The Pre-Intake Problem

I.The shift nobody named.

For roughly the last twenty years, the path from "I need a lawyer" to "I'm calling this lawyer" looked the same. A person would have a problem — a car wreck, a slip and fall, a wrongful termination. They would open Google. They would type some version of "best [practice area] lawyer in [their city]." Google would return a ranked list of blue links and a local map pack. The person would click two or three of those links, look at the firms behind them, and pick up the phone.

The marketing industry built itself around that path. Search engine optimization existed to dominate the blue links. Pay-per-click existed to dominate the slots above and around them. Local SEO existed to dominate the map pack. Intake teams existed to convert the call into a signed case. Every metric in legal marketing — keyword ranking, cost per click, click-through rate, conversion rate, signed-case rate — assumed that the journey began at the Google search bar.

Beginning sometime in late 2023, that assumption stopped being correct. Consumers started routing the early part of that journey through ChatGPT, then Perplexity, then Gemini, and most recently through Google AI Overviews — the AI-generated summary that now appears above the traditional search results on roughly half of all relevant queries.¹ By early 2026 the shift is mainstream enough that you can ask any room of consumers under forty and find people who have not opened a normal Google search results page in months.

What changed is not where consumers end up. They still call lawyers. They still sign retainer agreements. They still walk into intake meetings. What changed is the part of the journey that happens before the call — the stage between recognizing the problem and reaching for the phone. That stage used to take twenty minutes and pass through Google. It now takes ten seconds and passes through an AI engine that returns three firm names.

That stage is what this essay is about. It does not have an industry-standard name yet, partly because the marketing world has not caught up to its existence. I have been calling it pre-intake, and I am going to keep calling it that here. It is the most important and least-measured stage in legal client acquisition in 2026.

II.What pre-intake means.

The clearest way to define pre-intake is by drawing a clean line through the consumer journey and labeling the stages.

Stage one is recognition. Something happens to the consumer — an accident, a death in the family, a wage dispute, a divorce decision. They form the thought I need a lawyer. Recognition happens whenever it happens; marketing cannot move it, and most firms do not try.

Stage two is pre-intake. The consumer has decided they need a lawyer but has not chosen which one. They begin investigating. Twenty years ago this stage took them through asking friends, opening a phone book, or driving past billboards. Ten years ago it took them through Google search. Today it increasingly takes them through an AI engine.

Stage three is intake. The consumer has chosen — usually from a short list of three to five names — which firm to contact first. They call, fill out a form, or open a chat. The firm's intake process now begins.

Stage four is engagement. The consumer signs a retainer. The case is opened.

Pre-intake — stage two — is where the shortlist gets built. In the Google era, the shortlist was constructed by the consumer themselves out of the search results they read. In the AI era, the shortlist is constructed for them by a model that names three to five firms in the body of its answer.

That single observation contains most of the implication. If the shortlist is built during pre-intake, and pre-intake is now adjudicated by AI engines, then the firms named by AI engines win the case before intake even begins. The firms not named by AI engines are competing against an empty list — they cannot be chosen because they were never on it.

III.Why pre-intake is invisible to firms.

If pre-intake is so important, why are most law firms not actively competing in it?

The honest answer is that pre-intake is invisible to the dashboards that firms look at. Every traditional intake metric begins counting at the moment the phone rings or the form is submitted. Conversion rate, cost per signed case, attorney response time, lead quality score — all of them measure what happens after the consumer has chosen the firm. None of them can see what happened beforehand.

To put it concretely: a firm running Google Ads can see its click-through rate and its cost per lead. It can see exactly how often consumers clicked the ad and how often the click became a phone call. What it cannot see — and what its agency cannot see, and what its analytics platform cannot see — is the larger universe of consumers who never clicked the ad because they never saw it, because the consumer asked ChatGPT instead, and ChatGPT named three firms that did not include this one.

The missing leads do not appear in the dashboard as failures. They appear as nothing at all. A firm with a 35% intake conversion rate and a steady caseload looks, by every metric available, to be performing well. The fact that its addressable market has quietly shrunk by 20 or 30 or 40 percent — because that share of the market is now being routed to a different firm by an AI engine — does not show up anywhere on the report.

The result is a class of firms that look healthy by their own metrics and are losing market share by metrics no one has built. This is the most dangerous category, because they have no internal alarm telling them to act.

A firm with a 35% intake conversion rate and a steady caseload looks, by every metric available, to be performing well. The fact that its addressable market has quietly shrunk by 30 percent does not show up anywhere on the report.

IV.Who's winning pre-intake right now.

It is tempting to assume that the firms winning pre-intake are the largest, most expensive, most heavily-advertised firms in their market. They are not. Volume of advertising spend is essentially uncorrelated with appearance in AI engine answers, because AI engines do not pull from the paid auction. They pull from a completely different set of signals.

