Digital Marketing

What a website conversion review should check: CRO, SEO, AI search and tracking

10 min readBy Davinder Kumar Meen

A useful conversion review checks four things on the pages that matter: whether a visitor can see what to do next and do it easily (conversion), whether search engines can find and understand the page (SEO), whether AI answer engines can fetch and quote it (AI search), and whether your analytics would tell you if any of that changed (tracking). About half of these checks can be run by a script in minutes. The other half, such as whether the offer is right and whether the page matches the ad that sent the visitor, still need a person who knows your customers.

Key takeaways

  • Review the pages that bring in enquiries or sales, not the whole site: the home page, the main service or category pages, landing pages for paid campaigns, and the contact or checkout flow.
  • Automated checks are good at presence and size: is there a call to action, how many form fields, is the page indexable, are Core Web Vitals within Google's "good" thresholds, are the AI search crawlers allowed, is the analytics tag present.
  • Automated checks cannot judge the offer, the message match with your ads, or why visitors leave. Those need a person, session recordings or interviews.
  • Tracking comes first. If your key events are not firing, every later improvement is a guess you cannot verify.

What is a conversion review, and how is it different from an SEO audit?

A conversion review asks one question of each important page: does this page turn the right visitors into enquiries or orders, and would we know if it stopped doing so? An SEO audit asks whether search engines can find and rank the page. The two overlap, because a page that nobody finds cannot convert, and a page that ranks but confuses visitors wastes the traffic. A review that also looks at AI search and tracking covers the full path: found, read, acted on, measured.

What can an automated check see on a page?

Quite a lot, as long as you read the results as signals rather than verdicts. These are the checks a script can run reliably against the HTML, the response headers and a performance test.

Area Check What a script can tell you What it cannot
Conversion Call to action near the top Whether any of the first links or buttons uses an action word such as "book", "quote" or "call" Whether that is the right action for this visitor
Conversion Lead form Number of visible fields on each form Whether each field is needed
Conversion Trust signals Whether reviews, testimonials or ratings appear on the page or in structured data Whether the proof is believable or recent
Conversion Contact details Phone, email or contact links present Whether anyone answers them
Conversion Mobile viewport width=device-width viewport tag present How the page feels on a real phone
Speed LCP, CLS, INP Lab and field values against Google's thresholds Which fix your developers can ship first
SEO Title, meta description, H1 Present, unique, sensible length Whether they would earn the click
SEO Indexability noindex, canonical target, robots.txt, XML sitemap Whether Google chooses to rank it
AI search Crawler access robots.txt rules for search crawlers such as OAI-SearchBot Whether any engine cites you
AI search Answer-first content A direct answer near the top, question-style headings Whether the answer is the best one
AI search Entity and authorship Organization structured data, named author, visible dates Whether those facts are accurate
Tracking Tags present Google tag or Tag Manager container in the page Whether events fire with the right parameters

Take the speed row as an example of how precise a script can be. Google's Core Web Vitals thresholds for a "good" page are a Largest Contentful Paint of 2.5 seconds or less, an Interaction to Next Paint of 200 milliseconds or less and a Cumulative Layout Shift of 0.1 or less, measured at the 75th percentile of page loads. INP replaced First Input Delay as a Core Web Vital on 12 March 2024, so any report that still grades FID is out of date. PageSpeed Insights shows two kinds of numbers: field data from the Chrome User Experience Report over the previous 28 days, and lab data from a single simulated Lighthouse run. Field data is what your visitors experienced; lab data is for debugging. If a review quotes only the lab score, ask for the field values too.

For a step-by-step look at the speed side on WordPress, see our guide to speeding up a WordPress site.

Which conversion checks still need a person?

The checks that decide most outcomes are the ones a script cannot make:

  1. The offer. Is there a clear reason to act now: a consultation, a quote within a stated time, a sample, a price range? A page can pass every automated check and still offer nothing worth a form.
  2. Message match. Google's advice to advertisers is to "choose a landing page that closely matches your ad and keywords". A person compares the ad, the search query and the page headline side by side.
  3. Objections. What would stop a buyer here: price, risk, time, trust? The page should answer the top two before the form.
  4. The form experience. Nielsen Norman Group's form guidance is to keep forms short, put labels close to their fields, avoid placeholder text as labels, and make errors visible and specific rather than signalled by colour alone. A script counts fields; a person fills the form in on a phone and notices what is annoying.
  5. Real behaviour. Where do visitors stop scrolling, what do they click that is not a link, where does the form get abandoned? Session recordings, on-site surveys and five short customer conversations answer "why" in a way no audit does.

Our landing page checklist goes through these page by page.

