Most people approach this backwards. They hire first, then find out their site had problems no backlink or content plan could fix.

Run the checks below first. They take an afternoon, cost nothing, and they tell you whether you need an ai seo consultancy at all, or whether you have cheaper problems to solve first.

Score yourself honestly on each. The failures are your actual brief.

Check 1: Can Your Most Important Page Be Quoted?

Open your single most commercially important page. Read the first three sentences.

Pass: they answer the question a buyer would arrive with, in plain self contained statements.

Fail: they describe your company, or warm up slowly, or bury the answer below a hero image and three scroll lengths.

This matters more than it used to. Generated answers lift self contained passages and attribute them. A page that defines its own value in the opening lines is quotable. One that eases in is not, no matter how good the content lower down.

Most sites fail this on their highest value page, and fixing it costs nothing.

Check 2: Do Two Pages Compete for the Same Thing?

Search your own site for your top five topics. Count how many pages plausibly answer each.

Pass: one clear page per topic.

Fail: two or three pages circling the same intent, splitting authority and leaving the search engine to guess which one matters.

This is the most common fixable problem, and consolidation usually produces movement faster than new content. It is editing work, not production work, which makes it cheap to test before committing to anything larger.

If you fail this check, you have found your first project, and it does not require a consultancy to diagnose.

Check 3: Are Your Claims Specific Enough to Cite?

Scan a key page for its factual claims. Look at how they are phrased.

Pass: specific figures, named entities, dates, concrete detail.

Fail: general assertions that could appear on any competitor’s site.

Vague claims get paraphrased and lose attribution. Specific ones get quoted and carry your name with them. This is a writing fix rather than a technical one, and it serves human readers too, which is a reasonable sign it will hold as search behavior shifts.

Check 4: Can the Important Pages Actually Be Crawled?

Check that your key pages return cleanly, load reasonably fast, and are not blocked in robots or buried behind scripts.

Pass: the pages that matter are reachable and quick.

Fail: content that cannot be crawled, which cannot be ranked or cited regardless of quality.

This is unglamorous and decisive. No amount of content or link work overcomes a page a crawler cannot read. If you fail here, nothing else on the list matters until it is fixed.

Check 5: Do You Know Which Queries Bring Clicks vs Citations?

Look at your search console. Split your terms into informational and commercial, and compare clicks against impressions.

Pass: you can see which queries convert attention into visits and which mostly feed generated answers.

Fail: you judge everything by ranking position, which no longer predicts traffic reliably.

A page can sit near the top for a definitional term and receive almost no clicks, because a generated summary answered the query. That is normal now, not a failure. But if you cannot see the split, you cannot tell a real problem from an expected one.

Check 6: When You Ask an Assistant Your Buyer’s Questions, Do You Appear?

Take your five most important buyer questions. Ask them to a generative assistant. Note which sources it draws on.

Pass: you show up, or you can see clearly what the cited sources have in common structurally.

Fail: you are absent and cannot tell why.

This is imperfect and currently the most honest visibility signal available. Repeat it monthly and the pattern tells you more than most paid reporting. It also costs nothing, which is the theme of this entire list.

Scoring Yourself

Count your passes.

Five or six passes: your foundation is sound. A consultancy here is about acceleration and competitive terms, not repair. Scope the engagement around growth.

Three or four: you have fixable problems you can start on yourself. Do the free fixes first, then bring in help for what remains. You will pay less and get more, because the consultancy is not spending your budget on things you could have done in an afternoon.

Two or fewer: do not hire yet. Fix the crawlability and consolidation problems first, because any money spent on content or links will leak out through the gaps until you do.

Where Outside Help Actually Earns Its Fee

Once the free fixes are done, the remaining work is where a consultancy is genuinely worth it.

Intent research that goes beyond volume, into what buyers actually want when they ask. Competitive analysis of why specific pages currently rank. Ongoing measurement in an environment where position and traffic have come apart. These are harder to do well alone, and they compound slowly, which is why this discipline rewards a longer relationship than a single quarter.

The connection to broader strategy is worth noting too. The work of structuring information so machines interpret it reliably is the same discipline whether the consumer is a search engine, a generative assistant, or an internal system. A firm like Notionmind frames its own positioning around being found on Google and cited by AI, which is the right way to think about it. Those are two outcomes, built on one foundation.

Run the six checks first. If you fail most of them, the honest answer is that you have homework before you have a hiring decision, and that homework is free.