- Google's claim Reduce render-blocking resources — “save 2,330 ms”
- Truth A critical render-blocking dependency chain is present in the trace.
- Materiality The 2,330 ms is a counterfactual inside the simulator's own graph. It cannot be subtracted from the 2,784 ms field p75.
- Jurisdiction The stylesheet and its loading strategy are controllable in this stack.
- The fix The corrected template block, verbatim for your platform — not “reduce render-blocking resources”.
- Predicted movement An arithmetic range recomputed against your field data, with the working shown — never copied from the simulator.
- Verification Re-run after deployment and compare reality with the predicted range.
PageSpeed Audit
Every warning adjudicated. $497. No subscription.
Know which PageSpeed warnings are real, what Google's report missed, and exactly what to ship — then use 2 included re-runs to hold the report to its own predictions.
$497 one-time · one commercial URL · plus applicable tax
The published worked example, in full: inspect the framework and its case file →
- Every PSI warning adjudicated, with evidence
- The Silent Bottleneck the warning list never names
- Platform-native code fixes, not generic advice
- One URL on mobile and desktop
- 2 verification re-runs
- Nothing locks at checkout. You choose the URL after purchase — your dashboard can read your sitemap and recommend the strongest target first.
- The clock starts when you do. Your 30-day window begins at the first run, not on the purchase date.
- No account needed before paying. Access arrives by email once the payment settles.
- Invoice, VAT or purchase-order question? Ask before you buy — purchasing questions get a direct answer.
See what arrives
Not another score. A decision record your team can ship.
Every finding connects Google's original claim to captured evidence, a verdict, the exact implementation change, a predicted range with its arithmetic, and a verification step. This is the shape of all of them.
- Evidence quoted, not summarised. Every verdict cites the line of your trace, markup or headers that decides it.
- Four verdicts, not a score. Real, Wrong, Trivial or Locked — including the two that conventional audits have no words for.
- Held to account. A prediction with no re-run is a marketing claim; 2 re-runs are included to test it.
Everything included
Know what matters, what to ship, and whether it worked.
The same deliverable, grouped by the three questions you are actually deciding — before, during and after implementation.
01
Know what matters
- Every PSI flag adjudicated — Real, Wrong, Trivial or Locked, each verdict with its evidence quoted
- The Noise Rate — which of Google's warnings deserve zero engineering time, with proof
- The Silent Bottleneck — the real-user constraint no flag names
- Findings in no free report — why lab and field disagree, your CDN's live settings, the named plugin or app, back/forward-cache blockers, font fallback overrides, the legacy JavaScript still shipped to modern browsers — measured, not re-read from PSI
02
Know exactly what to ship
- Platform-native fixes — the
theme.liquidsnippet, thenext.config.jschange, the corrected<head>block, verbatim - Edge apply pack — only the settings that should change on your CDN
- Prioritised, with predicted ranges — the arithmetic shown, recomputed against your field data
- 1 URL on mobile + desktop — lab (Lighthouse) and real-user (CrUX) data, never conflated
Supported commerce and edge platforms
Deep commerce modules for Shopify, WooCommerce, BigCommerce, Magento / Adobe Commerce, PrestaShop and Salesforce Commerce Cloud. Edge apply packs for Cloudflare, Akamai, Fastly, Bunny, Azure Front Door, AWS and Google Cloud.
03
Know whether it worked
- Executive summary + JSON — one plain-language page for leadership, every finding via the REST API
- Help choosing the URL — after purchase your dashboard reads your sitemap and Google's field data, groups pages into templates and recommends the best one, before your URL locks in
- 2 verification re-runs within 30 days — hold the report to its own predictions
Why one score is not proof a change worked →
Measure your page's own run-to-run variance, free →
Choose the right service model
Between a free score and a $3,000–$8,000 engagement.
This audit is deliberately narrow: one commercial URL, a complete evidence-backed decision record, and a fast path to implementation. That is a different purchase from a consulting engagement, and the honest comparison says so — including when not to buy this.
| Decision factor | This audit | Specialist engagement |
|---|---|---|
| Price | $497 once | $3,000–$8,000 typical |
| Delivery | Less than 24 hours | Usually 2–4 weeks |
| Scope | One commercial URL, mobile + desktop | Broader site, architecture and stakeholders |
| Implementation output | Exact platform-native fixes and edge settings | Varies — often workshops, backlog and advisory |
| Verification | 2 re-runs included | Depends on the contract |
| Best fit | A page that needs an evidence-backed fix plan now | Systemic architecture or organisational change |
The full comparison — DIY, monitoring, agency, or this — including when it is the wrong tool →
Why a prediction is not proof
A forecast is only worth something when reality gets a vote.
We ran a controlled pilot on Lighthouse's own “estimated savings”. In 8 of 8 cases the simulated metric moved by at least the amount the tool predicted — a ratio of 1.0–1.9×.
That is Lighthouse agreeing with Lighthouse. It is internal consistency, not an independently timed saving, and it is precisely why this audit recomputes impact against your field data instead of quoting the simulator — and why 2 re-runs are included rather than sold as an add-on.
From payment to proof
What happens after payment?
-
Account, automatically
Created from your checkout email — you receive a secure link to set your password. Returning customers just sign in.
-
Your window, your pace
Enter the URL whenever you're ready. It locks to this purchase at the first run, and your 30-day window starts then — buy today, audit when it suits you. Not sure which page? The dashboard analyzes your site and recommends one, with the evidence.
-
Report in less than 24 hours
Every verdict, the deep findings, the executive summary and the JSON — delivered to your dashboard and inbox within 24 hours. The report ships when the method completes: every check run, nothing skipped.
-
Ship, then verify
Every fix carries a predicted range with its arithmetic shown. Deploy, then spend your 2 included re-runs comparing reality against that range.
Not deciding today?
Pressure-test the economics — or take the three-email brief.
One path helps you quantify the opportunity on your own numbers. The other explains exactly what you are buying, when it is the wrong tool, and what happens if the report is wrong.
Free decision tool
Will a faster page plausibly repay the audit?
The Speed → Revenue calculator pulls your real-user speed from Google's field data and prices a faster page against your own traffic, conversion rate and order value — as an evidence-bounded range, not a fake point estimate. Free, no account.
Evidence base: the millisecond register, graded by research design.
Three emails, then it stops
Take the decision brief
Everything you'd want checked before spending $497, in writing:
- Email 1What $497 actually buys — the deliverable, item by item.
- Email 2DIY, a $3,000 agency, or this — the honest comparison, including when not to buy.
- Email 3If the report is wrong — predicted ranges, re-runs, and the money-back terms.
One page. Every warning decided.
$497 once · plus applicable tax · delivered in less than 24 hours · 2 verification re-runs included
- Report or money back
- Stripe checkout
- Delivered < 24 h
- No subscription
Every report is our Noise & Silence framework, executed end to end — and if your page is already fast, the report says so. We never pad. Questions first? Talk to us — we typically reply within one business day.