What should a review check on search and AI search?

The basics still decide whether a page is found at all: a unique title and meta description, one H1, a canonical tag pointing at itself (or at the version you want indexed), no accidental noindex, and an XML sitemap that lists the page. These take minutes to check and are the most common reason a good page gets no search traffic.

For AI search, Google is explicit that there are "no additional requirements to appear in AI Overviews or AI Mode", and no special structured data or AI text files are needed; a page must be indexed and eligible to show a snippet. So the AI search checks are mostly about access and clarity:

  • Crawler access. OpenAI says sites opted out of OAI-SearchBot "will not be shown in ChatGPT search answers", and recommends allowing it in robots.txt. Check that your robots.txt and CDN rules do not block the search crawlers you want.
  • Answer-first writing. Does each key page answer its main question in the first paragraph, under headings phrased as questions buyers ask?
  • Entity clarity. Organization structured data helps Google "disambiguate your organization in search results"; check the name, URL, logo, address and sameAs profiles match what is on the page.
  • Authorship and dates. Articles should show a named author and a visible published or updated date.

Our guide to getting cited by AI search engines covers this in depth.

How do you check that tracking works?

Seeing a tag in the page source only proves the tag is there. To trust the numbers you need to see events arrive:

  1. Open GA4 DebugView (Admin, Data display, DebugView) and enable debug mode with Tag Assistant or Tag Manager preview. Google's documentation notes that events will not appear in debug mode if consent mode is in use and analytics consent has not been given, so accept analytics cookies in your test session.
  2. Complete each conversion yourself: submit the form, click the phone number, start the checkout. Each should send one event with the parameters you expect, and only once.
  3. Check the key events. In GA4 you mark the events that matter as key events (Admin, Data display, Events, then the star). A standard property allows up to 30.
  4. Check consent. Google's consent mode has four consent types: ad_storage, analytics_storage, ad_user_data and ad_personalization. Confirm the defaults are what your privacy policy says, and that tags behave differently after a visitor accepts or declines.

If tracking fails here, fix it before anything else. A conversion test you cannot measure is not a test. Our GA4 event plan for lead-generation sites lists the events to set up.

How should you prioritise what the review finds?

Sort every finding by two things: how many visitors it affects and how much effort it takes. A broken form on the contact page beats a slow image on a blog post. A practical order:

  1. Anything that stops a conversion or its measurement (broken forms, missing key events, a noindex on a service page).
  2. Anything on the top five pages that fails a Google threshold or an accessibility basic, such as text contrast below the WCAG 2.2 minimum of 4.5:1.
  3. Message match and offer changes on paid landing pages, where you pay for every visit.
  4. Content and AI search improvements on informational pages.

Write each finding as "page, problem, evidence, fix, owner".

Impinge Solutions reviews conversion, search, AI search and tracking together as part of its conversion rate optimisation service. Run our free website audit on the page that brings in most of your enquiries to see where to start.

FAQ

How many pages should a conversion review cover?

Start with five to ten: the home page, your main service or category pages, any paid landing pages and the contact or checkout flow. These usually carry most enquiries and sales. Reviewing every page spreads the effort thin and buries the findings that matter under hundreds of minor ones.

Can an automated tool replace a conversion review?

No. A tool can check presence and size: whether a call to action, a form, a viewport tag, structured data or an analytics tag exists, and whether speed metrics meet Google's thresholds. It cannot judge whether the offer is compelling, whether the page matches the ad, or why visitors leave. Use the tool to clear the basics, then spend people's time on those questions.

Which speed numbers matter for conversions?

Google's Core Web Vitals: Largest Contentful Paint (good at 2.5 seconds or less), Interaction to Next Paint (good at 200 milliseconds or less) and Cumulative Layout Shift (good at 0.1 or less), measured at the 75th percentile of page loads. Prefer field data from PageSpeed Insights, which reflects real visitors over 28 days, to a single lab score.

Do I need special markup to appear in AI search answers?

Not according to Google, which says there are no additional requirements or special structured data needed for AI Overviews or AI Mode beyond being indexed and eligible for a snippet. For ChatGPT search, OpenAI says sites that block OAI-SearchBot will not be shown in its search answers, so check your robots.txt allows it.

Why check tracking in a conversion review?

Because every improvement you make later has to be measured. If the contact form does not send a key event, or sends it twice, you cannot tell whether a change helped. Test each conversion in GA4 DebugView with analytics consent accepted, and confirm each one is marked as a key event.

Sources

Are you ready to partner with us?

Our good-natured team is waiting to hear your business requirements.

Book a free project call

Stay informed with our newsletter