Count the question before the brand
We ran three fixed U.S. shopping questions through Cuebase’s Perplexity probe, three times each. The prompts covered a general commuting jacket, a packable jacket, and a jacket under $150. All nine requests completed across 14 and 15 October 2026.
We counted a model or brand when an answer positively proposed it for the shopper’s task. A conditional alternative counted once. A source link did not count, and a repeated name within one answer did not count again. That produced 13 proposal appearances, not 13 unique products or a ranked list.
| Fixed prompt | Completed answers | Positive proposal appearances | What appeared |
|---|---|---|---|
| Cycling to work: waterproofing, reflectivity, and fit | 3 | 6 | PEARL iZUMi Canyon 2.5L WxB and a Proviz REFLECT360 alternative appeared in every answer. The exact Proviz model was inconsistent, so its returned water-column claim is not repeated as fact. |
| Packable rain jacket: waterproof versus water-resistant | 3 | 3 | Castelli Emergency 2 appeared in every answer. Castelli’s checked regional page says it folds into half a jersey pocket; exact U.S. packed dimensions are unknown. |
| Under $150: model, price, and conditions | 3 | 4 | PEARL iZUMi Canyon appeared three times. One answer conditionally added REI Co-op Rainier. The saved price claims varied by color, size, and sale, so no current price is claimed here. |
The prompt totals add up to 13: six plus three plus four. PEARL iZUMi appeared in six proposal appearances, Castelli in three, Proviz in three, and REI Rainier in one. Recurrence is an observation from this small, fixed test. It is not proof of product quality, demand, or a buying recommendation from Cuebase.
Check the product record before using the pattern
A repeated product name can give a merchant a useful inspection list. It cannot replace the product page. We checked the manufacturer pages below on 15 October 2026. The official PEARL iZUMi Canyon 2.5L WxB page lists a 10k waterproof / 5k breathability rating, sealed seams, BioViz reflective elements, and a Relaxed Fit. That last detail matters: some saved answers described a close or athletic fit. The manufacturer page does not support that description.
Castelli’s checked Emergency 2 page calls it a 10,000 mm, 2.5-layer waterproof jacket that folds into half a jersey pocket. The page is regional, so it does not confirm U.S. availability, price, or an exact packed size. Proviz’s checked REFLECT360 City page lists 360-degree reflective fabric and a 5,000 mm waterproof rating; the answers that said only “Reflect360” did not identify the model well enough to support a universal 10,000 mm claim.
The price prompt makes the same point. The saved answers reported a PEARL iZUMi range from $74.83 to $150 and called the lower figure sale- or variant-dependent. That is a dated answer observation, not a price promise. One answer proposed the REI Co-op Rainier as a conditional alternative. REI describes that model as waterproof and packable, but not as a cycling-specific jacket.
What the tracked-brand report measured
For the report, we fixed five names: Showers Pass as the research baseline, then Proviz, Endura, GOREWEAR, and REI. No merchant data or customer account was used. The automatic detector found Proviz in three of nine answers and REI in three; Showers Pass, Endura, and GOREWEAR did not appear. It recorded zero matching tracked-domain citations.
That detector result is separate from the 13 product proposals. “REI” appeared in the text of every price answer because the answer cited or referred to REI, while only one answer positively proposed the Rainier jacket. Fixed tracked names are useful for measuring a defined comparison set. They are not the whole market, and a detection is not a product recommendation.
Sources need a second count
The saved responses contained 90 section-level source annotations: 32 distinct URLs on 23 exact hostnames. REI appeared in 22 entries and Cycling Weekly in 12. Those are returned source lists, not endorsements of every claim or a measure of product rank.
The automatic tracked-domain citation field was empty. That does not mean the answers had no sources. This provider placed saved URLs in section-level annotations, and the separate research harness preserves them in the source log. Cuebase’s normal Export CSV is a metrics export; it does not export raw answers or a source-by-source annotation log.
Use the same method for your catalog
Write three fixed shopper questions
In Share of Voice, use a broad question, a requirement question, and a buying question. Keep the wording and market statement fixed when you compare runs.
Choose a narrow comparison set
Add the competitor names and domains you actually want to measure. A share of voice is a share of that set, not a share of every product in the market.
Run and save the result
Choose Run probe now, then use Export CSV for the recorded metrics. Keep raw answer text and returned URLs in a separate research log when a claim needs source-level review.
Verify the returned claim on the product record
Check material, waterproofing, fit, size, care, price, and availability against the specific product record. Do not copy an AI answer into your catalog as evidence.
For the reporting definitions, read AI share of voice and what to count.
Method, limits, and downloads
These were Perplexity sonar responses from the DataForSEO AI Optimization LLM Responses API, run through Cuebase’s real provider and probe workflow with isolated research storage. They were not consumer Perplexity searches. Web search was requested, the output cap was 400 tokens, and the U.S. market was written into each prompt; no provider geolocation was configured.
We tested one engine, nine responses, two dates, and a fixed five-name report set. The study does not measure clicks, orders, product quality, demand, or the effect of a catalog edit. A 400-token cap can truncate an answer. The source annotations and automated citation fields are different data paths, so the downloads keep them separate.
- Download the prompt-level findings (CSV).
- Download the nine saved answers, method, and source annotations (JSON).
- Download the 90-entry source log (CSV).
- Read DataForSEO’s Perplexity response documentation.
Use the report to find a claim to verify, then use the product record to make it true. See Cuebase’s paid AI visibility tracking →
For your first measurement, use the AI visibility baseline guide.