Page 2916
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
Content Delivery Readiness of Philippine Higher Education
Institution Websites: Evidence from HTTP Archive and Chrome
UX Report
Romelyn J. Banaybanay
1*
, Reagan B. Ricafort
2
1
Initao College, Initao, Misamis Oriental, Philippines
2
AMA University, Makati City, Philippines
*Corresponding author
DOI: https://doi.org/10.51583/IJLTEMAS.2026.150600214
Received: 11 July 2026; Accepted: 16 July 2026; Published: 22 July 2026
ABSTRACT
This study examined the content delivery readiness of selected Philippine Higher Education Institution websites
using public performance data from the Chrome User Experience Report (CrUX) and HTTP Archive. The study
covered 30 official website origins, consisting of 10 State Universities and Colleges, 10 private HEIs, and 10
Local Universities and Colleges. Each website was validated and normalized before data extraction. CrUX data
for May 2026 and HTTP Archive page-level data from the May 1, 2026 crawl were collected through BigQuery.
Of the 30 websites, 25 had complete data from both sources, four had CrUX data only, and one had no usable
match. The readiness framework combined field indicators related to actual user experience with page-level
indicators describing technical delivery conditions. Among the 25 complete-data websites, five showed high
readiness, 13 moderate readiness, and seven low readiness. Private HEIs recorded the highest average score
among the three groups, although the comparison was interpreted descriptively because the corpus was
purposively selected. The most common issues were heavy page weight, slow visual loading, high request count,
slow Time to First Byte, and slow Largest Contentful Paint. Sensitivity testing using three alternative weighting
schemes showed that 84 percent to 96 percent of classifications remained unchanged, while rank correlations
ranged from 0.962 to 0.980. These results indicate that the main findings remained generally stable across the
alternative models, with most category changes occurring near the score boundaries. The resulting framework
provides a practical basis for identifying and prioritizing improvements in payload size, loading behavior, request
management, server response, and routine website performance monitoring.
Keywords: content delivery readiness, Core Web Vitals, Chrome User Experience Report, HTTP Archive,
Philippine higher education websites
INTRODUCTION
Higher education institutions increasingly depend on official websites to deliver academic information,
admission updates, enrollment instructions, program details, public advisories, and institutional services. For
students, applicants, parents, employees, alumni, and public stakeholders, the official institutional website often
serves as the first point of contact with a college or university. Website availability is therefore only a starting
point. The website must also deliver content efficiently, reliably, and measurably.
In the Philippine higher education context, institutional websites vary across State Universities and Colleges,
private higher education institutions, and Local Universities and Colleges. The Commission on Higher Education
provides a national entry point for identifying higher education institutions, but a listed institution does not
automatically mean that its official website performs well as a content delivery system [7]. A website may be
official and reachable while still carrying slow loading behavior, large resource payloads, delayed visual
rendering, or unstable layout movement.
Page 2917
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
Content delivery readiness refers to the degree to which a website delivers web content efficiently to end users.
From an Advanced Content Networking perspective, this includes loading performance, page weight, request
volume, server-response behavior, visual stability, mobile and desktop experience, and the availability of
measurable performance data. These indicators matter because academic websites often carry image-heavy
announcements, admission pages, enrollment notices, program information, portals, calendars, downloadable
documents, and public advisories. These content elements support institutional communication, but they also
increase delivery load when not optimized.
Public web measurement datasets support this type of assessment without collecting private institutional records
or personal user data. The Chrome User Experience Report provides origin-level field data based on real-world
Chrome user experience, while HTTP Archive provides page-level crawl data about how websites are built and
delivered [5], [6]. Together, these sources allow a study to examine both user-experience signals and technical
delivery indicators.
Many institutional website studies focus on accessibility, usability, information quality, design quality, service
quality, or e-learning readiness. These areas remain important. This study takes a narrower technical direction.
It examines the website as a content delivery system. This focus is useful because a website may look complete
in a browser while still performing weakly under measurable delivery indicators such as Largest Contentful
Paint, Time to First Byte, page weight, request count, Speed Index, or fully loaded time.
The research gap addressed in this study is the limited availability of transparent, public-data-based assessments
of Philippine HEI website content delivery readiness using both CrUX and HTTP Archive evidence. This study
responds to that gap by validating a corpus of Philippine HEI website origins, extracting available public
performance indicators, applying a readiness scoring model, and reporting the limits of the available data.
This study assessed 30 official Philippine HEI website origins. The corpus followed a balanced purposive
structure composed of 10 State Universities and Colleges, 10 private HEIs, and 10 Local Universities and
Colleges (LUCs). The design supports comparison across selected HEI groups, but it does not serve as a national
census of all Philippine HEI websites.
Objectives of the Study
The general objective of the study was to assess the content delivery readiness of selected Philippine Higher
Education Institution websites using public web performance evidence from HTTP Archive and the Chrome
User Experience Report.
Specifically, the study aimed to:
1. validate official Philippine HEI website origins and extract available CrUX and HTTP Archive page-level
indicators.
2. classify content delivery readiness and compare patterns across selected HEI groups.
3. identify recurring delivery issues and propose practical content networking recommendations.
Research Questions
1. What is the data availability profile of the selected Philippine HEI website origins in CrUX and HTTP
Archive?
2. What content delivery readiness levels are observed among the selected HEI websites?
3. How do readiness patterns differ across SUCs, private HEIs, and LUCs?
4. What common loading and delivery issues appear in the selected websites based on the available public
performance indicators?
5. What content networking recommendations follow from the readiness results?
Page 2918
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
Significance of the Study
The study contributes a technical baseline for examining Philippine HEI websites as content delivery systems.
For institutional administrators and website managers, the findings support practical actions such as reducing
homepage payloads, improving visual loading, controlling request overhead, and monitoring user-experience
indicators. For researchers, the study offers a public-data workflow for assessing website delivery readiness
without relying on private logs or survey responses.
The study does not rank academic quality, service quality, or overall digital maturity. It evaluates content
delivery readiness only from the selected public indicators. This boundary is important because a website
readiness score does not measure institutional quality or educational performance.
RELATED LITERATURE
Content Delivery and Web Performance
Content delivery refers to the technical process of serving web content from a host or delivery infrastructure to
the end user. For institutional websites, this process includes the transfer of HTML, images, scripts, style sheets,
fonts, and other page resources. A content delivery problem may appear as slow loading, large page weight,
excessive requests, delayed server response, or unstable visual rendering.
Web performance is part of content delivery because users experience a website through both content and
delivery speed. Users encounter how fast the page responds, how soon the main content appears, how stable the
layout remains, and how quickly the page becomes usable. In this sense, web performance connects directly to
accessibility of information and reliability of institutional communication.
For Advanced Content Networking, these concerns connect to delivery efficiency. A visually complete website
may still show weak readiness if it uses large payloads, excessive resources, slow response paths, or unstable
rendering. This study therefore treats delivery readiness as a measurable technical condition, not merely as
website presence.
Core Web Vitals
Core Web Vitals provide a user-centered way to evaluate web experience. The stable Core Web Vitals include
Largest Contentful Paint, Interaction to Next Paint, and Cumulative Layout Shift. LCP measures loading
performance, INP measures responsiveness, and CLS measures visual stability [28].
These indicators fit the present study because they translate user experience into measurable performance
signals. LCP shows whether the main content becomes visible within an acceptable time. INP shows whether
the page responds well to user interaction. CLS shows whether page elements shift unexpectedly during loading.
The study also included supporting CrUX indicators such as First Contentful Paint and Time to First Byte. FCP
helps describe when initial content appears. TTFB reflects early server or network response. These indicators
help explain whether a website delay begins at the server-response layer or during front-end rendering [26], [27].
HTTP Archive and Public Web Measurement
Public web measurement datasets allow researchers to study website delivery without collecting private
institutional records. CrUX and HTTP Archive support this approach from different perspectives. CrUX
provides field data from real Chrome users, while HTTP Archive provides crawl-based data about website
structure and delivery behavior [5], [13].
CrUX on BigQuery gives origin-level and page-level distributions of web experience metrics. Its methodology
explains that origins must be publicly discoverable and must have enough visits to appear in the dataset [6]. This
Page 2919
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
rule is important because a missing CrUX row does not automatically mean weak performance. It may mean
that the origin did not satisfy dataset inclusion requirements for the selected month.
