Firms are watching organic traffic decline while their rankings hold. That is not an algorithm penalty. It is the answer layer absorbing the clicks that used to arrive.
The measurable shift
Pew Research Center’s behavioural study of 68,879 real Google searches found users clicked a traditional result 8% of the time when an AI Overview was present, against 15% when it was not. A randomised field experiment in 2026 measured a 38% reduction in organic clicks on queries where AI Overviews appeared, with zero-click outcomes rising from 54% to 72%.
Those are different methodologies pointing the same way. For informational queries, which is most of what a law firm blog targets, the click is no longer the default outcome of a good ranking.
Why legal is affected more than most
Legal search skews heavily informational. What is the time limit, do I need a lawyer, what happens at this stage, what does this cost. That is precisely the query shape the answer layer handles best, and precisely the content most firms have spent a decade building.
The transactional and local end of legal search is far more resilient. Someone looking for a specific firm, a phone number, an office location, or a lawyer in a named suburb still clicks. So the practical consequence is a reweighting rather than a collapse: general explainer content loses value, jurisdiction-specific and hiring-stage content gains it.
What actually gets a firm cited
Being in an AI answer is a function of whether the system holds a confident record of who you are and whether your content is structured enough to reproduce safely. In practice that means:
- A consolidated entity record: one canonical name, description and address everywhere the firm appears, with organisation and person schema carrying sameAs references to every profile you control.
- Answer-first structure, with the answer stated plainly before the reasoning.
- Explicit jurisdiction, dates and last-reviewed information. Undated legal content is low-confidence content.
- Named authors with verifiable credentials.
- Corroboration from sources the system already trusts: bar directories, industry bodies, legitimate trade press.
- Server-rendered pages. Client-side rendering blocks the crawlers that build these records, which is why a framework migration can quietly cause this problem.
The measurement change firms resist
Sessions will fall in this environment even when the work is succeeding. A firm that reports traffic to its partners will cancel a programme that was working. Move the reporting to branded impressions, citation appearances in a fixed monthly prompt set, and qualified enquiries from the CRM.
The impression-to-click gap in Search Console over eighteen months is the clearest single diagnostic available. A widening gap with stable positions is the signature of this shift rather than a performance failure.
What has not changed
The underlying requirement is the same one it has always been: be genuinely the most useful, most specific, most verifiable source on the question. The mechanism for rewarding that has changed. The standard has not.
| Pew Research Center’s study of 68,879 real Google searches found a traditional result clicked 8% of the time when an AI Overview was present, against 15% without. A 2026 randomised field experiment found a 38% reduction in organic clicks on triggered queries, with zero-click outcomes rising from 54% to 72%.Pew Research Center, 22 July 2025. Randomised field experiment reported April 2026. |
The eighteen-month check
In Search Console, chart branded and non-brand impressions against clicks for the last eighteen months. If positions held and the gap widened, this is your situation.
Get an AI search visibility assessment
We test what the answer layer currently says about your firm, and what it is drawing from.
Sources
- Pew Research Center, behavioural study of 68,879 Google searches, 22 July 2025.
- Randomised field experiment on AI Overviews and organic clicks, 2026.
- Clio, Legal Trends Report, 2026.
