Enterprise SEO Consulting, in the Age of AI

How we measured. A live Ahrefs SERP fetch for “seo consulting”, United States, top 10 organic positions, 9 September 2026 — with one caveat that governs everything below. That pull returned a local pack at position 5 and a Wichita, Kansas directory URL at position 6, so a United States fetch resolved to one city rather than the country. Nothing here is presented as a national top ten; only the non-local results are characterised. Demand from Ahrefs Keywords Explorer, US, same date. Search Console figures from indexableai.com, 90 days to 8 September 2026, 1,063 query rows, plus a separate 28-day window. Compensation and scope from 29 public senior SEO and AI-search job postings collected June to August 2026, 16 disclosing an annual band — aggregate only, no employer named. One domain, one window, one self-selected sample: a case study, not a law.

What is enterprise SEO consulting in the age of AI?

---

Enterprise SEO consulting is advisory work sold into organisations too large to change direction quickly and too busy to run the experiments themselves. The scale problem is unchanged: thousands of pages, competing internal owners, a release train nobody in marketing controls. What changed is what the buyer should pay for.

The traditional output was a document — an audit, a roadmap, a prioritised backlog — valuable because a competent outsider could see the estate the client could not. That asymmetry survives. The document's shelf life does not, because the thing it measured, position in a list of ten blue links, stopped describing the whole outcome.

A page can rank and never be retrieved. On our own estate 18 pages sat in Google's top 10 across a 28-day scan and 3 were ever cited by an AI engine (n=223, Indexable, 2026). A roadmap that moves pages up the ranking and reports only that has described half of what happened.

Start by asking what a consultant leaves behind that still works in month seven, once the slides are archived.

Why do a buyer and a job-seeker land on the same results page?

---

Because the query does not separate them and the results page does not either. Coursera ranks fourth on “seo consulting” with a career explainer written for people deciding whether to become a consultant (Ahrefs, 2026) — a course marketplace holding a top-five position on a term with obvious commercial intent.

Google's People Also Ask block makes the split explicit, carrying “What is SEO job salary?” directly beside “How much does an SEO consultant cost?” (Ahrefs, 2026). Two audiences, one query, and the engine hedges across both because it cannot tell which one is asking.

One warning before anyone reads more into that results page than it can carry. This SERP is geo-anchored: the same pull returned a local pack at position 5 and a Wichita, Kansas directory URL at position 6, so a United States fetch resolved to one city (Ahrefs, 2026). Only the non-local results are safe to characterise.

Within that limit the pattern holds. Bruce Clay sits at position 2 with roughly 19,642 monthly organic visits on 34 referring domains, Coursera at position 4 with 9,823 visits on 79, and a page with zero referring domains ranks ninth (Ahrefs, 2026). Each picks a side, and nobody serves both.

What does the work pay, and why should a buyer care?

---

Because the salary is the buyer's real alternative and, unlike a consulting rate card, it is published. Across 29 public senior SEO and AI-search job postings collected June to August 2026 the median advertised band was $176K–$221K, with 16 of the 29 disclosing an annual figure (Indexable, 2026). The scope attached to those bands is the more useful half of the corpus.

Read the second and third rows together. More than half the postings want at least eight specialisms covered, and a majority of those name no team (Indexable, 2026). They describe a department compressed into one line item, which is exactly why consulting still sells: few individual hires cover technical SEO, content, analytics and engineering at enterprise scale, and 11 of the 29 asked for that combination in one person (Indexable, 2026).

Two limits. The sample is self-selected — public postings, mixed geographies and seniorities — so it shows what employers ask for, not what the market pays. And pay against scope here is not a benchmark.

Why did the deck stop being the deliverable?

---

Because the thing a deck measured stopped describing the outcome. Rankings are real and worth improving. They stopped predicting whether anyone reads the page, and they never predicted whether a generated answer would quote it.

Our own numbers carry the point. Over 90 days the head term “enterprise seo consultant” put one of our pages at position 91.3 across 468 impressions and returned zero clicks, while full-sentence buyer questions on the same property sat between positions 1.9 and 5.3 (Indexable, 2026). Demand arrived in a shape the keyword plan had no row for.

Ranking well did not rescue it. In a separate 28-day scan at least 543 page-one queries returned zero clicks against 7,038 impressions, and the only eight queries earning a click were brand or navigational (Indexable, 2026). That sits alongside the market figure: 68.01% of US searches ended without a click between January and April 2026 (SparkToro, 2026).

A consultant who presents a ranking improvement and stops has described an input. Whether it produced a visit or a citation is a separate measurement, and it is the one the client gets asked about three months later.

What instrumentation should an engagement leave behind?

---

Four artefacts, none of them a document. Each is something the client's own team runs every month once the consultant has gone, and each answers a question a ranking report cannot.

A retrieval measurement beside the ranking report. Join your top pages by impressions to the URLs AI engines cite in your category. Anything in the top 10 with no citation belongs on the gap list. Ours came back at 15 of 18 pages (Indexable, 2026).