HTTP Archive complements CrUX because it records page loads and stores data in public BigQuery tables. The
crawl.pages table supports page-level analysis, including delivery indicators such as page size, request totals,
resource sizes, and performance timing fields [14], [15]. In this study, the request-level table was not used
because the query exceeded the available free BigQuery query-byte quota. BigQuery documentation explains
that query costs and query-byte limits must be managed when working with large public datasets [11], [12].
HEI Website Performance Studies
Prior HEI website studies show that institutional websites support academic communication, public information
access, institutional visibility, and online service delivery. Aydin analyzed higher education website visitor data
and showed the value of studying website behavior as institutional evidence [2]. Subiyakto and colleagues
examined institutional repository website design through usability testing, which reflects the broader concern
for quality in academic web systems [23].
These studies provide useful background, but many of them focus on human perception, usability, accessibility,
information quality, service quality, or website design. These areas are important, but they do not fully measure
content delivery readiness. A website may be usable in layout but still weak in page weight, loading speed,
server-response behavior, or resource delivery.
The present study builds from this literature by adding a content networking perspective. It does not replace
user-centered website evaluation. It adds a delivery-based layer using public field data and page-level crawl data.
Philippine HEI Website Context
The Philippine higher education environment includes SUCs, private HEIs, LUCs, and other institution types
listed through CHED resources [7]. These institutions use official websites to publish admission requirements,
enrollment schedules, academic offerings, public advisories, and institutional services. Content delivery
readiness therefore matters for public access to institutional information.
Philippine HEIs differ in resources, governance structures, location, and website management capacity. These
differences may affect how websites are maintained and optimized. Public and private institutions may use
different web teams, hosting arrangements, procurement processes, and update practices.
Philippine HEI digital studies often discuss e-learning readiness, faculty and student readiness, and institutional
technology adoption. Lucero and colleagues assessed e-learning readiness across government and private HEIs
in the Philippines [19]. That work is relevant because it shows continuing interest in digital capacity in Philippine
higher education. The present study differs by focusing on public website delivery performance rather than e-
learning readiness.
Technical Basis for the Selected Indicators
Indicator-specific web performance literature further supports the use of LCP, INP, CLS, FCP, and TTFB in this
study. Core Web Vitals guidance supports the use of LCP, INP, and CLS as user-centered indicators of loading
performance, responsiveness, and visual stability [13], [28]. FCP and TTFB add supporting evidence by showing
when first content appears and whether delay begins at the server-response layer [26], [27].
Content delivery also depends on standard HTTP behavior. HTTP semantics define how clients and servers
exchange representations, while HTTP caching explains how reusable responses reduce latency and network
overhead. Cache-Control rules, compression, and newer transport protocols such as HTTP/3 show that delivery
readiness is affected by both page design and underlying protocol behavior [10], [9], [20], [21], [24], [4].
Page 2920
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
Browser timing standards also support the measurement logic of this study. Navigation Timing and Resource
Timing describe how page navigation and individual resource loads can be measured in browsers. These
standards support the idea that web performance can be studied through measurable timing and resource
indicators rather than subjective website impressions alone [29], [30].
Recent web measurement studies also show why resource size, images, request behavior, dependency loading,
and site maintenance matter. Image-heavy pages can affect page load time, fine-grained page-load dependency
studies show that object dependencies shape loading behavior, and HTTP Archive has been used to study
observable web deployment patterns at scale [8], [22], [25], [16], [1].
Recent HEI-related studies reinforce the need to examine institutional websites as public information systems.
Philippine HEI studies have addressed web accessibility and open access dissemination through institutional and
library websites, while international university website research has examined website quality through
quantifiable technical attributes. These works support the present study, but they also show the need for a
narrower content delivery readiness view [18], [3], [17].
Research Gap
Existing literature shows that HEI websites support public and academic communication, while most
institutional website studies emphasize usability, accessibility, information quality, service quality, or e-learning
readiness. Core Web Vitals, browser timing standards, and public datasets such as CrUX and HTTP Archive
provide measurable evidence of loading and delivery performance. HTTP behavior, caching, compression, and
protocol design further show that readiness depends on both page resources and network delivery. Few studies
apply these measures to Philippine HEI websites through a transparent public-data workflow.
To address this gap, the study validates a 30-origin Philippine HEI website corpus, extracts available CrUX and
HTTP Archive page-level indicators, classifies content delivery readiness, and reports data limitations. The
resulting baseline supports future HEI website performance research without extending claims beyond the
selected public indicators.
METHODOLOGY
Research Design
The research used a quantitative descriptive-comparative design to assess the content delivery readiness of
selected Philippine HEI websites. The institutional website origin served as the unit of analysis. This design
matched the study because it measured observable delivery indicators from public datasets and compared
readiness patterns across selected HEI categories.
Corpus and Sampling Frame
The corpus included 30 Philippine HEI website origins in a balanced 10/10/10 purposive structure: 10 SUCs, 10
private HEIs, and 10 LUCs. Official institutional websites were identified through the CHED HEI directory
entry point and institution-level verification. The sample was not intended as a national census of all Philippine
HEI websites.
Table 1. Sample composition and public-data coverage by HEI type
HEI type
Total origins
Complete data
CrUX only
No data match
SUC
10
8
2
0
Private HEI
10
9
1
0
LUC
10
8
1
1
Page 2921
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
URL Validation and Origin Normalization
Before extraction, each listed website underwent URL validation. The process checked whether the homepage
opened, redirected to another official address, required JavaScript rendering, blocked automated access, or
lacked a confirmable official site. Validation identified 19 active sites, three redirects, five JavaScript-rendered
sites, and three access-blocked domains. No inactive or unconfirmed official sites remained in the corpus.
Normalized origins were prepared during URL validation by reducing each validated institutional address to its
scheme and host. The extraction notebook then removed surrounding spaces, converted the stored origins to
lowercase, and removed trailing slashes before creating the CrUX origin list and HTTP Archive root-page list.
This rule kept the extraction consistent with CrUX methodology and supported origin-level matching across the
two public datasets [6].
Table 2. URL validation status
Status group
Count
URL validation status
3
URL validation status
19
URL validation status
5
URL validation status
3
Data Sources and Extraction Period
CrUX and HTTP Archive served as the two public web performance data sources. CrUX provided real-user field
data for website origins, while HTTP Archive provided page-level crawl data from the crawl.pages table. The
extraction covered the May 2026 CrUX release and the May 1, 2026 HTTP Archive crawl [5], [14], [15].
CrUX extraction produced 29 matched origins from the 30-origin corpus. HTTP Archive page-level extraction
produced 50 page rows across desktop and mobile client results, covering 25 website origins. The HTTP Archive
request-level query was attempted but was not completed because it exceeded the available free BigQuery query-
byte quota. Its empty worksheet was retained as a transparency record. Because request-level output was
unavailable, the study does not report cache-control header, compression-header, or request-header-level claims
[11], [12].
Methodological Basis of the Content Delivery Readiness Framework
Content delivery readiness refers to the overall capability of a website to deliver web content efficiently and
consistently based on observable technical performance indicators. In this study, it is treated as a composite
construct because website delivery depends on multiple technical dimensions rather than a single performance
measure. A website may load quickly but still require excessive network requests, carry unnecessarily large
resources, or experience delayed server response and unstable visual rendering. Evaluating only one indicator
would therefore provide an incomplete representation of delivery readiness. A composite framework allows
these complementary aspects of website delivery to be assessed together, providing a broader view of how
efficiently an institutional website delivers content to its users.
No single public dataset provides a complete representation of website content delivery readiness. For this
reason, the framework integrates evidence from the Chrome User Experience Report (CrUX) and HTTP Archive
because they measure complementary aspects of website performance. CrUX provides origin-level field data
derived from aggregated real-user browsing experience, reflecting how users encounter website performance
under actual usage conditions. In contrast, HTTP Archive provides standardized page-level observations
collected through controlled web crawls, describing the technical characteristics of homepage delivery such as
resource size, request volume, and loading behavior. These datasets therefore serve different but complementary
purposes. CrUX reflects observed user experience, while HTTP Archive describes the underlying technical
Page 2922
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
conditions that influence that experience. Using both sources provides a more balanced assessment of content
delivery readiness than relying on either dataset alone [5], [6], [14], [15].
