Persipica

Sample report

What a Persipica report looks like

One weekly observation of Fellside, a made-up brand of walking socks. The brand, answers, sources and numbers are invented for this example. The way they are collected and counted is how Persipica works.

Download the PDF How Persipica measures

Sample report from persipica.com/sample-report. Fellside and every number in it are made up.

Sample: Fellside is not a real brand

Fellside, walking socks

Observation
Weekly, collected 5 October 2026
Market
United Kingdom, in English. Set per brand.
Assistants
ChatGPT, Claude, Gemini and Perplexity, weighted equally. Perplexity is asked at two settings.
Prompts
27: 24 that don't name Fellside, and 3 comparisons that do, kept apart
Audiences
3 simulated audiences. Not real people or customer data.
Plan
Business. Every plan has every section; plans differ in brands, credits and, on Free, the assistants asked.

At a glance

The mention rate first, then the coverage and identity behind it, kept apart so one can't hide another. They cover the 24 prompts that don't name Fellside: 360 planned answers.

Rate

51% likely 32–69%

mention rate, averaged across 24 prompts and all four assistants, each counting equally, with its likely range. The previous observation found 48%, likely 30–67%. The ranges overlap, so this isn't a clear change.

Coverage

352 of 360

answers usable. 7 timed out on Gemini after retries, and 1 came back blank from Claude. 24 of 24 prompts answered by every assistant.

Identity

188 unverified

answers match Fellside's name or domain and count as mentions. If some are another company or a common word, the mention rates are lower than shown, and the likely ranges don't cover that. A separate name check rated a mix-up as low risk.

Brand identity has not been verified. Shared names may produce false matches.

By assistant

The main view. Each rate is the average across the assistant's 24 prompts, with its likely range.

AssistantNamedRateCoverage
ChatGPT46%33 of 72, likely 28–65%72 of 72 usable
Claude54%38 of 71, likely 35–72%71 of 72 usable, 1 blank
Gemini42%27 of 65, likely 24–61%65 of 72 usable, 7 timed out
Perplexity63%90 of 144, likely 43–79%144 of 144 usable, at two settings

By audience

The dashed outline is the brand named most often instead of Fellside. Each rate is averaged across all four assistants, and its range is across the audience's 8 prompts.

Simulated audienceNamedRateNamed instead
Weekend hikers8 prompts. 4 timed out76%likely 41–93%Ridgeway, 44%, likely 17–74%
Students on a budget8 prompts. 3 timed out, 1 blank71%likely 37–91%Tor Outdoor, 38%, likely 14–70%
Parents buying for growing kids8 prompts. All 120 answers usable8%likely 1–42%Pennine Mills, 79%, likely 45–95%

Findings

Each finding shows the prompts, the answers and the sources it rests on. In Persipica you can open every answer; this sample shows one per finding.

Finding 1

Parents buying for growing kids rarely hear about Fellside

Pennine Mills79%likely 45–95%
Fellside8%likely 1–42%

8 prompts, all 120 answers usable, averaged across all four assistants. The ranges don't overlap, so the gap is clear.

Asked as a simulated parentWhich kids' walking socks last the longest?

Claude, 5 October 2026, ask 2 of 3

For kids who wear through socks quickly, Pennine Mills' Junior Trek is the one most reviewers agree on. It's sold with wash-test results and a size guide for growing feet. Ridgeway's kids' merino is softer but thinner at the heel.

All 3 answers

Mentioned in 1 of 3 answers.

  1. Ask 1. Fellside not named. Pennine Mills' Junior Trek comes up most for durability, with Ridgeway's kids' merino as the softer choice.
  2. Ask 2. Fellside not named. The answer shown above.
  3. Ask 3. Fellside named. Pennine Mills' Junior Trek is the usual pick for hard wear. Fellside's Trail Junior is a cheaper option with a reinforced heel.
  • Most citedA review site's list of the longest-lasting kids' socks, cited in 48 answers. Fellside isn't on it.
  • Their pagePennine Mills' kids' range shows wash tests and a growing-feet size guide.
  • Your pageNo durability claims, so answers have nothing to quote.

Suggested fixDrafted, waiting for your review

Drafted for your kids' range pageTrail Junior has a reinforced heel and toe, and each size stretches across a full shoe size, so a pair keeps fitting as feet grow.