A reporting split that survives the handover. Impressions and clicks measure different things now, so they need two lines, two owners and two explanations. Implement the split before the final readout or it will not outlive the engagement.

A review window matched to citation lag. Profound's 2026 analysis of ChatGPT and Claude agent logs across roughly 900 pages put the median citation lag at 6.81 days and the 90th percentile at 37.10 days (Profound, 2026). Schedule the review at 45 days.

A named list of the domains cited instead of you. In our category the most-cited domains were semrush.com with 85 citing responses, youtube.com with 83, ahrefs.com with 62, indexableai.com with 52 and reddit.com with 52 (Indexable, 2026). Two of the top five are platforms nobody publishes to on your behalf — a channel decision, not a content one. Underneath all four, JSON-LD carries a measured citation lift of 6.5 percentage points (AirOps, 2026).

What decision does the executive now own?

---

Three, and none is which firm to hire. Each belongs to whoever signs the invoice rather than whoever runs the channel, because each sets the terms the work will be judged on: what you are buying, whether the internal alternative exists, and how long you wait before calling the result.

Advice or apparatus. A recommendation is consumed once; a measurement runs every month afterwards. Price the two differently and put the second in the statement of work as a named deliverable.

Whether the internal alternative is real. The median advertised band across 29 public senior postings was $176K–$221K, and 10 of the 17 postings demanding eight or more specialisms named no team (Indexable, 2026). If your own job description resembles those, the hire you are comparing against may not exist at that price.

How long you wait. With a 90th-percentile citation lag of 37.10 days (Profound, 2026), a 30-day verdict on content-led work is a coin toss dressed as a decision. Agree the window before kickoff, and agree in writing that a click decline alongside a citation rise is acceptable.

What are the anti-patterns?

---

Five recur, each with a correction you can apply inside one procurement cycle. None is exotic: they are the default behaviours of a market whose reporting conventions were set when ranking and reading were the same event, and every one is visible in a statement of work before the engagement starts.

Buying a document. An audit unaccompanied by a repeatable measurement expires the first time the estate changes.

Reporting rankings as the outcome. Our page at 91.3 on the head term and our pages at 1.9 to 5.3 on the buyer questions tell opposite stories about one programme (Indexable, 2026).

Judging the engagement at 30 days. The 90th-percentile citation lag is 37.10 days (Profound, 2026), so a month produces an impression rather than a verdict.

Treating a results page as a national picture. The pull behind this article returned a local pack and a Wichita, Kansas URL on a United States fetch (Ahrefs, 2026). If a consultant shows you a top ten without saying where it resolved, ask.

Assuming authority decides the term. A domain with zero referring domains ranks ninth here (Ahrefs, 2026). At a keyword difficulty of 10 the binding constraint is the answer, not the links.

How do you brief and evaluate an SEO consultant?

---

Six steps. The first two happen before you send a brief; the remaining four belong in the contract rather than a kickoff conversation, because each one constrains how the engagement gets judged. None requires a tool you do not already own.

1. Step 1 — write down what you will still be able to measure in month seven. If the honest answer is “the deck”, change the brief. You should be buying a measurement your team can run unaided.
2. Step 2 — check your own head term the way we checked ours. Pull the live top ten and confirm no local pack is present before treating it as national. Ours resolved to Wichita, Kansas (Ahrefs, 2026).
3. Step 3 — ask for the retrieval gap list as a named deliverable. Top-10 pages with no AI citation, by URL. Ours came back at 15 of 18 (Indexable, 2026).
4. Step 4 — require the reporting split in the contract. Impressions and clicks on separate lines, with a written rule that a click decline alongside a citation rise is acceptable.
5. Step 5 — schedule the review at 45 days before kickoff. Apply the 37.10-day 90th-percentile citation lag as the stated reason (Profound, 2026).
6. Step 6 — price against the real alternative. Next, set the fee beside the median band of $176K–$221K across 29 postings (Indexable, 2026).

If you only do one, do step 3. The gap list costs an afternoon and it shows whether a ranking programme is producing retrieval.

In summary

---

Enterprise SEO consulting is not in trouble; its deliverable is. The market is large and cheaply entered — 5,700 US searches a month at a keyword difficulty of 10, with traffic potential of 20,000 (Ahrefs, 2026) — and served badly, because every page competing for it answers one of the two questions being asked.

Serve one audience and you lose half the demand; serve both and you need numbers on the career side, which is where 29 public postings and a median band of $176K–$221K earn their place (Indexable, 2026).

What a buyer should insist on is apparatus rather than advice: a retrieval measurement beside the ranking report, a reporting split with two owners, and a 45-day window matched to a 37.10-day citation lag (Profound, 2026). On our estate 15 of 18 top-10 pages had never been cited by an AI engine (Indexable, 2026) — a case study, not a law.

Start with step 3. Ask whoever you are considering for the gap list, and see whether they can produce it.