The indicators included in the readiness framework were selected because each represents a distinct aspect of
website content delivery. User-centered indicators such as Largest Contentful Paint (LCP), Interaction to Next
Paint (INP), Cumulative Layout Shift (CLS), and Time to First Byte (TTFB) describe how visitors experience
loading performance, responsiveness, visual stability, and initial server response during actual website use.
Supporting page-level indicators such as page weight, request count, Speed Index, fully loaded time, HTTPS
adoption, and CDN presence describe technical conditions that influence content delivery efficiency. Although
these indicators measure different characteristics of website performance, they collectively represent the major
observable factors affecting how efficiently web content is delivered [27], [28].
The weighting framework was developed to reflect the relative contribution of each indicator to observable
content delivery readiness. Indicators that directly represent user experience, particularly loading performance,
responsiveness, visual stability, and initial server response, received greater influence because they reflect the
aspects of website delivery experienced most immediately by users. Supporting technical indicators such as page
weight, request count, Speed Index, fully loaded time, HTTPS adoption, and CDN presence were weighted
according to their contribution to the underlying delivery process. Indicators associated with resource efficiency
and loading behavior received greater emphasis than infrastructure-related indicators because they have a more
direct influence on measurable delivery performance. The resulting weighting structure reflects a logical
hierarchy of observable evidence, progressing from user-experience indicators to technical delivery indicators
and supporting delivery infrastructure.
The readiness score was developed as an analytical summary of the selected public performance indicators rather
than as an absolute measure of website quality or institutional performance. Its purpose is to support transparent
and consistent comparison among the selected HEI website origins using a common evaluation framework. The
resulting classifications should therefore be interpreted as descriptive summaries of observable content delivery
conditions during the selected measurement period. They do not establish institutional superiority, overall digital
maturity, or compliance with any official performance standard. Instead, they provide a structured basis for
identifying delivery strengths, recurring technical issues, and opportunities for website optimization using
publicly available evidence.
Organization of the Measurement Framework
The measurement framework was organized into two complementary layers that correspond to the two public
datasets used in the study. The first layer consists of CrUX field indicators, which capture observable user
experience through loading performance, responsiveness, visual stability, and initial server response. The second
layer consists of HTTP Archive page-level indicators, which describe the technical characteristics of homepage
delivery, including resource demand, loading behavior, and supporting delivery infrastructure. Organizing the
indicators in this manner distinguishes user-observed performance from the technical conditions that influence
it while allowing both perspectives to contribute to a common readiness framework.
Variables and Indicators
The variables included in this study were selected because they provide observable evidence of the different
technical dimensions that collectively define content delivery readiness. Rather than relying on a single
performance measure, the framework incorporates indicators that capture user experience, technical delivery
characteristics, and supporting delivery infrastructure. Each variable contributes distinct information about
website content delivery, allowing the readiness framework to evaluate multiple dimensions of the construct
using publicly available measurements from CrUX and HTTP Archive.
The CrUX variables included 75th percentile LCP, INP, CLS, FCP, and TTFB, together with phone and desktop
density values. LCP, INP, CLS, and TTFB contributed to the CrUX component score, while FCP and form-
factor densities supported interpretation. The HTTP Archive variables included page weight, request count,
JavaScript size, image size, Speed Index, fully loaded time, HTTPS ratio, and detected CDN provider. Page
Page 2923
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
weight, request count, Speed Index, fully loaded time, HTTPS ratio, and CDN detection contributed to the HTTP
Archive component score. JavaScript and image size were retained as descriptive resource indicators.
Table 3. Public-data coverage retained for analysis
Coverage item
Count
Interpretation
Total retained origins
30
Official website origins retained after
validation
CrUX matched origins
29
Origin-level real-user field data available
HTTP Archive matched origins
25
Page-level crawl data available
Complete CrUX and HTTP
Archive matches
25
Primary comparative dataset
CrUX-only origins
4
Retained but treated as limited-data cases
No extracted dataset match
1
Retained for corpus accounting but
excluded from performance comparison
Scoring Model
The scoring model was designed to convert the selected CrUX and HTTP Archive indicators into a common and
transparent basis for origin-level comparison. The raw indicators use different units, including milliseconds,
megabytes, request counts, proportions, and categorical infrastructure evidence, so their values could not be
combined directly. Each indicator was therefore translated into a points-based score using fixed cut-off values
and then weighted according to its role in the content delivery readiness framework. This process produced
separate CrUX and HTTP Archive component scores before the two components were combined for complete-
data origins.
The cut-off values served as translation rules that converted indicators with different units into comparable point
bands. For the CrUX component, the LCP, INP, and CLS cut-offs followed Core Web Vitals guidance, while
TTFB followed supporting web-performance guidance. For the HTTP Archive component, fixed analytic bands
distinguished lower, intermediate, and higher levels of resource demand and loading delay. These cut-offs were
used as study-specific scoring rules rather than universal website performance standards. The complete point
assignments, cut-off values, weights, and formulas are reported in the reproducibility appendix. The influence
of the researcher-defined weighting structure was also examined through sensitivity analysis using alternative
weighting schemes [27], [28].
The CrUX component assigned 30 percent to LCP, 25 percent to INP, 25 percent to CLS, and 20 percent to
TTFB. The HTTP Archive component assigned 25 percent to page weight, 20 percent to request count, 25
percent to Speed Index, 15 percent to fully loaded time, 10 percent to HTTPS ratio, and 5 percent to CDN
detection. Within each component, the points assigned through the cut-off rules were multiplied by their
respective weights and summed to produce a score from 0 to 100. For origins with complete data, the overall
readiness score was calculated as the mean of the CrUX and HTTP Archive component scores. Giving equal
influence to the two component scores prevented either real-user field evidence or page-level crawl evidence
from dominating the final classification.
Readiness classifications were assigned using fixed score ranges: high readiness for scores greater than or equal
to 80, moderate readiness for scores greater than or equal to 60 and less than 80, and low readiness for scores
below 60. These ranges functioned as an analytic rubric for the study and were not treated as an official national
standard. The final comparative analysis used only the 25 origins with complete CrUX and HTTP Archive data.
CrUX-only origins retained their CrUX component score but were labeled as limited-data cases and reported
separately from complete-data classifications. Origins with no extracted evidence were classified as insufficient
data and excluded from performance comparison. This separation ensured that websites scored from different
amounts of evidence were not interpreted as directly equivalent.
Page 2924
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
For the 25 HTTP Archive-matched origins, desktop and mobile page rows were available. Because the
institutional website origin was the unit of analysis, continuous page-level values were averaged before scoring.
The HTTPS ratio was calculated from the combined desktop and mobile request totals, while CDN evidence
was retained at the origin level. This produced one origin-level value for each indicator. The original client-
specific observations were retained in the source dataset to preserve device-level evidence for reproducibility
and future analysis.
Sensitivity Analysis
To examine whether the results depended heavily on the original weighting structure, three alternative schemes
were tested using the 25 complete-data origins. The first scheme assigned equal weights to all indicators within
each component. The second gave greater influence to CrUX field evidence through a 60:40 CrUX-to-HTTP
Archive ratio. The third gave greater influence to HTTP Archive page-level evidence through a 40:60 ratio. The
analysis compared readiness classifications, score changes, rank order, and HEI-type averages against the
original model.
Across the three alternative schemes, 84 percent to 96 percent of readiness classifications remained unchanged.
Spearman rank correlations ranged from 0.962 to 0.980, showing that the ordering of websites remained highly
consistent. Moderate readiness remained the dominant classification under every scheme. Private HEIs also
retained the highest complete-data mean across all weighting structures. The few category changes involved
websites with scores near the 60-point and 80-point classification boundaries. These results show that the main
findings remained stable under reasonable alternative weighting schemes, while classifications near the cut-off
values require cautious interpretation.
Data Audit and Interpretation Rules
A post-extraction audit checked corpus balance, validation status, extraction coverage, request-level data
availability, limited-data cases, scoring defensibility, and group comparability. The audit retained all 30 origins
as the validated corpus. The 25 complete CrUX and HTTP Archive matches formed the primary comparative
dataset. Four CrUX-only rows and one no-match row were reported as data caveats.
