B2B Services AEO: Why Consulting Firms, Agencies, and Law Firms Are Disappearing From AI Search

When a CFO asks ChatGPT \"who should I hire for a digital transformation project,\" the same seven firms appear in 91% of responses. If you are a $20 million services firm with twelve years of operating history, partner-led delivery, and a Net Promoter Score that your sales team likes to mention, the bad news is that you are almost certainly not one of those seven.

The worse news is that the AI search results your prospective clients are looking at — the ones shaping the shortlist before any RFP goes out — are increasingly the only results they look at.

B2B services firms have spent the last fifteen years building inbound marketing programs against a Google SERP that listed ten blue links. That game is ending. The AI search interface that is replacing it is not just a different ranking algorithm — it is a different referral economy with a different set of winners. And the firms that won the old game by being slightly better at title tags and slightly more aggressive at gated content are not, by default, the firms winning the new one.

This is the most under-discussed shift in B2B services strategy in 2026. The accountancy practice in Birmingham, the digital agency in Brooklyn, the IP litigation boutique in San Francisco — all of them have AI search visibility that is structurally near zero, and most of them do not yet know it.

How Services Firms Lost AEO Before They Knew It Was Happening

The story of B2B services and AI search starts with a misread of what AI assistants actually do when asked a buyer-intent question.

When a senior buyer asks ChatGPT, Perplexity, or Claude something like \"who should I consider for a finance transformation in a $500M industrial business,\" the assistant is not running a directory search. It is composing a synthesis of every piece of relevant content it has indexed, weighted by authority signals, recency, and citation density. That synthesis produces a small set of named firms — usually three to seven — that appear consistently across rephrasings of the same question.

The mechanism that determines which firms appear in that small set is the same mechanism Signal has covered in detail across the ChatGPT citation engineering playbook and the share-of-model measurement framework. It rewards openly-readable content, named-author bylines, consistent entity metadata, and citation backlinks from other authoritative sources.

It does not reward, and largely cannot see, the things mid-market services firms have spent two decades investing in: gated case studies behind email forms, sales enablement decks shared only over a Calendly call, partner LinkedIn posts that disappear into the algorithmic ether after 72 hours, and capability presentations that exist only as PDFs on the firm's server. These assets generate revenue. They do not generate citations.

The result is that the services category has a citation-surface-area problem that took two decades of marketing-team incentives to create. Marketing leaders at services firms were rewarded for pipeline contribution. Pipeline contribution favored gated content. Gated content is invisible to AI crawlers. AI crawlers are now the discovery layer. The funnel ate itself.

There is a second, less obvious problem. The training data that powers most production AI assistants in 2026 is heavily weighted toward a few high-volume sources — large publishers, Wikipedia, Reddit, GitHub, and the canonical content libraries of a handful of established institutions. Signal has written before about how every major LLM cites Reddit, and the equivalent dynamic in B2B services is that the LLMs disproportionately cite the same handful of consulting libraries because those libraries dominated the open business-content web at the moment the training corpora were assembled. A mid-market firm that started publishing thought leadership in 2023 is not just behind on volume — it is behind on training-data inclusion, and that gap does not close until the next major training-data refresh cycle picks up the new content.

The Two-Tier Citation Problem

If you actually log the firms cited across a representative basket of B2B services queries — \"best management consulting firm for retail strategy,\" \"top boutique law firms for fintech regulation,\" \"leading agencies for B2B brand positioning,\" and so on — a striking pattern emerges. AI assistants cite from two tiers, with almost nothing in between.

Tier one: the incumbent firms. McKinsey, BCG, Bain. Deloitte, PwC, KPMG, EY. Accenture. For law, the magic circle and white-shoe firms. For agencies, Ogilvy, WPP, Publicis. These names appear with a frequency wildly disproportionate to their actual market share of relevant projects. They appear because the content footprint they have built — McKinsey Quarterly going back to 1964, BCG Perspectives, Bain Insights, Deloitte Insights, the Big Four research libraries — collectively comprises hundreds of thousands of indexed, openly accessible, named-author pages that AI training pipelines ingested liberally before anyone in the services industry had heard the acronym AEO.

Tier two: the publications and intermediaries. Harvard Business Review. McKinsey Quarterly (yes, again, as a publication signal as well as a firm signal). The Drum and AdAge for agencies. Law.com and The Lawyer for law firms. Industry analyst houses like Forrester, Gartner, and IDC. These outlets show up in AI search results as authoritative aggregators — \"according to a recent Forrester report\" or \"a 2025 HBR analysis of digital transformation suggests\" — and they often determine which firms get mentioned as supporting examples.

