A 29-Second Application-Page Lab Result
A May 2026 independent research snapshot of a public nonprofit application page, including what the measurements can and cannot establish.
Independent research · not client work
Independent public-site research. The organization did not hire Toledo Technologies, and this note does not claim a delivered fix or measured outcome.
Limits
The measurements are point-in-time synthetic tests. Toledo Technologies had no applicant analytics, server telemetry, or deployment access.
Recorded evidence
Captured mobile Largest Contentful Paint: 29.1 seconds
Captured mobile First Contentful Paint: 6.98 seconds
Captured mobile Lighthouse accessibility score: 100/100
No conversion, completion-rate, or post-fix outcome data was available
Evidence Label
This is an independent research note, not client work. The organization did not request or pay for the audit. The numbers below are dated public-page measurements, not a claim about every visitor or a delivered result.
The Subject
A New Haven nonprofit running an accelerator program for founders from underrepresented backgrounds. Application-driven, mission-driven, well-staffed, respected. The website hosts the program-application form, the case for support for funders, and the board page.
The Premise of the Audit
The subject did not ask for this review. We selected a public application page to compare Lighthouse’s accessibility and performance signals on the same page. Toledo Technologies had no private access to the site or organization.
It did not.
What We Found
| Metric | Mobile | Desktop |
|---|---|---|
| Performance | 40/100 | 35/100 |
| Largest Contentful Paint | 29.1s | 3.3s |
| First Contentful Paint | 6.98s | 1.95s |
| Total Blocking Time | 574ms | 1.82s |
| Accessibility | 100 | 96 |
| SEO | 92 | 92 |
| Best Practices | 96 | 96 |
The captured accessibility and performance scores point in different directions. The test demonstrates a slow synthetic render under that profile; it does not establish who abandoned the page or how many applications were affected.
What Is Causing It
Without server access, the run supports three investigation paths rather than a root-cause claim:
- Desktop main-thread blocking is 1.82 seconds. That is too much JavaScript executing on page load. Not a network problem; a payload problem.
- First Contentful Paint at 6.98 seconds on mobile warrants separating server response time from render-blocking resources in a trace.
- The site timed out from one test network while PSI completed elsewhere. Geography, transient availability, network policy, and hosting are all possible explanations. The public test cannot choose among them.
The next responsible step would be repeated regional tests plus a trace of the critical request chain. A hosting change should only be recommended if those tests isolate server or regional latency; JavaScript work should only be scoped after the main-thread profile identifies the blocking code.
What We Did Not Do
We did not pitch a rebuild. The site is good. The accessibility work is real. The technical investment is uneven, not absent.
We did not contact the subject or represent this as delivered work. This public version exists to show the reasoning and its limits.
What This Looks Like For A Subject Who Wants Help
For a nonprofit at this performance level, a paid audit should end with a smallest-credible intervention: measured baseline, named bottleneck, affected pages, acceptance target, and a quote. The public snapshot is not sufficient to promise a sub-five-second result or prescribe a migration.
What This Means For Other Nonprofits
If an automated accessibility score is high and performance is low, test both with their proper tools: manual accessibility checks, repeatable performance traces, and real-user data where available. A Lighthouse score is evidence, not certification or a complete user study.
Need the same question answered on your system?
Paid discovery produces a scoped document with explicit evidence and limitations. It does not begin with a promised outcome.