Ethical and Technical Boundaries
The analysis used public institutional website URLs and aggregate web performance datasets. It did not collect
personal information, login-protected data, student records, or user-identifiable browsing records. Unsupported
claims were excluded. The study examined selected content delivery indicators only and did not assess
institutional academic quality, web accessibility compliance, security posture, or user satisfaction.
RESULTS AND DISCUSSION
Dataset Coverage and Extraction Completeness
After validation, the final corpus retained 30 official Philippine HEI website origins. The balanced 10/10/10
structure remained intact, with 10 SUCs, 10 private HEIs, and 10 LUCs. The sampling frame therefore remained
balanced after cleaning.
Data availability was strong but not complete. CrUX matched 29 of the 30 origins. HTTP Archive page-level
data matched 25 origins. Overall, 25 origins had complete CrUX and HTTP Archive coverage, four had CrUX-
only coverage, and one had no extracted dataset match for the selected month. The 25 complete rows form the
best basis for comparative performance interpretation.
Page 2925
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
Overall Content Delivery Readiness
Among the 25 complete-data origins, five were classified as high readiness, 13 as moderate readiness, and seven
as low readiness. This primary comparative distribution corresponded to 20.0 percent high readiness, 52.0
percent moderate readiness, and 28.0 percent low readiness.
The four CrUX-only origins and the one no-match origin were retained for corpus accounting and reported
separately as limited-data cases.
Table 4. Readiness-level distribution for complete-data origins
Readiness level
n
%
High readiness
5
20.0%
Moderate readiness
13
52.0%
Low readiness
7
28.0%
Moderate readiness dominated the results. Many selected HEI websites were reachable and measurable, but still
carried delivery inefficiencies. The main problems appeared in large page payloads, slow visual loading, high
request volume, and weak loading-response indicators.
Readiness by HEI Type
Group averages showed a cautious pattern. Among complete-data cases, private HEIs recorded the highest mean
score at 71.3, followed by LUCs at 68.7 and SUCs at 67.7. These values do not prove that one sector generally
performs better. They describe only the selected websites and the available public indicators for the extraction
month.
Table 5. Complete-data readiness distribution and mean score by HEI type
HEI type
n
Mean
High
Moderate
Low
SUC
8
67.7
1
4
3
Private HEI
9
71.3
4
2
3
LUC
8
68.7
0
7
1
SUCs showed mixed performance across high, moderate, and low readiness. Private HEIs produced the largest
number of high-readiness cases, but also included low-readiness rows. LUCs were mostly moderate, with one
low-readiness row and one no-match row. This pattern supports the descriptive-comparative framing of the
study.
CrUX Field-Performance Findings
CrUX covered 29 origins and provided field evidence from real users rather than a single synthetic test.
Interactivity and visual stability were relatively strong for many sites, while loading indicators were weaker
across a notable part of the corpus.
Among the 29 CrUX-matched origins, 16 met the good LCP threshold of 2.5 seconds or less, while 13 exceeded
it. For INP, 23 origins met the 200-millisecond threshold, while six exceeded it. For CLS, 26 origins stayed at
or below 0.1, while three exceeded it. For FCP and TTFB, only 13 origins met the reference thresholds. These
figures suggest that interaction and layout stability were stronger than initial loading and server-response
behavior.
Page 2926
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
Table 6. CrUX threshold results for matched origins
Indicator
Interpretation
Good reference
threshold
Met good
threshold
Above good
threshold
LCP
Loading
performance
<= 2,500 ms
16
13
INP
Interactivity
<= 200 ms
23
6
CLS
Visual stability
<= 0.1
26
3
FCP
Early content
visibility
<= 1,800 ms
13
16
TTFB
Server-response
support metric
<= 800 ms
13
16
From a content-networking perspective, loading speed depends on both front-end resource design and network
delivery choices. HEI websites often carry banners, announcements, images, program pages, scripts, and portal
links. Without optimization, these assets allow the site to remain functional while weakening loading
performance.
HTTP Archive Page-Level Findings
HTTP Archive page-level data covered 25 origins. The median average page weight was 15.4 MB, the median
average request count was 80.0, the median Speed Index was 8949 ms, and the median fully loaded time was
21860 ms. These values show that several websites carried resource-heavy homepages and slow visual
completion.
Table 7. HTTP Archive page-level descriptive statistics
Metric
Unit
Matched
origins
Mean
Median
Minimum
Maximum
Average page
weight
MB
25
24.1
15.4
1.4
108.0
Average request
count
requests
25
110.3
80.0
32.5
274
Average JavaScript
size
MB
25
0.9
0.65
0.0
2.9
Average image size
MB
25
18.8
7.4
0.2
77.0
Average Speed
Index
ms
25
12675.7
8949.5
3647.5
61136
Average fully
loaded time
ms
25
23495.9
21860.5
4996
60690
HTTPS ratio
percent
25
99.88%
100.00%
97.59%
100.00%
HTTP Archive
score
points
25
56.3
52.5
38.5
90
The issue-frequency audit showed the same pattern. Heavy page weight appeared in 18 websites, slow visual
loading in 16, and high request count in seven. These were the most common delivery issues. For LCP and
TTFB, Table 6 counts all origins outside the good reference threshold, including needs-improvement cases.
Table 8 counts only origins that met the study’s poor issue threshold.
Page 2927
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
Table 8. Frequency of observed content delivery issues
Observed issue
Affected
origins
Interpretation
Heavy page weight
18
The dominant content delivery problem points to large
images, scripts, media, or homepage assets.
Slow visual loading
16
Several sites render meaningful visual content slowly,
affecting perceived loading performance.
High request count
7
Many resources must be fetched before the page stabilizes,
raising network overhead.
Slow TTFB
6
Affected origins showed poor server-response timing under
the issue rule.
Slow LCP
6
Affected origins showed poor main-content loading under
the issue rule.
No HTTP Archive page data
5
This is a data-availability issue, not a direct performance
failure.
High CLS
3
Affected sites show layout instability risk during loading.
No major issue from selected
indicators
3
These rows serve as better-performing cases under the
selected metrics.
From a network delivery perspective, heavy page weight increases transfer demand and delays meaningful
rendering. High request count increases connection, scheduling, and resource-loading overhead. Slow visual
loading affects perceived responsiveness because users wait longer before meaningful page content becomes
visible.
Highest- and Lowest-Scoring Complete-Data Cases
The complete-data subset offered the clearest basis for case-level comparison. Adamson University, Bicol
University, De La Salle University, AMA University or AMA Education System, and the University of San
Carlos recorded the five highest overall readiness scores.
Table 9. Highest-scoring complete-data cases
HEI website
HEI type
Overall readiness
score
Readiness level
Main observed
issue(s)
Adamson
University
Private HEI
95
High readiness
No major issue
from selected
indicators
Bicol University
SUC
91
High readiness
No major issue
from selected
indicators
De La Salle
University
Private HEI
86.2
High readiness
Heavy page weight
AMA University /
AMA Education
System
Private HEI
84
High readiness
No major issue
from selected
indicators
University of San
Carlos
Private HEI
82.2
High readiness
Heavy page weight
Page 2928
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
The five lowest-scoring complete-data cases were the University of Santo Tomas, Western Mindanao State
University, Mapua University, City of Malabon University, and West Visayas State University. Their low-
readiness classifications reflected combinations of slow LCP, slow TTFB, high CLS, heavy page weight, high
request count, and slow visual loading.
Table 10. Lowest-scoring complete-data cases
HEI website
HEI type
Overall
readiness
score
Readiness
level
Main observed issue(s)
University of Santo
Tomas
Private HEI
48.5
Low readiness
High CLS; Slow TTFB; Heavy page
weight; High request count
Western Mindanao
State University
SUC
49.8
Low readiness
Slow LCP; Slow TTFB; Heavy page
weight; Slow visual loading
Mapua University
Private HEI
51.5
Low readiness
High CLS; Heavy page weight; High
request count; Slow visual loading
City of Malabon
University
LUC
52.8
Low readiness
Slow LCP; Heavy page weight; Slow
visual loading
West Visayas State
University
SUC
54.8
Low readiness
Slow LCP; Heavy page weight; High
request count; Slow visual loading
These lists do not rank institutional quality. The scores measure selected content delivery indicators for official
website origins during the extraction month. A high score does not establish completeness, accessibility, full
security, or institutional superiority. A low score does not indicate academic weakness. It means that the selected
public indicators showed delivery problems during the study period.