The firms that consistently appear when consumers ask ChatGPT or Perplexity for a lawyer in their city share, in our experience auditing roughly 200 PI firms during 2025 and the first half of 2026, three properties.

First, they have citation density across third-party authority sources. The firm name appears in association rosters, state bar profiles, regional press, legal directory listings with substantive content, and the kind of practice-vertical publications that AI engines treat as reliable. Citation density is not the same as link volume; ten high-authority citations beat ten thousand low-authority directory listings. Authority of the source matters more than volume.

Second, they have structured data on their own website — clean Attorney, LegalService, Review, Organization, and FAQPage schema. AI engines parse schema directly when retrieving content at query time. A firm whose website emits schema reads to the model as a recognizable entity with discoverable attributes. A firm whose website emits no schema reads as undifferentiated marketing text. The difference is not small.

Third, they have practice-area-by-jurisdiction content depth. Not "we handle car accidents," but "uninsured motorist claim process in Florida," "rideshare passenger injury liability Texas," "delayed-onset whiplash statute of limitations Georgia." Narrow pages that answer narrow questions earn semantic weight that generic practice pages do not. When the consumer's actual question is narrow, the model retrieves the firm whose pages answer that exact question.

None of those three properties require enormous budgets. They require deliberate engineering. The largest firm in a metro can score zero on all three; a four-attorney boutique can score high on all three. The competitive surface is genuinely different from the surface advertising spend competes on.

V.What pre-intake looks like inside the AI engines.

To make this concrete, here is what a real consumer query produces in a real engine, today.

The prompt is the kind a person would actually type: "best personal injury lawyer in Tampa Florida." Submitted to ChatGPT in May 2026, the response is roughly two hundred words of prose. The prose names four specific firms by name, gives a one-line description of each, and closes with a generic recommendation to "consult several attorneys before making a decision." A reasonable consumer reading that answer will pick one of the four named firms and call.

The same prompt submitted to Perplexity produces a similar shape with cited sources. The same prompt submitted to Gemini produces a similar shape weighted more heavily toward local Google Business Profile data and reviews. The same prompt as a Google search now also produces an AI Overview at the top — a summary citing two to three firms, displayed above the traditional blue links. On a substantial share of queries, the consumer never scrolls past the AI Overview.²

Across the four engines combined, that means a typical PI consumer in a typical metro is being shown three to five firm names before they have done any of their own research. If your firm is not in those names, you are not in the consideration set. The mechanics of the funnel make this true regardless of how well your billboard is performing.

Most firms have never actually checked what AI returns for their own market. The exercise is uncomfortable. The firms named tend not to be the ones the principal of the firm assumes will be named. The largest firm in a metro is frequently absent. A smaller competitor that has engineered for AI citation is frequently present. The exercise is also revealing: it surfaces, for the first time, what the AI-mediated shortlist actually looks like — which is the shortlist that matters now.

If you want to see it for your own firm, the diagnostic at isyourfirmaiinvisible.com runs an abbreviated version of this test across all four engines in about sixty seconds. It will not tell you everything, but it will tell you whether your firm appears, and that is the threshold question.

VI.The economic case.

The pre-intake question becomes uncomfortable when you do the economics.

A firm that appears in AI answers receives a call from a consumer who has already chosen, in their mind, to consider that firm seriously. The model named the firm. The consumer treats that as an implicit recommendation. The intake call begins with what is effectively warm consideration; the firm did not pay for that consumer's attention.

A firm that does not appear in AI answers is competing for the same consumer further down the funnel, after the AI has handed them a shortlist that excludes the firm. The path to reaching that consumer is now paid: the firm has to interrupt the consumer with an ad, with a billboard, with a TV spot, with a sponsorship — channels whose unit economics have been getting worse for a decade. The current going rate for a personal injury Google Ads click in a competitive metro is in the $300 to $600 range, depending on practice area and time of day.³ Roughly one in seven clicks becomes a signed case in well-run accounts. That math gets you somewhere between $2,100 and $4,200 in paid acquisition cost per signed case, before the firm's intake labor, before contingency outlay, before anything else.

A firm winning pre-intake gets a meaningful share of those same cases for the marginal cost of having engineered for AI citation in the first place. The engineering work has a one-time cost; the citation density it produces compounds for years. The asymmetry between the two business models is significant and is widening every quarter as AI search share grows.

Put differently: in the AI-mediated funnel, citation surface is now an asset that produces ongoing return. Advertising spend is an expense that produces no asset. A firm that converts its budget from one to the other is, in essence, switching from renting consideration to owning it.

VII.Why most firms aren't fixing this.

Given the math, the obvious question is why firms have not, in large numbers, started doing the work to win pre-intake. Three reasons recur.

