A 200-point checklist is not an audit. It's a spreadsheet. The audits that change a business look at three layers — technical, experiential, and commercial — and tell you the story those layers add up to.
We get asked for website audits constantly, and the request usually comes pre-shaped: someone has been told their Core Web Vitals are bad, or that a competitor's site "feels better," and they want a report that confirms what they already suspect.
We don't write that report. We write the one underneath it — the one that connects what's broken to what it's costing, and prioritizes the fix list against revenue, not severity.
The three layers we look at
Most agencies audit one layer well. The good ones audit two. A complete audit covers all three and shows how they interact, because the bug at layer one usually shows up as a conversion problem at layer three.
1. Technical foundation
Crawlability, indexation, schema, page speed, accessibility, security headers, mobile rendering, JavaScript hydration, third-party tag bloat. The plumbing. If this layer is broken, nothing else gets a chance to perform.
2. Experience and content
Information architecture, navigation clarity, page composition, hierarchy of CTAs, content depth against the topic's actual SERP, tone consistency, image and video weight. This is the layer most teams think of as design, but it's really decision design — how easy is it to do the next right thing on this page?
3. Commercial outcome
Conversion rate by template, drop-off by step, form abandonment, lifecycle handoff, attribution integrity, revenue per session by source. This is where the audit either earns its fee or doesn't, because if the report can't connect to dollars, it won't survive the next budget review.
What most audits miss
- Template-level math. Conversion rate is meaningless in aggregate. The PDP converts at 4%. The blog-to-PDP converts at 0.3%. Those need to be looked at separately.
- Cross-page friction. Audits that grade pages in isolation miss that the real problem is the journey — the form on page four that asks for what page two already collected.
- The competitor's actual page. If the audit doesn't open up the SERP and show you exactly what's outranking each priority page, you can't make calibrated calls about effort.
- The rendering reality. An audit run only against the rendered HTML misses everything that depends on a JavaScript framework loading correctly. We render. We diff. We compare.
How AI changed the audit
Two years ago, half of the audit was data collection. Crawls, screenshots, manual review of forty templates. Now we run a fleet of agents that crawl, render, screenshot, score, and cluster findings before a human ever opens the report.
What that frees us to do is the part that used to get short-changed: the prioritization. Instead of handing over 280 issues sorted by severity, we hand over twelve recommendations sorted by expected impact, with the math behind each one.
The audit isn't the deliverable. The decision the audit unlocks is.
Reading an audit you didn't pay for
If a vendor is pitching you with a free audit, you can still get value out of the document — but read it differently. Ignore the score at the top. Ignore the volume of issues. Look for two things: whether the report names a specific business outcome, and whether the recommendations are sequenced or just listed.
A free audit that says "your H1 tag is missing on 43 pages" is a product demo. A free audit that says "your top three category pages are losing 18% of qualified traffic to a JavaScript bug" is a conversation worth having.
When to run one
Before a redesign. Before a migration. Before a re-platform. Before you set next year's marketing goals. After a traffic drop that nobody can explain in one sentence. Twice a year if the site is the primary revenue surface for your business.
If it's been more than 12 months since the last one, you probably already know what the audit is going to say. The question is what you're going to do about it.