It depends on your sector and how contested it is. In most professional services, being cited in 20 to 30 per cent of relevant queries is strong performance. Leaders in less contested niches reach 50 per cent or more. Any benchmark quoted without a query set attached is close to meaningless.
The short version
- 20 to 30 per cent of relevant queries is strong in most professional services.
- 50 per cent or more is achievable in narrow, less contested niches.
- The rate is meaningless without the query set it was measured against.
- Which queries you are cited for matters far more than how many.
- Your only genuinely useful benchmark is your own baseline plus your named competitors.
Why the headline number misleads
A citation rate is a fraction, and almost all the argument is in the denominator. Cited in 60 per cent of queries sounds excellent until you learn the query set was twelve terms your firm already dominated. Cited in 15 per cent sounds weak until you learn it covers 200 competitive buying-intent queries across five platforms.
Anyone quoting a citation rate without telling you the query set is not telling you anything. That includes agencies quoting their own results, and it includes us.
The second problem is that not all citations are equal. Being named for “what is generative engine optimisation” is a definitional query from somebody learning a term. Being named for “best GEO agency for a regulated financial services firm” is a buyer with budget. A firm at 15 per cent concentrated in the second group is in a far better commercial position than a firm at 40 per cent concentrated in the first.
Third, rates vary sharply by platform. It is normal to see materially different performance across ChatGPT, Perplexity, Gemini, Claude and Google AI Overviews from the same content, because they retrieve and select differently. A single blended number hides that, and the platform breakdown is usually where the actionable detail is.
How to check it yourself
Build a fixed set of 20 to 50 queries that a real buyer would ask, in their language. Run them across the platforms that matter to you. Record whether you are named, and who is named instead. Repeat monthly with the same queries on the same day of the month.
The first run is your baseline. It will probably be lower than you expect, and that is useful rather than discouraging: it is the only number against which later work can be judged.
How to set a benchmark that means something
- Fix the query set before you measure, so you cannot be tempted to curate it afterwards.
- Weight by intent. Track buying-intent queries separately from informational ones.
- Track named competitors on the same set. Their share is the context your own number needs.
- Segment by platform rather than blending.
- Judge trajectory over absolute level. Moving from 4 per cent to 20 per cent in two quarters tells you the programme works. Sitting at 30 per cent and flat tells you it has stalled.
The measurement approach is covered fully in how to measure AI search visibility and GEO metrics.