The first is that pre-intake has not been named, in most firms' internal vocabulary, as a stage that exists. The intake director cannot improve a stage that does not appear on her org chart. The marketing director cannot budget for a stage that does not appear in his line items. Until firms start saying pre-intake out loud, internally, the stage continues to be invisible operationally even when it is visible economically.

The second is that firms assume the work is part of SEO and that their existing agency handles it. It is not, and they do not. The correlation between Google search ranking and AI visibility — across the dataset we have, drawn from roughly 200 PI firms — is approximately 0.076 percent. SEO and AEO are different algorithms with different inputs. A firm whose agency is doing strong SEO work and assumes the AI side is taken care of is mistaken in a way that takes about eighteen months to become obvious in the numbers.

The third is that firms are investing in the wrong stage — pouring resources into intake training, call response time, lead-scoring software, and CRM optimization, all of which compound performance after the consumer has already chosen the firm. None of that work moves the pre-intake shortlist. None of it solves the problem of not being named at all. It is possible to have the best intake conversion rate in the metro and still be losing market share, because the conversion rate is being computed on a smaller and smaller share of the available consumer pool.

The cumulative effect of those three patterns is that the firms most in danger of losing share to pre-intake competitors are also the firms least likely to recognize what is happening. They are not flying blind because they cannot see; they are flying blind because they have not been told there is anything to look at.

A firm whose agency is doing strong SEO work and assumes the AI side is taken care of is mistaken in a way that takes about eighteen months to become obvious in the numbers.

VIII.What to do.

The fix for the pre-intake problem is not a single intervention; it is a stack of related ones, run in order. There are five layers.

1. Acknowledge that pre-intake exists. Add the stage to the funnel diagram. Add it to the marketing budget line items. Add a single metric to the dashboard that measures it — engine-appearance rate across the prompts a real consumer in this metro would use. Without acknowledgment, none of the rest happens.
2. Measure where the firm currently scores. Run the diagnostic. Score the firm against ChatGPT, Perplexity, Gemini, and Google AI Overviews. Record the result. This is the baseline. The honest baseline for most firms is invisible or thin, which is uncomfortable but useful.
3. Audit the signal stack. Citation density on authority sources, structured data on the firm's website, practice-area-by-jurisdiction content depth, reputation substance, entity consistency. Each of these is a workstream. Most firms have zero on three or more of them.
4. Rebuild the signal stack on a deliberate cadence. Six to twelve weeks of focused work, in roughly that order. The early weeks are foundational (entity reconciliation, schema markup), the later weeks are compounding (citation building, content depth, reputation engineering).
5. Re-measure quarterly. Engine retrieval behaviors change. Competitors do work. The signal stack needs maintenance. Firms that build and then ignore drift back toward invisibility within two to three quarters.

None of those steps require a budget that a serious firm cannot allocate. The first two cost almost nothing. The third is consulting time. The fourth is the only real expense, and it produces a citation asset that pays back across years. The fifth is monitoring overhead.

The harder problem is not budget. The harder problem is that pre-intake work does not have the satisfying short-term feedback loop of paid acquisition. A Google Ads campaign produces leads this week. A pre-intake rebuild produces engine citations beginning four to twelve weeks out, with the largest gains six to nine months in. Firms accustomed to weekly performance reports often quit pre-intake work before it has compounded. That is its own problem, and it has caused real engagements to fail.

IX.The window.

The reason any of this matters with any urgency is that pre-intake is, for the moment, an uncontested competitive surface in most metros. Roughly five percent of personal injury firms appear in AI answers at all. The remaining ninety-five percent are invisible. The dominant firm in any given metro tends to be one firm — sometimes two — across all four engines, which means competing for that position currently means competing against at most two opponents, not the field of dozens a firm would face in a paid auction.

That is the position now. It does not last. The firms that have already moved are accumulating citation density that compounds. Their entity surface expands. Their structured data gets richer. Their topical depth grows. Each month a competitor stays invisible, the gap widens. By 2028 — and this is conservative — the dominant AI-visible firm in a top-fifty metro will have three to five times the citation surface area of the average invisible competitor in the same metro. Closing that gap retroactively will be possible but will cost meaningfully more than starting first.

There is a real first-mover advantage in citation work that does not exist in paid acquisition. Once a firm is established as a recognized entity in an AI engine's retrieval graph, displacing it requires both building a competing entity and waiting for the model to update its understanding. Both of those are slow processes. The leader board for AI visibility in personal injury, in any given metro, is being written in 2026 and 2027. It will be relatively stable thereafter.

The firms that recognize this and act now will spend the second half of the 2020s collecting the upside. The firms that do not recognize it, or recognize it but treat it as next quarter's problem, will be doing the same advertising work they are doing today, against a shrinking addressable market, and wondering why their unit economics keep eroding.

Pre-intake is the stage that matters. It does not have a name in most firms yet. The name does not matter. The work does.