Handling Limited-Data Cases
Five sites required limited-data treatment. Bulacan State University, Mindanao State University - Iligan Institute
of Technology, Ateneo de Manila University, and University of Makati had CrUX-only coverage. Pamantasan
ng Lungsod ng Valenzuela had no extracted dataset match for the selected month. These sites remained in the
validated corpus but were excluded from direct comparison with complete-data rows.
Table 11. Limited-data cases and reporting caveats
HEI website
HEI
type
Data
coverage
Limited-data status
Reporting caveat
Bulacan State University
SUC
CrUX only
High CrUX score
(limited data)
High CLS; No HTTP
Archive page data
Mindanao State University
- Iligan Institute of
Technology
SUC
CrUX only
High CrUX score
(limited data)
No HTTP Archive page
data
Ateneo de Manila
University
Private
HEI
CrUX only
High CrUX score
(limited data)
No HTTP Archive page
data
University of Makati
LUC
CrUX only
Moderate CrUX
score (limited data)
No HTTP Archive page
data
Pamantasan ng Lungsod ng
Valenzuela
LUC
No extracted
dataset match
Insufficient data
No CrUX field data; No
HTTP Archive page
data
Page 2929
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
Reporting these sites separately protects the comparison. A missing HTTP Archive page row indicates limited
data availability, not poor website performance. A no-match row means that the selected public datasets lacked
enough evidence for scoring in the extraction month. Later monthly releases should be checked before stronger
conclusions are made.
DISCUSSION
Moderate readiness was the main result for the selected Philippine HEI website corpus. Most sites were not
critically failing, but many were also not highly optimized. They occupied a middle condition in which the
website worked while content delivery efficiency remained uneven.
The strongest technical issue was heavy page weight. This finding matters because page weight affects transfer
size, loading time, and mobile access. HEI websites often publish high-resolution images, event banners,
announcements, sliders, embedded scripts, and portal links. These elements support communication, but they
also create delivery costs when left uncompressed or unmanaged.
Slow visual loading was also frequent. The likely technical sources include large images, render-blocking
resources, multiple scripts, and weak above-the-fold optimization. These issues affect how quickly users see
meaningful content. In a school website, delays may affect applicants checking admission information, students
checking enrollment notices, and stakeholders accessing advisories.
Private HEIs recorded the highest mean score in the selected complete-data subset. This result requires caution
because the sample was purposive and not nationally representative. It may reflect differences in hosting, web
maintenance capacity, design practices, update frequency, or vendor support among the selected websites. It
does not support a national claim about public and private HEI performance.
Public web measurement also proved useful for HEI research. CrUX and HTTP Archive support delivery-
readiness analysis without private analytics or personal data. Their limits remain important: some origins have
incomplete public coverage, and large BigQuery tables require quota or cost management. These limits should
be reported rather than hidden.
Limitations of the Results
The findings are descriptive and comparative, not causal. The study did not measure all pages of each
institutional website. It used official origins and homepage-level or origin-level public data. HTTP Archive tests
pages in a controlled crawl environment, while CrUX reports aggregate real-user field data. These sources
complement each other but do not produce the same type of evidence.
The request-level HTTP Archive table was not included because the query exceeded the free BigQuery query-
byte quota. The study therefore avoided claims about cache-control headers, compression headers, request-level
protocol behavior, and detailed third-party request patterns. Future work should include request-level extraction
in a controlled BigQuery environment.
Even with these limits, the dataset supports the study purpose. It provides a transparent baseline for selected HEI
website origins using public, repeatable, and technically relevant indicators. It also identifies clear action areas
for website teams: reduce page weight, improve visual loading, manage request count, monitor TTFB, and
review performance over time.
Summary of Findings
Four findings summarize the results. First, the 30-origin corpus remained balanced and valid after URL
validation. Second, 25 origins had complete CrUX and HTTP Archive data and formed the primary comparative
dataset. Third, moderate readiness dominated, while high readiness was less common and low readiness
remained present. Fourth, heavy page weight and slow visual loading were the most frequent delivery issues.
Page 2930
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
CONCLUSION AND RECOMMENDATIONS
Conclusion
This study examined the content delivery readiness of 30 selected Philippine HEI websites using data from CrUX
and HTTP Archive. Of the 30 websites, 25 had complete data and were included in the main comparison. The
four CrUX-only cases and one no-match case were reported separately.
Most of the websites with complete data showed moderate readiness. Common problems included heavy page
weight, slow visual loading, too many requests, slow server response, and slow Largest Contentful Paint. Private
HEIs had the highest average score in the selected group, but results still varied within all three HEI types. This
means that HEI type alone does not explain website performance.
The sensitivity analysis showed that the main results changed very little when different weighting schemes were
used. However, websites with scores close to the cut-off points should still be interpreted with care.
Overall, the study shows that a working website is not always ready to deliver content efficiently. The proposed
framework gives selected HEIs a practical way to identify weak areas, set priorities for improvement, and
regularly monitor website performance using publicly available data.
Recommendations
Reduce page weight. HEIs with heavy page weight should compress and resize images, remove unused media,
minify CSS and JavaScript, and avoid oversized homepage banners. This should be the first technical priority
because heavy page weight was the most frequent issue in the dataset.
Improve visual loading performance. HEIs with slow visual loading should optimize above-the-fold content,
defer non-critical scripts, preload critical assets where appropriate, and avoid render-blocking resources. These
actions support better perceived loading performance and stronger LCP-related results.
Lower request overhead. Websites with high request counts should consolidate repeated resources, reduce
unnecessary third-party embeds, apply lazy loading, and load non-essential scripts after the main content.
Improve server response performance. Websites with slow TTFB should review hosting quality, database
response time, origin server configuration, DNS performance, and CDN or edge delivery options. Server-side
delay affects the rest of the loading process.
Control layout instability. HEIs with high CLS should reserve dimensions for images, sliders, videos, and
embedded content. They should avoid late-loading banners and shifting page elements because these affect user
experience during page loading.
Enforce secure content delivery. All HEI websites should enforce HTTPS across the main document and page
resources. Secure delivery is a baseline requirement for public institutional websites.
Use regular monitoring. HEIs should review CrUX and HTTP Archive indicators at least once per term, with
monthly checks where feasible. Monitoring should include LCP, INP, CLS, FCP, TTFB, page weight, request
count, resource sizes, and mobile and desktop results.
Adopt a content delivery checklist. Institutional web teams should use a checklist before publishing major
homepage updates, admission pages, enrollment announcements, portal links, or public advisories. The checklist
should cover image size, script loading, HTTPS, mobile rendering, and page weight.
Repeat the study with request-level data. Future extraction should include HTTP Archive request-level data
in a controlled BigQuery environment. This would support stronger analysis of cache-control headers,
compression, CDN behavior, protocol use, and third-party resource patterns.
Page 2931
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
Expand the corpus and conduct longitudinal analysis. Future studies should include more HEIs across regions
and track several monthly releases. A longitudinal design would show whether institutional websites improve,
decline, or remain stable over time.
Recommended Technical Action by Readiness Level
For high-readiness websites, the main action is maintenance. These websites should continue monitoring and
avoid new homepage elements that increase page weight or layout instability. For moderate-readiness websites,
the main action is optimization. These websites should reduce page weight, improve visual loading, and review
request overhead. For low-readiness websites, the main action is remediation. These websites should review
hosting, server response, page structure, resource size, and front-end loading behavior. For insufficient-data
websites, the main action is monitoring. These origins should be rechecked in future CrUX and HTTP Archive
releases.
Contribution of the Study
The study contributes a content-networking perspective to the assessment of Philippine HEI websites by shifting
attention from mere website presence to the technical readiness of websites to deliver content efficiently. Its
main contribution is a transparent and low-cost assessment framework that combines CrUX field evidence with
HTTP Archive page-level indicators. The framework supports comparative evaluation, identifies recurring
delivery problems, and helps institutions prioritize practical improvements using publicly available data. It also
offers a reproducible approach that other researchers and educational institutions may adapt for periodic website
performance assessment.
Declarations
Ethical Statement
This study used public institutional website URLs and aggregate public datasets. It did not involve human
participants, animals, personal information, login-protected records, or private institutional data. Formal ethics
approval was not required.
