Methodology
Every number on this site links here. Each one names the client, the date window, the sample size, and the instrument it was measured with.
We sell rankings and citations, and both are earned the same way: with sources that are attributable, dated, and specific. It would be strange to ask you to trust numbers we wouldn't expect a model to. If you find a figure on this site that doesn't resolve to an entry below, tell us and we'll take it down.
¹
Care Dale
- Industry
- D2C consumer health (shower filtration)
- Window
- April – July 2026 (13 weeks)
- Sample
- 30 prompts at the April baseline, expanded to 38 high-intent prompts across 4 AI engines.
- Measurement
- Prompt-level visibility measured against live ChatGPT, Perplexity, Gemini and Grok responses. Clicks and impressions from the client's own Google Search Console property over the same window.
- Attribution
- Named with the client's consent, given on the record.
Every figure we quote from this engagement
| #9 → #1 | Most-cited domain | Share of AI citations in the category, April vs July |
| 4× | Google clicks & impressions | Daily average, April vs July |
| 3.3% → 44.7% | ChatGPT visibility | Share of ChatGPT's tracked prompts, April vs July. April was one unlinked mention. |
| 30% → 76% | AI prompt visibility | Visible on 9 of 30 tracked prompts in April |
²
Legal services client
- Industry
- Legal & compliance services
- Window
- 18 Mar – 21 Apr 2026 (5 weeks)
- Sample
- 114 tracked prompts, expanded to 144. 23 articles published.
- Measurement
- Answers pulled from the live ChatGPT, Gemini, Perplexity and Grok web interfaces, not vendor APIs, on a fixed daily schedule. Referral sessions and impressions from the client's own Google Analytics 4 and Google Search Console properties over the same window.
- Attribution
- Anonymised. This client has not cleared us to publish their name, so we don't. We will name them here if and when they do.
Every figure we quote from this engagement
| 0 → 25 | Gemini referrals/day | From zero. Gemini sent no traffic before the engagement. |
| +44% | AI referral sessions in 5 weeks | 43 → 62 sessions/day |
| +43.8% | Total citations | 80 → 115 |
| +5.8pp | Mention rate | 25.3% → 31.1% |
| +49.2% | Organic impressions | 12.6M → 18.9M |
| +31.2% | Organic clicks | 67.5K → 88.5K |
³
Nuyug
- Industry
- D2C fashion jewellery
- Window
- 20 May – 14 Jul 2026 (8 weeks)
- Sample
- The brand's whole Google Search Console property: every query it surfaces for, not a selected keyword set.
- Measurement
- Impressions from the client's own Google Search Console property, read off the daily report on two named days: 20 May 2026, the baseline, and 14 Jul 2026, the latest day reported. Both are single days rather than averages, which is why the dates are published alongside the figures. The growth figure is those two readings divided, not a separately measured number.
- Attribution
- Named with the client's consent, given on the record.
Every figure we quote from this engagement
| 889 → 3.5K | Google impressions/day | 889 → 3,529 on 20 May vs 14 Jul 2026 |
| +297% | Google impressions in 8 weeks | 889 → 3,529 impressions/day, the same two days |
⁴
D2C innerwear brand
- Industry
- D2C apparel
- Window
- 2-week engagement, Q1 2026 (2 weeks)
- Sample
- 75 high-intent prompts. One piece of content published daily.
- Measurement
- Share of voice measured as the brand's share of total category mentions across ChatGPT, Gemini, Perplexity and Grok responses to the tracked prompt set. Clicks from the client's Google Search Console property.
- Attribution
- Anonymised. This client has not cleared us to publish their name, so we don't. We will name them here if and when they do.
Every figure we quote from this engagement
| #2 | Category share of voice | 9.6%, ahead of Zivame (8.9%), behind Jockey (10.4%) |
| 2× | Clicks | 200 → 400 |
What we don't claim
- These are four engagements, not a cohort. They are not a promise of what your brand will see, and the sample is far too small to generalise from.
- AI referral attribution depends on the engine passing a referrer. Some don't, so true AI-driven traffic is likely undercounted rather than over.
- Answer engines are non-deterministic. Citation counts vary run to run; we sample on a fixed daily schedule to reduce that, not to eliminate it.
- Organic impression and click lift correlates with the engagement window. We do not claim AEO work is the sole cause.