What is missing from these tiers is a coherent third group of mid-market firms. The $20M boutique that delivers better work for clients in its niche than any Big Four firm could. The 60-person regional law firm with deeper sector expertise than the magic circle in its specialty. The independent agency producing better creative than the holding company on any reasonable measure of quality. These firms exist, they win pitches every week, and they are invisible to ChatGPT.

The citation rate for mid-market firms is not a function of how good their work is. It is a function of how much of their work is published in a form an AI assistant can read and cite.

Why Most Agency and Consultancy Websites Are AEO Disasters

If you spend an hour systematically reviewing the websites of, say, twenty mid-market digital agencies or twenty boutique management consultancies, you will see the same five structural problems repeat to the point of absurdity.

1. The case studies are behind email gates, or behind a \"Get in touch\" form, or simply do not exist as readable pages. The single richest content asset a services firm owns — the case study with a named client, a measurable outcome, and a methodology narrative — is the asset most consistently hidden from the open web. AI crawlers cannot complete a form. They cannot \"download the PDF.\" Content that requires interaction to access is content that does not exist for AEO purposes.

2. The thought leadership is anonymous. Most agency and consultancy blog posts are bylined by \"The Firm Team\" or \"Marketing\" or no one at all. Some are signed by the firm's name as if the firm itself were a person. This is the single most damaging signal a content program can send to an AI assistant trying to build an entity graph. AI search rewards named authors with verifiable expertise; firm-bylined content reads as low-confidence aggregate noise.

3. The website copy is generic to the point of opacity. \"We are a strategic partner that delivers transformative outcomes through human-centric methodologies that put clients at the heart of everything we do.\" An AI assistant trying to figure out what your firm actually does cannot infer it from that sentence. It will move on to the next firm whose homepage clearly states \"We help mid-market industrial businesses execute post-merger IT integrations within 180 days.\" Specificity is now an AEO requirement, not a copywriting nicety.

4. Schema markup is either absent or wildly incomplete. As Signal has covered in detail, schema markup is shifting from a search-ranking signal to an entity-context signal for AI search. Most mid-market services firms have either zero schema beyond basic Organization markup, or a sprawl of generic markup that does not actually expose the firm's services, partners, or expertise as discrete entities. The firms with Person schema on every partner bio, Service schema on every offering page, and Article schema on every published piece are the firms that are showing up in AI citations even when they are smaller than their unschemafied competitors.

5. The partner bios are the worst pages on the site. A two-line bio listing the partner's degree and the year they joined the firm, with no published content, no LinkedIn link, no Person schema, no list of topics they speak on. This is a tragedy. The partner bio page should be the AEO crown jewel of a services firm — the entity-anchor that ties the firm's expertise to a verifiable individual the AI can model. Most are empty.

The cumulative effect is that the typical mid-market services firm website is something close to AEO-invisible — an interactive sales brochure designed for human readers who arrived via a referral, optimized for nothing the modern discovery layer can read.

The McKinsey / BCG / Bain Playbook (Won by Accident)

What is mildly infuriating, if you run a smaller firm, is that the dominant players in services AEO did not build their position on purpose. They built it because the brand strategy that worked in the pre-AI era — publish constantly, sign everything with a named partner, give it away free as a credibility marker, attract a flow of senior corporate readers — happened to produce exactly the content architecture that AI assistants would later reward.

McKinsey Quarterly has been publishing since 1964. McKinsey Insights now hosts roughly 30,000 indexed pages of research content, virtually all of it openly accessible, virtually all of it bylined by named partners with profile pages that themselves link to LinkedIn, university affiliations, and external press citations. The firm's Marvin internal AI platform is built on top of this corpus, and the AEO benefit is essentially a free byproduct of decades of brand investment.

BCG's pattern is similar in structure if smaller in scale. The BCG Henderson Institute, BCG Perspectives, and the firm's published frameworks — the Growth-Share Matrix, the Strategy Palette, the Adaptive Advantage construct — appear in AI search results both as direct citations and as conceptual references that other authors then cite back to BCG. Bain Insights, Deloitte Insights, EY's research library, KPMG's industry reports — same pattern, different volume.

The reason this matters for mid-market firms is not that you should attempt to replicate McKinsey's volume. You cannot, and trying to will produce thin, generic content the AI will ignore. The lesson is structural: the firms that won AI search in the services category did so by building a content architecture that AI assistants happen to read well — openly readable, named-author, topic-clustered, schema-supported, internally cross-linked, and externally cited.

