Your peers spend hours optimising their ChatGPT prompts to save time on drafting. Meanwhile, your clients ask that same AI a question: "Which firm for my case?" And it doesn't answer with your name.
For a law firm to be cited by ChatGPT, Claude or Perplexity in response to a legal query, three technical conditions must be met: Cross-Platform Consistency across all your digital sources, Specialised Semantic Precision on your areas of expertise, and an External Authority Score consolidated by trusted third parties. Without these three foundations, your firm is invisible to generative AI — regardless of your reputation, your Google PageRank, or your marketing budget. Our 2026 KaiZen audit reveals a brutal reality: 82% of French law firms are invisible on Perplexity, and fewer than one in ten has correctly configured OpenAI's GPTBot crawler in its robots.txt. On the French market, a single firm simultaneously dominates "user" (individuals) and "market" (B2B) queries: Bredin Prat, with a KaiZen Visibility Score of 100, versus 64 for Gide Loyrette Nouel.
The reality the profession still ignores
According to the latest March 2026 data, 27% of Google searches now end without a click — a client asking a question gets the answer directly in Google's AI Overview or in ChatGPT, without ever visiting a lawyer's website. AI-engine traffic grew 225% in a year. And, critically for the legal profession: Claude (Anthropic) is now deployed in 32% of companies worldwide, a majority of them in legal departments.
The consequence is mechanical. When a general counsel asks Claude "which firm for my M&A dispute in France?", the AI doesn't consult Google. It draws on its training corpus and its structured real-time sources. If your firm isn't in those sources, you don't exist in the answer. And the prospect will never know you could have handled their case.
More counter-intuitive still: traffic from ChatGPT converts at 15.9% versus 1.76% for Google (Seer Interactive, 2025). In other words, 100 visitors from an AI citation produce as many client engagements as 900 Google visitors. A firm cited by ChatGPT captures purchase intent 9 times higher than a firm ranked well in traditional search.
ChatGPT traffic converts 9 times better than Google traffic for premium B2B services. In January 2026 alone, ChatGPT generated 87.4% of all global "AI referrals" — a near-monopoly on algorithmic provider recommendations.
Our estimate, from field observations and LLM growth data, suggests that roughly one legal search in four now goes through a generative AI rather than Google. That ratio is set to grow mechanically as AI-native generations reach decision-making roles in companies.
Why classic SEO no longer protects you
If you're reading this, you've probably already invested in SEO, or perhaps in directories like Doctrine, Lexbase or Le Monde du Droit. That strategy proved itself for fifteen years. But it rests on an assumption that has become fragile: that your clients still use Google as their entry point. That's less and less true — here's why.
1. Legal directories are not priority sources for LLMs
ChatGPT and Claude don't scan Doctrine or Lexbase in real time. Their training sources are dominated by Wikipedia, institutional sites (.gov, professional bodies), LinkedIn, and company pages structured in JSON-LD. An exhaustive Doctrine profile ranks you on Doctrine — not in ChatGPT.
2. Traditional SEO and GEO follow opposite logics
Google rewards accumulated authority — backlinks, seniority, long content, trust signals. LLMs reward semantic structure: sourced figures, explicit FAQs, Schema.org markup, direct and factual phrasing. A firm can be number 1 on Google without ever being cited by an AI, and vice versa.
3. Your AI crawlers may not even be allowed to read you
A critical technical detail nobody in the profession mentions: for OpenAI to include your firm in ChatGPT's reference corpus, its GPTBot crawler must be explicitly allowed in your robots.txt. Same for ClaudeBot (Anthropic) and PerplexityBot. Our 2026 KaiZen audit shows that fewer than one French law firm in ten has done this. Even firms investing heavily in SEO often block, by default, the very robot that would open OpenAI's door — the digital equivalent of closing your office to clients without knowing it.
4. Professional ethics aren't the obstacle you think
Many lawyers give up on digital visibility out of ethical over-caution. Yet France's National Internal Regulation (RIN) has explicitly authorised informational communication since 2014. GEO isn't advertising — it's about making your professional information legible to machines. No ethical difference from keeping your Bar listing or your Chambers entry up to date.
The 3 levers of the KaiZen method
Our KaiZen method, applied to more than 40 firms across Europe, rests on three foundations without which no durable AI visibility is possible. None is sufficient alone. All three must be deployed in parallel, in the continuous-improvement logic of the Japanese method our name draws on.
1. Cross-Platform Consistency (Unified Entity)
Generative AI doesn't blindly trust a single source. It validates your reliability by cross-checking institutional directories (National Bar Council, your local Bar), reference third-party sources (Chambers and Partners, Legal 500 EMEA, Décideurs), your digital profiles (official site, partners' LinkedIn, Google Business profile) and your press mentions (trade and legal press).
Any dissonance between these sources — a practice area mentioned on LinkedIn but absent from the site, a different address between the Bar and Google Business, inconsistent education between Chambers and the official site — is read by LLMs as a risk signal. The authority score drops, and your firm falls out of recommendations.
Concretely, Entity Reconciliation means auditing then aligning every mention of your firm across the web so it forms one unambiguous signal. It's the most thankless work in the KaiZen method, but also the one that produces the most immediate effect on AI visibility.
