The 5 criteria that matter

1. Empirical probing, not screenshots. One person asking ChatGPT once is an anecdote. A real audit runs a repeatable question set across all 4 major assistants and records every response. Ask the provider how many probes, which providers, and whether you can see the raw results.
2. Dimension coverage, not a single score. Visibility has causes. We score 5: citation (do the assistants name you), schema (structured data a machine can read), knowledge graph (does it know who you are), crawlability (can AI crawlers reach you) and social (the wider signal trail). An audit that returns one number without decomposing it can’t tell you what to fix. The distinction between ranking and being cited is covered in SEO versus GEO.
3. A benchmark cohort. A score of 44 means nothing on its own. Against a scored cohort of comparable employers it becomes a position, a gap and a priority. Ask what you’re being measured against and how often that cohort is rescored.
4. An evidence trail. Every finding should cite the specific probe, page or record it came from, with a date. If the report can’t show its working, you’re buying confidence, not measurement.
5. Repeatability. Models and indexes shift constantly. A methodology that can’t be rerun on the same terms can’t show progress. One-off audits age like fruit.

AI visibility is not AI transparency

Some employer brand indexes now score “AI transparency”: how clearly an employer explains its own use of AI in hiring, from CV screening to candidate chatbots. That is a fair question, and a policy one. It is answered by what you publish about your own process.

AI visibility is a different measure. It asks whether ChatGPT, Claude, Gemini and Perplexity name your organisation, and describe it accurately, when a candidate asks where to work. It is answered by evidence outside your careers site as much as inside it: structured data, knowledge graph records, crawler access, and whether other sites corroborate the story you tell.

The 2 need different evidence and different fixes. An employer can publish an exemplary AI-in-hiring policy and still never be named by an assistant. Another can be named in half of candidate prompts while saying nothing at all about how it screens CVs. A score that blends the two tells you neither.

So before you buy, ask the provider which one they measure. A useful audit says so in its first paragraph, publishes its method, and lets you see the prompts it ran. Ours measures visibility. The methodology is published and versioned, so you can check any score against the rubric that produced it.

Red flags

• A guaranteed ranking or citation outcome. Nobody controls the models; anyone promising placement is guessing on your invoice.
• No named methodology and no version. If the scoring basis is a secret, so is the value.
• A price that only becomes clear after a sales call.
• Generic SEO deliverables relabelled. Keyword rankings are not AI citations.
• An “AI” score that never says whether it measures your visibility to assistants or your policy on using AI, or blends the two into one number.

How ours works

The SetpointHQ Recruitment Marketing Audit is £999, fixed, stated here because we think audit pricing should survive being written down. It scores your employer brand on the 5 dimensions named in criterion 2, benchmarked against The SetpointHQ Index, the monthly-scored cohort of 100 UK employer brands we publish openly. Every finding carries its evidence and every score its date, and it is delivered in 7 working days with a prioritised action plan and a 60-minute debrief. The assistant-visibility half of it is set out on its own page, the AI Visibility Audit, scorecard and method included, so you can hold it against criteria 2 and 5 before you talk to anyone.

If you’d rather understand the mechanics before buying anything, start with how to make your recruitment brand show up in AI answers: the method is not a secret, only the labour. And if you want the crawler-access and structured-data checks run on your careers site at no cost first, the free careers site audit comes back in 3 working days.

Source: The SetpointHQ Index, 100 UK employer brands, scored monthly across ChatGPT, Claude, Gemini and Perplexity. Median composite 54.9 at 1 September 2026, rubric 2026.06.5.