Funding
This research received no external funding.
Conflict of Interest
The authors declare no conflict of interest.
Data Availability Statement
The source data used in this study are publicly available through the Chrome User Experience Report and HTTP
Archive. Supporting research materials, including the source inventory, extraction notebook, cleaned dataset,
frozen scoring workbook, sensitivity-analysis workbook, and reproducibility documentation, may be obtained
from the corresponding author upon reasonable request.
REFERENCES
1. Alkinoon, M., Alabduljabbar, A., Althebeiti, H., Jang, R., Nyang, D., & Mohaisen, D. (2023). Understanding
the security and performance of the web presence of hospitals: A measurement study. arXiv.
https://doi.org/10.48550/arXiv.2304.13278
2. Aydin, O. (2021). Analysis of the visitor data of a higher education institution website. arXiv.
https://doi.org/10.48550/arXiv.2107.14107
Page 2932
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
3. Balbin, D. S. (2026). OA or nonchalant: A survey of open access dissemination initiatives of higher education
institution library websites and systems in the Philippines. College & Research Libraries, 87(2).
https://crl.acrl.org/index.php/crl/article/view/27268/35061
4. Bishop, M. (2022). HTTP/3 (RFC 9114). RFC Editor. https://www.rfc-editor.org/rfc/rfc9114
5. Chrome for Developers. (n.d.). CrUX on BigQuery. Retrieved July 10, 2026, from
https://developer.chrome.com/docs/crux/bigquery
6. Chrome for Developers. (n.d.). CrUX methodology. Retrieved July 10, 2026, from
https://developer.chrome.com/docs/crux/methodology
7. Commission on Higher Education. (n.d.). List of higher education institutions. Retrieved July 10, 2026, from
https://legacy.ched.gov.ph/list-of-higher-education-institutions-2/
8. Dornauer, B., & Felderer, M. (2023). Web image formats: Assessment of their real-world usage and performance
across popular web browsers. arXiv. https://doi.org/10.48550/arXiv.2310.00788
9. Fielding, R., Nottingham, M., & Reschke, J. (2022). HTTP caching (RFC 9111). RFC Editor. https://www.rfc-
editor.org/rfc/rfc9111
10. Fielding, R., Nottingham, M., & Reschke, J. (2022). HTTP semantics (RFC 9110). RFC Editor. https://www.rfc-
editor.org/rfc/rfc9110
11. Google Cloud. (n.d.). Estimate and control costs in BigQuery. Retrieved July 10, 2026, from
https://docs.cloud.google.com/bigquery/docs/best-practices-costs
12. Google Cloud. (n.d.). BigQuery quotas and limits. Retrieved July 10, 2026, from
https://docs.cloud.google.com/bigquery/quotas
13. Google Search Central. (n.d.). Understanding Core Web Vitals and Google Search results. Retrieved July 10,
2026, from https://developers.google.com/search/docs/appearance/core-web-vitals
14. HTTP Archive. (n.d.). Frequently asked questions. Retrieved July 10, 2026, from https://httparchive.org/faq
15. HTTP Archive. (n.d.). Getting started: Accessing HTTP Archive with BigQuery. Retrieved July 10, 2026, from
https://har.fyi/guides/getting-started/
16. Jeřábek, K., & Polčák, L. (2023). Network Error Logging: HTTP Archive analysis. arXiv.
https://doi.org/10.48550/arXiv.2305.01249
17. Król, K. (2026). From website quality metrics to web well-being friendliness: A quantitative assessment of
university websites in Poland. Quality & Quantity. https://doi.org/10.1007/s11135-026-02871-2
18. Lagura, G., Padios, I. J., Junsay, M., Enecito, H. J., Apellido, E., Balbastro, L. J., Daapong, L., & Dacut, S.
(2026). Web accessibility and institutional readiness for inclusive governance in Philippine higher education.
Journal of Public and Private Management, 33(1), 99-123. https://doi.org/10.14456/ppm.2026.7
19. Lucero, H. R., Victoriano, J. M., Carpio, J. T., & Fernando, P. G., Jr. (2022). Assessment of e-learning readiness
of faculty members and students in the government and private higher education institutions in the Philippines.
arXiv. https://doi.org/10.48550/arXiv.2202.06069
20. MDN Web Docs. (n.d.). Cache-Control header. Retrieved July 10, 2026, from https://developer.mozilla.org/en-
US/docs/Web/HTTP/Reference/Headers/Cache-Control
21. MDN Web Docs. (n.d.). Compression in HTTP. Retrieved July 10, 2026, from https://developer.mozilla.org/en-
US/docs/Web/HTTP/Guides/Compression
22. Netravali, R., Mickens, J., & Balakrishnan, H. (2016). Polaris: Faster page loads using fine-grained dependency
tracking. Proceedings of the 13th USENIX Symposium on Networked Systems Design and Implementation,
123-136. https://www.usenix.org/conference/nsdi16/technical-sessions/presentation/netravali
23. Subiyakto, A., Rahmi, Y., Kumaladewi, N., Huda, M. Q., Hasanati, N., & Haryanto, T. (2021). Investigating
quality of institutional repository website design using usability testing framework. arXiv.
https://doi.org/10.48550/arXiv.2101.04880
24. Trevisan, M., Giordano, D., Drago, I., & Khatouni, A. S. (2021). Measuring HTTP/3: Adoption and
performance. arXiv. https://doi.org/10.48550/arXiv.2102.12358
25. Wang, X. S., Balasubramanian, A., Krishnamurthy, A., & Wetherall, D. (2013). Demystifying page load
performance with WProf. Proceedings of the 10th USENIX Symposium on Networked Systems Design and
Implementation, 473-485. https://www.usenix.org/conference/nsdi13/technical-
sessions/presentation/wang_xiao_sophia
26. web.dev. (n.d.). First Contentful Paint (FCP). Retrieved July 10, 2026, from https://web.dev/articles/fcp
27. web.dev. (n.d.). Time to First Byte (TTFB). Retrieved July 10, 2026, from https://web.dev/articles/ttfb
28. web.dev. (n.d.). Web Vitals. Retrieved July 10, 2026, from https://web.dev/articles/vitals
Page 2933
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
29. World Wide Web Consortium. (2022). Navigation Timing Level 2. W3C Recommendation.
https://www.w3.org/TR/navigation-timing-2/
30. World Wide Web Consortium. (2025). Resource Timing. W3C Working Draft.
https://www.w3.org/TR/resource-timing/
APPENDIX A. REPRODUCIBILITY DOCUMENTATION
This appendix records the origins, datasets, variables, transformations, matching rules, scoring procedures,
device aggregation, sensitivity analysis, and supporting files needed to trace the reported readiness
classifications.
A.1 Normalized HEI Origins and Dataset Coverage
Table A1 lists the normalized origins used to match CrUX and HTTP Archive records. Coverage was
classified as complete data, CrUX only, or insufficient data according to the evidence returned by the two
datasets.
Table A1. Normalized HEI origins and dataset coverage
No.
HEI name
Type
Region
Normalized origin
CrUX
HTTP
Archive
Analysis
status
1
University
of the
Philippines
System
SUC
NCR
https://up.edu.ph
Yes
Yes
Complete data
2
Polytechni
c
University
of the
Philippines
SUC
NCR
https://www.pup.edu.ph
Yes
Yes
Complete data
3
Batangas
State
University,
The
National
Engineerin
g
University
SUC
Region IV-A
https://batstateu.edu.ph
Yes
Yes
Complete data
4
Bulacan
State
University
SUC
Region III
https://bulsu.edu.ph
Yes
No
CrUX only
5
Bicol
University
SUC
Region V
https://bicol-u.edu.ph
Yes
Yes
Complete data
6
Central
Mindanao
University
SUC
Region X
https://www.cmu.edu.ph
Yes
Yes
Complete data
7
Cagayan
State
University
SUC
Region II
https://csu.edu.ph
Yes
Yes
Complete data
8
Western
Mindanao
State
University
SUC
Region IX
https://wmsu.edu.ph
Yes
Yes
Complete data
9
West
Visayas
State
University
SUC
Region VI
https://wvsu.edu.ph
Yes
Yes
Complete data
Page 2934
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
No.