2. Specialised Semantic Precision
LLMs favour sources that demonstrate niche authority over broad generality. A firm presenting itself as "specialised in business law" will be systematically outranked by one presenting itself as "specialised in post-acquisition disputes in secondary LBO private-equity transactions." Precision creates relevance.
This specialisation must be expressed in three technical dimensions. First, standardised legal vocabulary: the exact terms of statutes, EU directives, and international conventions relevant to your practice. Second, concrete use cases: describing specific client situations rather than abstract skills. Third, Schema Markup in the LegalService format, which formally declares your practice areas to machines in a language they natively understand.
A page without Schema is interpreted at 30-40% precision by LLMs. A page with complete LegalService Schema is understood at 95%. That's the difference between appearing in an answer and being ignored.
3. External Authority Score
Your own pages aren't enough. LLMs grant disproportionate authority to third-party sources that talk about you: a mention in a trade-press article weighs more than a release on your own site. Wikipedia (ideally with a dedicated firm article), partners' actively maintained LinkedIn profiles, citations in mainstream business press, institutional listings (National Bar Council), professional rankings (Chambers, Legal 500) form the knowledge graph AIs use to distinguish genuine authority from a merely well-optimised site.
Building this external mesh is what separates a "technically optimised" firm from a "genuinely cited" one. It's also what takes the most time — hence the importance of starting now.
The KaiZen Muda: your firm's wasted knowledge
In the Japanese KaiZen method of continuous improvement, the concept of Muda (無駄) means the waste of resources — any activity or output that adds no value. It's one of the three pillars of the Toyota system that transformed the global car industry in the 1970s. Applied to a law firm in 2026, this lens reveals massive, invisible waste.
In a law firm, digital Muda is knowledge produced but not indexable. It's the brilliant case note emailed to three clients then archived. It's the analysis of an EU directive written for an internal file and never published. It's the conference talk only the title of which appears on LinkedIn. It's the technical memo whose relevance would have justified ten AI citations, but which has slept in a shared folder for three years.
Adopting a KaiZen approach to your AI visibility means turning each internal piece of expertise into a structured external signal. Not producing more content — making the most of what already exists. The KaiZen method takes the knowledge capital accumulated over ten or twenty years of practice and makes it machine-legible via Schema.org markup, FAQ structuring, targeted publication and Entity Reconciliation. It's continuous improvement, not revolution.
Comparison: before / after GEO implementation
| Criterion | Without a GEO strategy | With the KaiZen method |
|---|---|---|
| Monthly AI citations | 0 to 2 | 15 to 45 |
| robots.txt / GPTBot configuration | Missing in 9 firms out of 10 | Optimised and audited |
| Cross-Platform Consistency | Dissonance across 3 to 7 sources | Unified, reconciled entity |
| Visitor conversion rate | 1.76% (Google traffic) | 15.9% (AI traffic) |
| Time to appear in ChatGPT | Random (often never) | 3 to 6 months |
| Presence in Google AI Overviews | Very rare | Recurring on targeted queries |
| Marginal cost per qualified lead | High (paid + SEO) | Low (cumulative capital) |
Why Bredin Prat crushes its competitors in AIs
Our KVS (KaiZen Visibility Score) barometer measured the presence of 28 French and international firms in ChatGPT, Claude and Perplexity across 145 legal queries in French and English. The result is unambiguous: Bredin Prat scores a KVS of 100, crushing the market. Gide Loyrette Nouel caps at 64, and the Anglo-Saxon Magic Circle (Linklaters, Clifford Chance) doesn't exceed 18. The gap between the leader and its first French challenger is massive — the first-mover premium is extreme.
That gap isn't explained by historical reputation — all these firms are established references going back decades. It's explained by three technical choices Bredin Prat made earlier than its competitors. Exhaustive Schema structuring of its site that makes each practice area, each partner, each landmark matter indexable by LLMs. Dense, factual editorial content on its practice areas, designed to answer concrete questions rather than produce institutional discourse. A very active partner presence on LinkedIn with regular editorial posts, creating the external validation mesh. That's exactly the KaiZen method's triple combo: Cross-Platform Consistency + Semantic Precision + External Authority Score.
More important still: this advantage is cumulative and self-reinforcing. The more Bredin Prat is cited, the more it becomes a reference source for LLMs, the more it is cited. Firms falling behind today will pay an exponential catch-up cost in 18 months. The strategic window to join the top of French AI visibility is closing fast.
Key takeaways
- Roughly one legal search in four now goes through a generative AI — your firm may not be in the shortlist offered to your next prospect.
- 82% of French firms are invisible on Perplexity, and 9 in 10 haven't correctly configured GPTBot in their robots.txt.
- ChatGPT traffic converts 9 times better than Google traffic: being AI-cited means capturing very-high-intent prospects.
- The KaiZen method rests on three levers: Cross-Platform Consistency, Specialised Semantic Precision, External Authority Score. None is sufficient alone.
- The KaiZen Muda — knowledge produced but not indexable — is your firm's invisible waste. GEO turns it into a signal.
- The advantage is cumulative: the earlier you start, the higher your competitors' catch-up cost.