  • Your catalogueThe heel, toe and sizing come from your Trail Junior product data and size chart.
  • Asked of youPennine Mills' page quotes wash tests. Yours has none, so Persipica asks for your results instead of writing a number.
  • OutreachThe review list isn't on your site. Persipica names it as one to approach; it doesn't contact anyone for you.

Finding 2

Gemini quotes last season's price for Trail Junior

  • Your price£11.00, from your catalogue.
  • Quoted£9.00 in 2 of the 6 answers that give a Trail Junior price, both from Gemini. ChatGPT quoted £11.00 once and Perplexity 3 times.
  • SourceTwo outdoor retailers' product pages still list £9.00. Gemini's 2 answers cite them.

Suggested fixWith retailer

Ask both retailers to update the price. Persipica can't change their pages; it checks the quoted price again on the next observation.

Finding 3

Gemini also says Trail Junior is sold out

Asked as a simulated parentGood walking socks for a ten-year-old?

Gemini, 5 October 2026, ask 1 of 3

For kids, Fellside's Trail Junior costs about £91 and comes in three colours, but it's sold out in most sizes2, so try Ridgeway.

  1. Price. Your price is £11.00. The £9 comes from a retailer's old page (finding 2).
  2. Stock. Your catalogue shows Trail Junior in stock in every size. The answer cites no page for it.

2 of Gemini's 65 answers say Trail Junior is sold out. Like every match here, they count as mentions and are unverified, as At a glance explains.

Finding 4

AI agents: not a clear difference from Ridgeway

120 runs of one buying task by AI agents, 40 each on Claude, GPT and Gemini; each rate is averaged across the three agents. 3 broke off before a final pick, so they count for found and shortlisted but not for picked. The ranges overlap at every stage.

Found
82% likely 67–91%
Ridgeway 90% likely 77–96%
Shortlisted
41% likely 27–56%
Ridgeway 63% likely 48–76%
Picked
22% likely 12–38%
Ridgeway 38% likely 24–54%
  • Why RidgewayAgents could read its size guide and see stock on the product page.

Fix: put the size guide in textDone

Shortlisted went from 32% (likely 19–47%) to 41% (likely 27–56%), averaged across the three agents. The ranges overlap, so the next two observations will show whether it holds.

Products

Which product the answers naming Fellside mention, averaged across all four assistants, with each rate's likely range across the 24 prompts. An answer can name more than one product, or none.

ProductShareRatePrice quoted
Peak Merino43%likely 26–62%£18.00 correct in 58 of 61
Everyday 3-pack35%likely 19–55%£24.00 correct in 20 of 20
Trail Junior5%likely 1–21%£9.00 in 2 of 6, see finding 2

Prompts that name you

3 comparison prompts, such as "Fellside or Ridgeway for blisters?". All 45 answers usable. Each share is averaged across all four assistants, with its likely range across the 3 prompts. They are kept apart and never count towards the 51%.

Ridgeway50%likely 13–87%
Fellside39%likely 8–82%
No preference11%likely 1–65%

Sources the answers cite most

Across the 352 usable answers. In a real report each one links to the page.

  1. Fellside's own site74
  2. Ridgeway's product pages70
  3. A review site's longest-lasting kids' socks list. Fellside isn't on it.48
  4. An outdoor retailer's walking socks category41
  5. A hiking forum thread on blisters30
  6. A student money blog's budget kit list22

How to read this report

  • The audiences are simulated. They're profiles used to phrase prompts, not real people or customer data.
  • Rates are worked out prompt by prompt and then averaged, with each assistant counting equally, set before the observation. So a rate can differ by a point or two from its answer count.
  • A likely range is a 95% Wilson interval across prompts, and says how many prompts it rests on. It covers the uncertainty from the prompts that were answered. It doesn't cover answers that never arrived; that's what coverage is for.
  • One result is called higher than another only when their ranges don't overlap.
  • Real people can get different answers. Assistants tailor answers to a person's account, location and past chats, which no outside tool can see.
  • A rate shows how often AI names you, not what people buy.

The full method is on How Persipica measures.

See this report for your own brand

Start free Book a demo

No card needed to join. A weekly report like this one asks 27 prompts a week, about £30 an observation in credits on a plan.