You can copy the architecture without copying the volume. That is the actual playbook.

The a16z and Stripe Press Anomaly

The most instructive non-consultancy examples of services-adjacent AEO success are Andreessen Horowitz and Stripe Press. Neither is a services firm in the classical sense. Both have built citation footprints that are radically disproportionate to their headcount, and both did it the same way: by treating content as durable infrastructure rather than as marketing collateral.

Andreessen Horowitz. A16z's content footprint includes the Future blog, the a16z Podcast network, the Marc Andreessen and Ben Horowitz essay archives, the bio pages of every general partner and operating partner, and the firm's published frameworks for portfolio construction, AI strategy, and crypto market structure. The result is that when a founder asks ChatGPT about Series A norms, portfolio construction at venture funds, or how to evaluate an AI startup, a16z content appears with a frequency that exceeds firms with materially larger AUM and longer histories. The firm essentially executed a thought-leadership-as-citation strategy starting in 2009, before anyone was thinking about AEO, and the compounding returns are now showing up in AI search at a magnitude that older firms cannot easily catch.

Stripe Press. Stripe has built a brand-adjacent publishing house — Stripe Press — that publishes long-form books and essays on topics adjacent to the company's economic-infrastructure mission. The Stripe Press catalog includes works on cities, growth, internet history, and progress studies, none of which directly sell Stripe's API products. The AEO consequence is that Stripe shows up as an authoritative voice in conversations about economic growth, developer culture, and infrastructure design that have nothing to do with payments processing. The brand halo from this citation footprint flows back to the core product in ways that are difficult to quantify but unambiguously real.

The principle in both cases is the same: content that exists primarily as durable, openly-readable, named-author work generates an authority signal that survives the transition from SEO to AEO. Content that exists primarily as gated lead-magnets does not.

For a mid-market services firm asking what to actually do, the implication is direct. Stop thinking of thought leadership as marketing collateral with a pipeline-attribution model. Start thinking of it as citation infrastructure with a brand-discovery model. The two activities produce different content, with different authorship norms, in different distribution channels.

The clearest tell that a services firm has not made this mental shift is when the head of marketing describes the firm's thought leadership program in terms of MQLs generated per article. That is the wrong unit. Citation infrastructure is measured in entity authority, not in marketing-qualified leads. The MQL framing leads to short, generic, SEO-shaped articles aimed at top-of-funnel keyword capture. The citation-infrastructure framing leads to longer, more specific, more opinionated pieces signed by senior partners — exactly the content shape that AI assistants reward and that most mid-market firms refuse to commission because it does not look like an obvious pipeline asset on the marketing dashboard.

The Named-Author Moat

Of all the AEO levers available to a services firm, the named-author moat is the highest-leverage and the most underused.

AI assistants build internal entity models of who is authoritative on a given topic. Those models are reinforced when content is consistently signed by an identifiable person, when that person has a stable profile page with structured Person schema, when the person's LinkedIn, conference bios, university affiliations, and external press appearances all reference the same name and topic cluster, and when other authoritative sources cite the person directly.

This is how McKinsey ended up with a long tail of partners who are themselves AI-search citations — Kevin Sneader, Vik Malhotra, Dominic Barton, and so on through the org chart. Each named partner is an entity in the AI's internal model, with a verified affiliation back to the firm. The compounding effect is that McKinsey content does not have to compete on its own merit — it is reinforced by the entity authority of the named partner who signed it, who is in turn reinforced by every other piece of content the same partner has signed.

A mid-market firm with twelve partners, all of whom currently publish under the firm's generic byline, can transform its AEO posture in roughly six months by doing five things:

1. Assign every substantive piece of content to a named partner. Not as a co-author. Not as the firm. As the principal author, with their photo, bio, and Person schema on the article page.

1. Build a real partner bio page for each individual. Not the two-line standard. A 600-800 word page covering background, expertise areas, representative engagements, published articles, speaking history, and external press mentions. Structured with Person schema and sameAs links to LinkedIn, the firm's about page, and any external profiles.

1. Cross-link aggressively. Every article the partner publishes should link back to their bio page. The bio page should link to all their articles. Their LinkedIn should link to the bio page. Their conference speaker bios should link to the bio page. Coherent entity signals require coherent linking.

1. Get them cited externally. Press mentions, podcast guest appearances, trade publication quotes, university lecture pages — each one is an external entity-reinforcement signal. Most mid-market partners are reluctant to do PR; the AEO returns are now high enough that the math has changed.