HEI name
Type
Region
Normalized origin
CrUX
HTTP
Archive
Analysis
status
10
Mindanao
State
University
- Iligan
Institute of
Technolog
y
SUC
Region X
https://msuiit.edu.ph
Yes
No
CrUX only
11
Ateneo de
Manila
University
Private HEI
NCR
https://www.ateneo.edu
Yes
No
CrUX only
12
De La Salle
University
Private HEI
NCR
https://www.dlsu.edu.ph
Yes
Yes
Complete data
13
University
of Santo
Tomas
Private HEI
NCR
https://www.ust.edu.ph
Yes
Yes
Complete data
14
Mapua
University
Private HEI
NCR
https://www.mapua.edu.ph
Yes
Yes
Complete data
15
Far Eastern
University
Private HEI
NCR
https://www.feu.edu.ph
Yes
Yes
Complete data
16
University
of San
Carlos
Private HEI
Region VII
https://usc.edu.ph
Yes
Yes
Complete data
17
Silliman
University
Private HEI
Region VII
https://su.edu.ph
Yes
Yes
Complete data
18
Adamson
University
Private HEI
NCR
https://www.adamson.edu.ph
Yes
Yes
Complete data
19
University
of the East
Private HEI
NCR
https://www.ue.edu.ph
Yes
Yes
Complete data
20
AMA
University
/ AMA
Education
System
Private HEI
NCR
https://www.amaes.edu.ph
Yes
Yes
Complete data
21
Pamantasa
n ng
Lungsod
ng Maynila
LUC
NCR
https://plm.edu.ph
Yes
Yes
Complete data
22
University
of Makati
LUC
NCR
https://www.umak.edu.ph
Yes
No
CrUX only
23
Quezon
City
University
LUC
NCR
https://qcu.edu.ph
Yes
Yes
Complete data
24
Taguig
City
University
LUC
NCR
https://tcu.edu.ph
Yes
Yes
Complete data
25
Pamantasa
n ng
Lungsod
ng
Muntinlup
a
LUC
NCR
https://www.plmun.edu.ph
Yes
Yes
Complete data
26
City
College of
Angeles
LUC
Region III
https://cca.edu.ph
Yes
Yes
Complete data
Page 2935
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
No.
HEI name
Type
Region
Normalized origin
CrUX
HTTP
Archive
Analysis
status
27
Mabalacat
City
College
LUC
Region III
https://mcc.edu.ph
Yes
Yes
Complete data
28
City of
Malabon
University
LUC
NCR
https://cityofmalabonuniversity.edu
.ph
Yes
Yes
Complete data
29
Pamantasa
n ng
Lungsod
ng
Valenzuela
LUC
NCR
https://valenzuela.plv.edu.ph
No
No
Insufficient
data
30
Navotas
Polytechni
c College
LUC
NCR
https://navotaspolytechniccollege.e
du.ph
Yes
Yes
Complete data
The primary scoring and sensitivity analyses used the 25 complete-data origins. Four CrUX-only origins and
one insufficient-data origin were retained and reported separately.
A.2 Data Sources, Extraction Periods, and Dataset Tables
CrUX data were extracted from chrome-ux-report.materialized.metrics_summary for May 2026, and HTTP
Archive page data were extracted from httparchive.crawl.pages for the May 1, 2026 crawl. Parameterized lists
contained the 30 normalized origins and corresponding root pages [5], [14], [15].
Table A2. Data sources and retained outputs
Source
BigQuery table
Measurement
level
Period
Retained
output
Chrome User
Experience
Report
chrome-ux-
report.materialized.metrics_summary
Website origin
May 2026
29 origin rows
HTTP Archive
httparchive.crawl.pages
Homepage and
client type
May 1, 2026
crawl
50 page rows
HTTP Archive
request table
httparchive.crawl.requests
Individual
requests
May 1, 2026
crawl
0 rows (query
not completed)
The 50 retained page rows represented desktop and mobile observations for 25 origins. The request-level query
was not completed because it exceeded the available free BigQuery query-byte quota, and no request-header
variables were analyzed [11], [12].
Core CrUX analysis query
SELECT yyyymm, origin,
p75_lcp AS p75_lcp_ms, p75_inp AS p75_inp_ms, p75_cls,
p75_fcp AS p75_fcp_ms, p75_ttfb AS p75_ttfb_ms,
phoneDensity, desktopDensity, tabletDensity
FROM `chrome-ux-report.materialized.metrics_summary`
WHERE yyyymm = @crux_yyyymm
AND origin IN UNNEST(@origins)
ORDER BY origin;
Core HTTP Archive analysis query
SELECT date AS crawl_date, client,
`httparchive.fn.GET_ORIGIN`(root_page) AS origin, root_page,
SAFE_CAST(JSON_VALUE(summary, '$.bytesTotal') AS INT64) AS bytes_total,
SAFE_CAST(JSON_VALUE(summary, '$.bytesJS') AS INT64) AS bytes_js,
SAFE_CAST(JSON_VALUE(summary, '$.bytesImg') AS INT64) AS bytes_img,
SAFE_CAST(JSON_VALUE(summary, '$.reqTotal') AS INT64) AS req_total,
SAFE_CAST(JSON_VALUE(summary, '$.numHttps') AS INT64) AS num_https,
SAFE_CAST(JSON_VALUE(summary, '$.SpeedIndex') AS INT64) AS speed_index_ms,
SAFE_CAST(JSON_VALUE(summary, '$.fullyLoaded') AS INT64) AS fully_loaded_ms,
Page 2936
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
JSON_VALUE(summary, '$.cdn') AS cdn
FROM `httparchive.crawl.pages`
WHERE date = @crawl_date AND client IN ('mobile', 'desktop')
AND is_root_page AND root_page IN UNNEST(@root_pages)
ORDER BY origin, client;
Availability-check query
SELECT
'CrUX matched origins' AS check_name,
COUNT(DISTINCT origin) AS matched
FROM `chrome-ux-report.materialized.metrics_summary`
WHERE yyyymm = @crux_yyyymm
AND origin IN UNNEST(@origins)
UNION ALL
SELECT
'HTTP Archive matched origins' AS check_name,
COUNT(DISTINCT `httparchive.fn.GET_ORIGIN`(root_page)) AS matched
FROM `httparchive.crawl.pages`
WHERE date = @crawl_date
AND is_root_page
AND root_page IN UNNEST(@root_pages);
The compact queries above reproduce the fields used in scoring and descriptive analysis. The broader original
extraction query and parameter setup remain in the extraction workbook and Google Colab notebook.
A.3 Variables, Source Fields, Units, and Transformations
Table A3 identifies each analysis variable, source field, unit, transformation, and role in scoring or
interpretation.
Table A3. Variable definitions and transformations
Dataset
Analysis variable
Source field
Unit
Transformation and use
CrUX
Largest Contentful
Paint
p75_lcp
Milliseconds
Renamed p75_lcp_ms; weighted
CrUX
Interaction to Next
Paint
p75_inp
Milliseconds
Renamed p75_inp_ms; weighted
CrUX
Cumulative Layout
Shift
p75_cls
Unitless
score
No unit conversion; weighted
CrUX
Time to First Byte
p75_ttfb
Milliseconds
Renamed p75_ttfb_ms; weighted
CrUX
First Contentful
Paint
p75_fcp
Milliseconds
Renamed p75_fcp_ms; descriptive
CrUX
Phone density
phoneDensity
Proportion
Retained as form-factor evidence
CrUX
Desktop density
desktopDensity
Proportion
Retained as form-factor evidence
CrUX
Tablet density
tabletDensity
Proportion
Retained as form-factor evidence
HTTP
Archive
Page weight
bytesTotal
Bytes
Divided by 1,000,000 to obtain MB;
weighted
HTTP
Archive
Request count
reqTotal
Requests
Desktop and mobile counts averaged;
weighted
HTTP
Archive
JavaScript size
bytesJS
Bytes
Divided by 1,000,000 to obtain MB;
descriptive
HTTP
Archive
Image size
bytesImg
Bytes
Divided by 1,000,000 to obtain MB;
descriptive
HTTP
Archive
Speed Index
SpeedIndex
Milliseconds
Desktop and mobile values averaged;
weighted
Page 2937
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
HTTP
Archive
Fully loaded time
fullyLoaded
Milliseconds
Desktop and mobile values averaged;
weighted
HTTP
Archive
HTTPS ratio
numHttps,
reqTotal
Proportion
Pooled HTTPS requests divided by pooled
requests; weighted
HTTP
Archive
CDN detection
cdn
Categorical
Provider detected or no provider detected;
weighted
The source workbook also retained CSS size, font size, redirect count, domain count, page-level TTFB, rendering
start, and visual completion fields as non-scoring audit variables.
