For performance marketers
If Perplexity skips you on a money prompt, that should not be a surprise.
AVOP is the trail: which surface cited whom, on which buying questions, with your competitors in frame. Less vanity counting, more "here is the page that lost."
What you are looking at
Surfaces
Six probes
The big chat apps plus Perplexity and Google AI Overviews when you run a full audit.
Prompt set (example)
Fifty prompts
Enough to split branded vs category vs “vs competitor” without boiling the ocean.
Brief turnaround
Same-week tasks
Engineering gets schema tickets, content gets the outline—no shared doc drift.
What your dashboard hides
- Position one for a head term does not prove an assistant will cite your pricing page when someone asks to buy.
- Listening tools trumpet mentions; they rarely show the precise answer text next to your competitor’s doc.
- AVOP ties misses to ship-ready fixes—schema, canonical mess, thin FAQ—so work does not stall in committee.
Three loops we see a lot
- 1
Monthly memo
Screenshot the score, paste the deltas, attach the three fixes you funded—same skeleton every time.
- 2
Competitor raid
Filter to prompts where you are absent but two rivals show up. Steal their structure, not their copy.
- 3
Brief from silence
Export the unanswered questions as a Notion board—content owns narrative, web owns markup.
When leadership asks for receipts
Brand mention graphs feel good until someone opens the assistant and sees a rival linked. AVOP is the receipt: URL, prompt, engine, date.
Early user notes
“We finally knew which comparison page actually cost us—not just 'AI mentions look soft.'”
“Product marketing and web eng stopped arguing in Slack without screenshots.”
After a full run you have
- A stack-ranked fix list—not “improve content,” but actual files and tags
- Competitor overlays on the prompts that overlap with pipeline, not generic share-of-voice
- A delta you can paste into the monthly performance memo without rewriting it
Fabricated strip—real layout
Numbers are made up; spacing and labels are not.
Pricing-adjacent prompts
Score & proof
Eighteen asks where you never appear
Competitors
Rival A owns eleven with a comparison doc you do not have
Do this next
Write the FAQ row that names them directly—ship schema after the copy lands
Engine quirks (demo)
Score & proof
Claude keeps citing your /docs, Gemini grabs blog tutorials
Competitors
Perplexity still likes head-to-head tables
Do this next
Stop forcing everything through the homepage—give each engine the page shape it keeps picking
Backlog sanity
Score & proof
Nine tickets, each tagged small/medium/large
Competitors
Most losses sit on “pricing vs X” and onboarding comparisons
Do this next
Eng takes JSON-LD, marketing takes the blunt comparison paragraphs
Your assumptions
Back-of-napkin pipeline
Multiply the lead count and deal size you already use internally. We annualise it—no magic, no promised lift.
Leads × deal size × 12 months — tweak either field and the total follows.
If those numbers held all year
This is arithmetic, not a model of AI traffic. Use it to sanity-check whether the problem is worth your time.
Marketer FAQ
Which engines?
Major LLMs in prod, plus Perplexity and Google AI Overviews on paid audits.
Free audit?
Teaser depth. Paid widens prompts, competitors, and history.
Export?
CSV + PDF on paid plans.
Start with one domain, embarrass fewer people in the Monday meeting
Pick the worst five prompts first—prove value before you expand the stack.