1. Measure the citation lift. Track each named partner's appearance in AI search results across a defined query set for their expertise topic. This is the single most useful health metric for a B2B services AEO program.

The named-author moat is partially defensible because it cannot be replicated by content velocity alone. A competitor cannot publish their way around your firm's senior partner being the recognized AI-search authority on, say, transfer pricing for cross-border SaaS businesses. They have to either build their own named expert in the same topic or concede the territory.

A practical objection arises here. Partners do not want to write. Partners do not have time to write. Partners feel uncomfortable being individually positioned because the firm's culture prefers the collective brand. These are all true, and they are also the same objections every consulting firm raised about LinkedIn in 2014, podcasts in 2018, and substack in 2022. The firms that pushed through the cultural discomfort each time ended up with disproportionate brand authority in the next channel. AEO is the 2026 version of the same dynamic. The firms that get their partners published under their own names, in 2026, will look prescient in 2028. The firms that wait for partner buy-in to be unanimous will not.

The operational pattern that works at most firms is to start with two or three willing partners — there is always at least one in any firm who quietly wants the personal brand — and use their early citation results as internal evidence to bring the more reluctant partners along. Once the firm can show in a partner meeting that Partner A is being cited by ChatGPT in 38% of queries on their topic while Partner B (who refuses to publish) is being cited in 0%, the political conversation changes.

Restructuring Case Studies for AEO

Case studies are the most squandered asset in B2B services marketing. They contain everything an AI assistant needs to confirm a firm's expertise — a named client, a defined problem, a specific methodology, a measurable outcome — and they are almost universally hidden behind gates that make them invisible to the discovery layer.

The restructuring move is straightforward, if politically uncomfortable inside many firms.

Open the case study. Publish the full case study as an indexable page on the firm's site. Yes, the client may need to be anonymized in some cases — \"a $400M industrial distributor\" instead of the actual company name — but the more the case study can be specific, the more useful it is to the AI. If you have explicit permission to name the client, name them. If you do not, get permission for at least 30% of your case studies before this becomes industry standard practice.

Structure with Article schema and CaseStudy markup. Author, datePublished, dateModified, mainEntityOfPage, and the specific service rendered. Connect the case study back to the relevant Service page and the partner who led the work.

Name the named delivery partner. The case study should be bylined by the lead partner on the engagement, with their Person schema and bio link. This ties the case study to the firm's named-author entity graph.

Include extractable facts. \"We reduced the client's procurement cycle time from 47 days to 18 days within six months\" is an extractable fact that an AI assistant can quote. \"We delivered transformative procurement outcomes\" is not. The case study should contain three to seven specific, quotable facts in declarative sentences near the top of the article.

Layer the gated artifact on top. The full case study is openly readable. The richer artifact — the board-ready deck, the methodology appendix, the financial model — can still sit behind a form. The first version powers AI citation. The second version captures intent. The two are not mutually exclusive; the historical mistake was treating them as the same artifact and putting both behind the gate.

A firm with twenty case studies, currently all gated, that ungates ten of them with proper schema, named clients (or specific descriptors), measurable outcomes, and partner bylines will see meaningful citation lift in AI search within 60-90 days. This is one of the few AEO moves with a tractable, near-term return.

The Four Metrics Services Firms Should Track

Most services firms currently measure marketing using a stack designed for the SEO era — sessions, qualified pipeline, content-attributed pipeline, MQL-to-SQL conversion. None of these metrics measure AEO performance, because AEO performance does not produce sessions until much later in the funnel and may produce conversion influence without ever producing a session at all.

The 2026 AEO measurement stack for a services firm centers on four metrics.

1. Citation rate by service line. For each of the firm's two to four primary service offerings, define a representative basket of 50-100 queries a buyer might ask an AI assistant. Sample each query monthly across ChatGPT, Perplexity, Gemini, and Claude. Measure the percentage of queries where the firm is cited. Track the trend. Tools like Profound, Bluefish, and SerpRecon now provide this data programmatically; a small internal scraper using each AI's API can produce equivalent data for under a few hundred dollars a month.

2. Share of citation versus named competitors. Pick three to five direct competitors — the firms that show up most often in your pitch processes. For your target query basket, measure your firm's citation rate versus each competitor's. This is the most important strategic metric, because absolute citation rate is meaningless without competitive context.

3. Named-author citation rate. For each of your senior partners, measure how often they personally are cited in AI search results — either by name or by attribution back to their published work — across the query set relevant to their expertise. This is the metric that tells you whether your named-author moat is real or theoretical.