A.4 Origin Normalization and Dataset-Matching Rules
The matching workflow applied six rules:
1. Each institutional website was validated as official, and redirects, JavaScript rendering, or blocked automated
access were recorded without automatic exclusion.
2. During inventory preparation, each validated address was reduced to its scheme and host. The notebook then
trimmed spaces, converted origins to lowercase, and removed trailing slashes while retaining the observed host
form required for exact CrUX matching [6].
3. CrUX matching required the normalized origin for month 202605. HTTP Archive matching required the
corresponding root page in the May 1, 2026 crawl with a desktop or mobile record.
4. The HTTP Archive root-page list was created by adding a trailing slash, and httparchive.fn.GET_ORIGIN was
used to derive a comparable origin from each matched page.
5. Missing cases were checked against the validated inventory. No unrelated or unofficial alternate domain was
substituted to increase coverage.
6. Coverage was labeled complete data when both sources matched, CrUX only when only field evidence matched,
and insufficient data when neither source returned usable evidence.
Validated URLs and normalized origins were stored separately to preserve redirect and matching traceability.
A.5 Indicator Cut-Offs, Point Assignments, and Weights
Weighted indicators were translated to 100, 60, or 30 points. CDN detection used 100 points when a provider
was detected and 60 points when none was detected.
Table A4. CrUX scoring rules
Indicator
100 points
60 points
30 points
Weight
LCP
<= 2,500 ms
> 2,500 to 4,000 ms
> 4,000 ms
30%
INP
<= 200 ms
> 200 to 500 ms
> 500 ms
25%
CLS
<= 0.10
> 0.10 to 0.25
> 0.25
25%
TTFB
<= 800 ms
> 800 to 1,800 ms
> 1,800 ms
20%
Table A5. HTTP Archive scoring rules
Indicator
100 points
60 points
30 points
Weight
Page weight
<= 2 MB
> 2 to 5 MB
> 5 MB
25%
Request count
<= 75
> 75 to 150
> 150
20%
Page 2938
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
Indicator
100 points
60 points
30 points
Weight
Speed Index
<= 3,000 ms
> 3,000 to 6,000 ms
> 6,000 ms
25%
Fully loaded time
<= 10,000 ms
> 10,000 to 20,000 ms
> 20,000 ms
15%
HTTPS ratio
>= 0.99
>= 0.95 to < 0.99
< 0.95
10%
CDN detection
Provider detected
No provider detected
Not used
5%
The LCP, INP, and CLS cut-offs followed Core Web Vitals guidance, while TTFB followed supporting web-
performance guidance. The HTTP Archive cut-offs were researcher-defined analytic bands used for this study
and were not treated as official HEI website standards [27], [28].
A.6 Component and Overall Score Formulas
Let P_LCP, P_INP, P_CLS, and P_TTFB represent the points assigned to the four CrUX indicators.
CrUX_i = 0.30(P_LCP) + 0.25(P_INP) + 0.25(P_CLS) + 0.20(P_TTFB)
Let P_PW, P_REQ, P_SI, P_FLT, P_HTTPS, and P_CDN represent the points assigned to the six HTTP Archive
indicators.
HA_i = 0.25(P_PW) + 0.20(P_REQ) + 0.25(P_SI) + 0.15(P_FLT) + 0.10(P_HTTPS) + 0.05(P_CDN)
For origins with both components, the overall readiness score was:
Readiness_i = (CrUX_i + HA_i) / 2
The resulting score was reported to one decimal place. Readiness levels were interpreted using the following
rules:
Score condition
Readiness classification
Score >= 80
High readiness
Score >= 60 and < 80
Moderate readiness
Score < 60
Low readiness
No CrUX or HTTP Archive evidence
Insufficient data
The decimal-safe boundaries were score >= 80, score >= 60 and < 80, and score < 60. CrUX-only component
scores were labeled as limited data rather than treated as complete overall scores.
A.7 Desktop and Mobile Aggregation Procedure
HTTP Archive returned one desktop and one mobile row for each of the 25 matched origins. Because the website
origin was the unit of analysis, the client records were combined before scoring.
For continuous page-level indicators, the origin-level value was the arithmetic mean of the desktop and mobile
observations:
x-bar_i = (x_desktop,i + x_mobile,i) / 2
This procedure was applied to page weight, request count, JavaScript size, image size, Speed Index, and fully
loaded time. Page weight, JavaScript size, and image size were first extracted in bytes and converted to decimal
megabytes using:
MB = bytes / 1,000,000
The origin-level HTTPS ratio was calculated from the combined request totals:
Page 2939
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
HTTPSRatio_i = (HTTPS_desktop,i + HTTPS_mobile,i) / (Requests_desktop,i + Requests_mobile,i)
CDN detection was present when a provider name appeared in either client record. The original client rows
remained in the HTTP Archive Raw worksheet for traceability.
A.8 Sensitivity-Analysis Schemes and Comparison Measures
Sensitivity analysis was conducted using the 25 complete-data origins. The original model was compared with
three alternative weighting schemes.
Table A6. Weighting schemes
Scheme
CrUX indicators
HTTP Archive
indicators
Component combination
Baseline
Original weights
Original weights
50% CrUX, 50% HTTP Archive
Equal within
components
25% each
16.67% each
50% CrUX, 50% HTTP Archive
CrUX emphasis
Original weights
Original weights
60% CrUX, 40% HTTP Archive
HTTP Archive
emphasis
Original weights
Original weights
40% CrUX, 60% HTTP Archive
Comparisons used score change, absolute score change, classification agreement, category changes, Spearman
rank correlation, readiness counts, and HEI-type means.
Table A7. Sensitivity-analysis results
Scheme
Unchanged
Agreement
Mean
absolute
change
Maximum
change
Spearman
rho
High
Moderate
Low
Baseline
25 of 25
100%
0.00
0.00
1.000
5
13
7
Equal within
components
22 of 25
88%
3.41
6.42
0.980
6
14
5
CrUX emphasis
21 of 25
84%
2.84
6.15
0.962
7
13
5
HTTP Archive
emphasis
24 of 25
96%
2.84
6.15
0.964
4
14
7
Across the alternative schemes, 84 percent to 96 percent of classifications remained unchanged and Spearman
correlations ranged from 0.962 to 0.980. Changes occurred mainly near the 60-point and 80-point boundaries.
Full calculations are recorded in ACN_HEI_R1_Sensitivity_Analysis.xlsx.
A.9 Reproducibility Files and Version Control
Table A8 identifies the version-controlled files used for validation, extraction, scoring, sensitivity analysis, and
manuscript revision.
Page 2940
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
Table A8. Reproducibility files
File
Purpose
Version-
control
status
ACN_HEI_Source_Inventory_v3_Extraction_Pack.xlsx
Validated URLs,
normalized origins,
root pages, raw-
output templates,
and SQL queries
Source
inventory
and
extraction
package
ACN_HEI_BigQuery_Extraction_Colab.ipynb
Parameterized
CrUX and HTTP
Archive extraction
workflow
Retained
extraction
notebook
ACN_HEI_Extraction_Cleaned_Scored_v4_FREEZE.xlsx
Raw data,
summaries,
component scores,
classifications, and
audit record
Frozen
July 9,
2026
ACN_HEI_R1_Sensitivity_Analysis.xlsx
Alternative
weighting schemes,
recalculated scores,
category changes,
and ranking
comparisons
Derived
revision
file
ACN_HEI_Final_Draft_IJLTEMAS_v16_Editorial_Refinement.docx
Manuscript
submitted for
journal evaluation
Retained
as the
original
submission
record
ACN_HEI_Final_Draft_IJLTEMAS_v20_R1_FINAL_SUBMISSION.docx
Final revised
manuscript for
journal resubmission
Final
revised
submission
copy
The frozen workbook remains the unchanged source for the original data and scores. Corrections require a new
version, while the sensitivity workbook remains a separate derived file.