Frequently asked questions
How the SEO audit works, where the keyword research data comes from, what connecting Google Search Console gets you, and what this SEO platform will not claim to measure.
Can’t find it here? Email us and we’ll answer, then add it to this page.
What is AI Marketing Lab?
An SEO platform for UK small businesses. It joins your Google Search Console rankings and your GA4 traffic in one place, audits your marketing site for technical problems, and shows whether AI answer engines like ChatGPT can read you.
Most SEO tools are built for agencies and assume you already know what a canonical tag or a backlink profile is. This one is built for the person who owns the business and wants to know what to do this week.
Everything it reports is measured rather than modelled. Your rankings come from your own Search Console property, your traffic from your own GA4, and the answer-engine findings from reading your live robots.txt and HTML. Where a figure cannot be measured, the tool says so instead of estimating it.
What does the SEO audit check?
The SEO audit crawls your site and checks titles, meta descriptions, heading structure, canonical tags, image alt text, structured data, Open Graph tags, robots.txt, your sitemap, broken internal links, and Core Web Vitals via Google PageSpeed Insights.
Every finding carries three things: what is wrong, which page it is on, and the fix. Where the fix is deterministic — a canonical tag, an Open Graph block, a robots.txt line — the tool generates the exact code from your own site's values, ready to paste.
Where the fix needs writing, such as a page title or a meta description, you can ask for suggestions generated from that page's actual heading and opening content rather than from the rule in the abstract.
The audit also folds in answer-engine readiness, so one run tells you both how a search engine sees the page and how an AI answer engine would try to quote it.
How do I connect Google Search Console?
Sign in with Google under Settings, then pick your property. We request read-only access to Search Console and nothing else. If you have a domain property, choose the sc-domain entry rather than a URL prefix so every subdomain is covered.
Search Console is free from Google and is the single most important thing to connect. It is the only source that knows which queries actually showed your site, at what average position, and how many people clicked. Every keyword figure in the tool comes from there.
Two things worth knowing. Search Console publishes on a two-to-three day lag, so the most recent days are always empty — that is Google, not a fault. And a brand-new property will have very little data for the first few weeks, which the tool says plainly rather than showing you an empty table and letting you draw your own conclusion.
Where does your keyword research data come from?
From your own Google Search Console queries and from Google Trends related searches. Both are free. We do not buy a third-party keyword database, so you will not see invented search volumes or difficulty scores anywhere in the tool.
This is a deliberate trade. A licensed keyword dataset would let us print a search volume next to every term, and those numbers are modelled estimates from a crawled ranking corpus rather than counts. We would rather show fewer columns that are true than more columns you cannot verify.
What you get instead is your real position, your real impressions, your real click-through rate, and Google Trends' own rising-query signal for terms you do not yet rank for. The keyword planner then sorts every term into buckets — nearly there, seen but not clicked, losing ground, no presence — based on what those measurements say.
Can I run an SEO audit without connecting anything?
Yes. The site audit and the answer-engine check both work on any public URL with no account connection, because they read the site itself. Connecting Search Console adds your ranking and traffic data on top.
You can point either at a competitor's domain as well as your own. Nothing about reading a public page requires you to own it.
What connection unlocks is the half we cannot see from outside: which queries you appear for, how many people click, and how that has moved. No amount of crawling a website reveals that — only the site owner's own Search Console does.
How do you check whether AI answer engines can read my site?
We read your robots.txt and report, crawler by crawler, whether each answer engine is allowed to fetch your pages. Then we fetch a page twice — once as ChatGPT's crawler, once as a browser — and show you the difference.
The distinction that matters most is between crawlers that build live, cited answers and crawlers that only gather training data. OAI-SearchBot powers ChatGPT's search results; GPTBot feeds model training. Blocking those two means completely different things, and most tools lump them together.
The second half is the one people find surprising. Answer engines do not run JavaScript. If your content only appears after a client-side render, the crawler receives an almost empty page no matter how good the site looks in a browser. Fetching as both shows you exactly what the engine gets.
Anything that claims to measure AI crawlers using a JavaScript tracking tag is measuring nothing, because the crawlers never execute the tag.
What is the difference between SEO and answer-engine optimisation?
SEO gets you ranked in a list of blue links. Answer-engine optimisation gets you quoted inside the answer itself. Search and SEO used to be the same problem; now a page can rank well and still never be cited by ChatGPT, Claude or Google's AI Overviews.
The work overlaps but is not identical. Both want clean structure and server-rendered content. Answer engines additionally reward a direct answer placed immediately under a heading, headings phrased as the question being asked, and content marked up so a machine knows what it is looking at rather than guessing.
They also need permission to fetch you at all, which is a robots.txt question and not a content question. It is entirely possible to do every piece of on-page work correctly and remain invisible because one line excludes the crawler.
How much does it cost?
Nothing during the closed beta. We are testing with a small group of UK small business owners through summer 2026 to find out whether the tool actually moves their numbers before charging anyone. No card, no trial expiry, no surprise bills.
Beta users keep access on whatever plan we land on. We would rather learn whether this is worth paying for than collect card details early.
How is this different from Ahrefs or SEMrush?
Those are excellent tools built for SEO professionals. They cost over £100 a month and assume expertise you may not have. If you are searching for SEO software as a business owner rather than a specialist, this is built for you: fewer levers, clearer actions, UK-focused.
The other difference is what we will and will not claim. Large SEO platforms show estimated traffic, domain authority scores and keyword difficulty for any domain you type. Those are models built on licensed data, not measurements, and they are frequently wrong at the scale a small business operates at.
We show what can be measured and name what cannot. On the competitor screen you will find a short list of metrics we deliberately do not display, with the reason next to each.
The answer-engine work is also genuinely different rather than a feature checkbox. Most tools have not caught up with AI search at all.
Will the AI write content for me?
No, and that is deliberate. There is no shortage of tools that generate a blog post. There is a shortage of tools that tell you which post is worth writing and why. We identify the opportunity, show the Search Console evidence, and leave the writing to you.
Where AI is used, it is used for judgement rather than volume: suggesting three title options built from a page's actual subject, writing the narrative that explains which audit findings to tackle first, or naming the companies an AI assistant would offer as alternatives to you.
Every AI-generated suggestion is labelled as one, and anything it names is verified before you see it. Competitor domains, for instance, are fetched to confirm they exist — a model will confidently invent a plausible company, and an unverified list is worse than a short one.
Is my data safe, and where is it stored?
Your data is held in a UK-region Postgres database with row-level security, so an account can only ever read its own rows. We do not sell it, share it, or train models on it. Google access is read-only and you can disconnect at any time.
We read your Search Console and Analytics metrics on demand through Google's APIs and display them. We do not store the underlying user-level data.
The privacy notice sets out exactly what is collected, the lawful basis for each purpose, how long it is kept and how to have it deleted.
When will it be available to everyone?
We are aiming to open access more widely in autumn 2026, after the beta. If you would like to be on the list, sign up now — beta users get continued access automatically.
The beta is small on purpose. We would rather fix what a handful of businesses find than scale something that has not been used in anger yet.