4. AI-referral conversion influence. When prospects do eventually land on your site or open a sales conversation, capture whether AI search played a role in their discovery journey. The simplest mechanism is a single optional field in your inbound contact form — \"How did you hear about us?\" with options including ChatGPT, Perplexity, Gemini, and Claude. The data will be noisy. It will still be useful.

These four metrics will not align cleanly with the legacy marketing dashboard. The CFO will resist them. The right response is to run both stacks in parallel through 2026 — the legacy SEO/pipeline metrics for continuity, the AEO metrics for forward direction — and renegotiate the dashboard structure for 2027 budget discussions.

The Expertise Schema Play

The technical AEO foundation for a services firm is a coordinated schema implementation that exposes the firm, its partners, its services, and its published content as a connected entity graph.

A serviceable minimum implementation includes five schema layers:

Organization schema on the homepage. Legal name, founding date, address, sameAs links to LinkedIn, Crunchbase, and any registry pages. The areaServed array listing markets the firm operates in. The knowsAbout array listing the firm's primary topic areas.

Person schema on every partner and senior practitioner bio. worksFor, jobTitle, alumniOf, knowsAbout, sameAs links to LinkedIn and any external bios. Image. Description in the partner's own voice.

Service schema on each service-line page. serviceType, provider linked back to the Organization, areaServed, audience, and a hasOfferCatalog if the firm publishes specific packaged offerings.

Article schema on every thought-leadership piece. author linked to the Person schema for the partner, datePublished, dateModified, mainEntityOfPage, and image. Citation properties where applicable.

FAQPage schema on key service pages. Six to ten questions a buyer might ask an AI assistant about that service, answered concisely on the page itself with FAQ schema markup. This is the highest-leverage piece of schema for AEO because it directly mirrors the format AI assistants extract.

Schema implementation is not glamorous work. It is also not optional. The firms that have implemented this stack are showing up in AI search citations at meaningfully higher rates than equivalent firms that have not, even controlling for content volume and brand recognition. As Signal has covered, the entity-context layer is the new currency of AI search, and schema is how you participate in that economy.

The Action Checklist for a $20M Services Firm

If you run marketing or operations for a $20M services firm and you have read this far, you probably want a concrete plan rather than another framework. Here is the one I would actually give you.

Quarter 1. Audit your current citation footprint. Run a baseline measurement across 100 representative queries on ChatGPT, Perplexity, Claude, and Gemini. Identify your top three direct competitors and measure their citation rates. Audit your existing content library: how many pieces are openly readable? How many are bylined by a named author? How many have Article schema? Build the dashboard you will use for the next two years.

Quarter 2. Restructure case studies. Identify the ten case studies with the strongest measurable outcomes. Get client permission to publish openly where possible. Republish with full Article schema, named delivery partner byline, named client (or specific descriptor), and three to seven extractable facts in the opening section. Implement Organization and Person schema across the site.

Quarter 3. Build the named-author program. For each partner you want to position as an AI-search-cited expert, define their topic territory (one or two specific areas), build a proper bio page with Person schema, and commit to a publishing cadence of one substantive article every six weeks. Begin external PR placement for these partners — podcast guest appearances, trade publication contributions, conference speaking. Implement FAQ schema on top service pages.

Quarter 4. Measure, iterate, and double down on the topic territories where citation rates are climbing. Decommission gated content that is generating little pipeline and is not eligible for AEO use. Publish llms.txt and llms-full.txt files exposing your full content corpus to AI crawlers. Begin the 2027 planning cycle with AEO as a named workstream with its own budget line — the days of running it out of the SEO budget are over.

This is roughly 18 months of work for a marketing team of three to five people, layered on top of existing inbound and outbound activity. It is achievable inside a $1-2M marketing budget. It will not produce immediate pipeline lift in Q1 or Q2. It will, by the end of 2026, change the firm's position in AI search materially — which is the only marketing investment that compounds against the structural shift the entire B2B services category is now navigating.

Takeaway: B2B services firms in the $5M-$100M revenue band are losing AI search visibility not because their work is worse than the Big Four's, but because their content architecture is structurally invisible to the discovery layer that is replacing Google. The fix is not more content — it is different content, structured differently, signed differently, and exposed differently. Open the case studies. Name the authors. Build the entity graph. Measure citation rate, not sessions. The firms that execute this transition in 2026 will define the mid-market services landscape for the next decade. The ones still optimizing for pipeline-attributed gated content will spend 2027 wondering why fewer prospects know they exist.