AI SEO Services | Umer Qureshi

AI SEO has become a label for two different jobs, and buying the wrong one wastes a year.

The first job is using AI to do SEO faster: generating drafts, clustering keywords, building internal links at scale. The second is making a site perform in a search environment where AI now writes the answer. Most agencies selling AI SEO are selling the first. The revenue problem almost always sits in the second.

This engagement treats them as one programme with one measurement: qualified enquiries. Technical foundations, ranked search and answer visibility are not separate workstreams here, because the same page has to satisfy a crawler, a ranking system and an extraction model at once.

The three layers, run together

A site that performs in AI search needs all three layers working, and they fail in a predictable order.

Technical. Crawlability, speed, canonical hygiene, clean URLs, correct status codes. Unglamorous, and the reason a surprising number of sites never had a chance. A page that returns the wrong status code or canonicalises to a competitor page cannot rank or be cited.

Ranked search. The classic work: matching pages to real demand, building the commercial pages that intent-heavy queries expect, earning the links that make them competitive.

Answer visibility. Structuring those same pages so an assistant can lift a claim and attribute it to you. This is the layer most programmes skip, and it is now where a growing share of high-intent traffic is decided.

Where AI genuinely helps, and where it does not

AI is very good at the parts of SEO that are wide and shallow: clustering thousands of queries, finding orphan pages, spotting patterns across a large estate, drafting a first pass. I use it for all of that, and it is why a one-person practice can cover ground that used to need a team.

It is bad at the parts that are narrow and deep: deciding what a business should actually say, knowing which claim is defensible, judging whether a source is real. Those decisions stay with a person, because the cost of getting them wrong is a client publishing something untrue.

The practical rule I work to: AI does the sweeping, a human does the deciding, and nothing ships without the source being opened and read.

What gets measured

Pageviews are a poor measure of an AI SEO programme, because the mechanism now separates visibility from clicks. A page can gain visibility and lose clicks in the same month.

So the reporting is built on four numbers instead. Qualified enquiries, which is the only one that pays. Citations won across the major assistants, tracked against a fixed question set. Non-brand impressions split from brand, because brand searches flatter a report and tell you nothing. And position for the specific commercial queries that map to a service page, rather than an average across every query the site touches.

Who this suits

This fits a business with something specific to sell and a real reason to be chosen: a supplier with genuine technical specifications, a service firm with a defined method, a software company with a clear category. Specifics are the raw material, and a business without them will struggle regardless of who does the work.

It suits less well a business that wants volume traffic for advertising revenue, or one that needs a local map presence above all else, where the levers are different.

What the first ninety days look like

Month one is diagnosis and foundations. The technical crawl, the brand versus non-brand split, the citation baseline, and whatever is actively blocking revenue gets fixed. Nothing new is published, because publishing into a site with structural faults wastes the work.

Month two builds the commercial layer: the service and comparison pages that intent-heavy queries expect, structured so they satisfy a ranking system and an extraction model at once. This is usually where the largest single gain sits, because most sites have never had these pages at all.

Month three measures and adjusts. Some pages move quickly, some do not, and the fourth month is planned from what actually happened rather than from the original assumption. Anyone presenting a fixed twelve-month content calendar before seeing the first results is selling a schedule, not a programme.

Deciding whether this is the right engagement

If your impressions are steady and your clicks are falling, this is the right engagement and the timing is now rather than after the rankings go.

If you have no commercial pages and a busy blog, start here too, because the fix is structural and publishing more will not touch it.

If the site is new, has few pages and no measurement, start with the audit instead. It costs less, it ends with a document you own, and it will tell you whether a programme is justified. Either way the first conversation is short and mostly questions.

What the engagement includes

A technical baseline. Crawl, index and status-code audit across the whole estate, with the fixes ordered by what is actually blocking revenue rather than by severity score.
Demand mapping. What your buyers search at each stage, priced against real difficulty, so effort goes where it can win rather than where volume looks biggest.
Commercial page architecture. The service, product and comparison pages that intent-heavy queries expect, built to satisfy ranking and extraction at the same time.
Answer visibility work. Structure, schema and corroboration so the same pages can be quoted by assistants, tracked against a fixed question set.
Reporting that separates brand from non-brand. Monthly, with citations and enquiries as headline numbers and brand search stripped out so the trend is real.

How the work runs

Establish the baseline. Technical crawl, current rankings, current citations, and the brand versus non-brand split. Most sites have never seen the last one.
Map demand to difficulty. Price every candidate cluster before committing to it, so the programme starts where the odds are good.
Fix the foundations. Clear the technical blockers first. There is no point publishing into a site that cannot be crawled cleanly.
Build the commercial layer. Service and comparison pages for the queries that carry buying intent, structured for both ranking and extraction.
Measure and compound. Monthly re-measurement, with the next month's work chosen from what moved.

Proof

The same three-layer approach is documented, with numbers, in the fleet software case study.

Talk about your situation

The first conversation is short and mostly questions. Get in touch and tell me what you are trying to fix. Or see how this is priced.

Frequently Asked Questions (FAQs)

What is AI SEO?

AI SEO covers two things that are often confused: using AI tooling to run SEO work faster, and optimising a site to perform in AI-driven search results such as AI Overviews and assistant answers. This engagement does both, but the second is where the commercial risk sits for most businesses right now.

Is AI SEO different from normal SEO?

The foundations are the same. Crawlability, page structure, matching real demand and earning authority all still decide outcomes. What is new is that a page must now also survive extraction: an assistant reading it, lifting a claim and attributing it. That changes how pages are written, not whether the fundamentals matter.

Will AI written content hurt my rankings?

Google's position is that it rewards helpful content regardless of how it was produced, and penalises content produced at scale to manipulate rankings. In practice the risk is not the tool, it is the absence of review: unverified statistics, invented citations and pages that say nothing specific. Everything published under this engagement has its sources opened and checked.

How is success measured?

Qualified enquiries first. Then citations won across the major assistants against a fixed question set, non-brand impressions separated from brand, and position on the specific commercial queries tied to a service page. Site-wide averages and raw pageviews are deliberately not headline metrics.