Commercial model
Start with a visible, honest next step.
The Visibility Sprint is a fixed-scope answer-engine review: real collection from the models behind ChatGPT, Claude, Gemini and Perplexity through their official APIs, real observation by hand where it matters, and a named reviewer on every record. It is delivered inside the software as a verified evidence ledger, a prioritised backlog and one client-ready report. The software is free and there is no subscription. Everything on this page is either a limit the product enforces or a commitment we hold ourselves to, and each one says which it is.
Available now · one-time engagement
Visibility Sprint
Your client’s buyer questions are asked, without naming the brand, of the models behind ChatGPT, Claude, Gemini and Perplexity — the OpenAI, Anthropic and Perplexity models with web search on, and Gemini from model knowledge — and of five open-API models for comparison. Every answer is stored verbatim with the model, the sources its search cited and the time, and a named reviewer signs each record off. These are API answers, not screenshots of the consumer apps; read the methodology page before you buy. You get the ledger, a prioritised backlog, and one link you can hand your client.
Every record is human-verified
A named reviewer signs off each observation. Unverified records cannot reach your client’s report.
No number is presented as a measurement
There is no estimated score anywhere in the product. Every figure counts observations that actually happened. Read the method.
- Kickoff
- Within five business days of your purchase
- Delivery window
- 20 business days from kickoff
- Prompts observed and recorded
- At least 10, up to the 50 per audit the product enforces — the working set is agreed at kickoff
- Scheduled calls included
- Two: a kickoff call and a findings readout
- Work performed by
- AnswerLift — a named AnswerLift reviewer reads every stored answer and signs off each record before it can reach your client
- Provider API usage costs
- Included in the fee — collection runs on AnswerLift's own gateway accounts under a fixed spending ceiling per Sprint, and provider usage is never billed on to you
What the software costs: nothing. Create an account, create a workspace, run audits, record observations by hand, import evidence and crawler logs, work the remediation backlog and issue client reports — all free, with or without a Sprint. The one thing a Sprint unlocks is automated and scheduled collection from the answer-engine panel, because that spends AnswerLift’s own money on every run and the fee is what covers it.
Kickoff, delivery, calls and the reviewer are engagement commitments. The 50-prompt ceiling and the provider-cost position follow from what the product does, not from a promise on this page. None of these is a commitment about rankings, citations, traffic or what an answer engine will say — nothing in this product can deliver one.
What the Sprint produces, counted
Each row below is enforced by the product itself, not by a promise on this page.
| Deliverable | How many | What it is |
|---|---|---|
| Brands and markets | 1 brand · 1 market | An audit records one brand, one domain and one target market. A second brand or a second market is a second engagement. |
| Collected automatically | 9 answer engines | Led by the models behind ChatGPT, Claude, Gemini, Perplexity (OpenAI GPT API (web search), with web search; Claude API (web search), with web search; Gemini API (assistant panel), from model knowledge; Perplexity Sonar API, with web search (always on)), then five open-API models for comparison. Each answer is stored verbatim with the exact model, the search method and any sources the search returned. These are the models asked through their official APIs, not the consumer apps, and are never labelled as if they were. Copilot, Grok, Meta AI, Google AI Overviews are not measured — see the methodology page before you buy. |
| Recorded by hand | 3 consumer surfaces | ChatGPT, Google AI Overviews, Perplexity. A person asks the question in the interface and records what they saw. Nothing is scraped. |
| Collection cadence | From every 6h | Set per prompt, up to 50 tracked prompts per audit. A run that could not reach a provider is stored and shown as a failed run — never quietly dropped. |
| Priority questions in the brief | 3–5 | The intake will not accept fewer than three or more than five. These frame the work; they are not the full prompt set. |
| Evidence ledger | Unlimited records | Each record carries the prompt, its version, the engine, the source label and URL, the capture and observation times, the mention-versus-citation call, confidence, volatility and a written note. |
| Human QA sign-off | Every record | A named reviewer verifies each observation. Unverified records are excluded from the client report by the query itself, not by the page. |
| Remediation backlog | Written from findings | Every task is written by the person who decided it was worth doing, with owner, priority, due date and status. Nothing is seeded from a template. |
| AI-crawler log ingestion | Up to 5,000 rows | Import your own server or CDN log summary for GPTBot, PerplexityBot, ClaudeBot, Google-Extended and the rest. A crawler hit is a fetch, not a citation, and the product says so wherever it shows one. |
| Client-facing report | 1 shareable link | Read-only, excluded from search engines, carrying no AnswerLift analytics, and expiring 30 days after it is issued. It shows verified observations and nothing else — there is no score on it, because there is no score anywhere in the product. |
Done by a person
- Choosing the prompts worth asking, and the cadence for each.
- Checking the consumer apps by hand where the client wants to see exactly what a person sees.
- Reading every collected answer and confirming or correcting its classification.
- Signing off every record — collected or hand-entered — before a client can see it.
- Writing the interpretation and the prioritised remediation work.
- Exporting and supplying the AI-crawler logs from the client's own servers.
Done by the software
- Asking every agreed prompt of the 9-engine panel, on a predetermined model per engine with a same-vendor failover chain.
- Storing each answer verbatim with the engine, the exact model, the search method, any cited sources and the timestamp.
- Counting, in code, which engines named the brand, at what position, and which competitors were named instead.
- Making a provisional cited / mentioned / absent call for a reviewer to check.
- Deriving volatility only from repeated observations of the same prompt.
- Enforcing the human-QA gate on what can reach a client report.
- Rendering, exporting and expiring the client report.
Important boundaries
- Does not guarantee rankings, citations, traffic, revenue, or answer-engine behavior.
- Does not measure Copilot, Grok, Meta AI, Google AI Overviews automatically.
- Does not drive or scrape the consumer ChatGPT, Claude, Gemini, Perplexity apps. It asks the models behind them through official APIs.
- Cited sources come only from the engines that search the web (the OpenAI, Anthropic and Perplexity models). Engines asked from model knowledge produce no citations, and none are invented for them.
- Does not treat an API answer as identical to what the consumer app would say to a given person — each record names its engine, model and search method.
- Does not establish that any change you make caused any change in an answer.
- Does not produce an estimated or computed score of any kind. If nothing was observed, nothing is reported.
- Does not automatically publish content or make changes to a client site.
- Does not replace your agency's client approvals, editorial judgment, or publishing control.
Decision guide
Snapshot, Sprint, or the free workspace.
Choose the step that matches your current need. Agencies planning a multi-client rollout can talk to us.
| Path | Free Snapshot | Visibility Sprint | Free workspace |
|---|---|---|---|
| Availability | Available now | Available now | Available now |
| Commercial model | Free, no obligation | £2,750 one-time engagement | Free, no subscription |
| What you receive | What each of four engines said to three buyer questions about one brand, with the sources it cited, every answer checked by a person | A verified evidence ledger, a prioritised remediation backlog, and one shareable client report | Audits, hand-recorded evidence, the QA gate, the backlog and client reports |
| Best for | Seeing real answers before you buy | Getting defensible evidence in front of one client | Teams recording their own observations |
| Measurement | Three questions, the models behind ChatGPT, Claude, Gemini and Perplexity, one pass, every record reviewed | Automated collection across the answer-engine panel plus human observation, every record reviewed | Your own hand-recorded observations |
Buyer responsibilities
Bring the client context.
To keep the Sprint focused, your team provides the client domain, three to five priority questions, the target market, and the agency owner who approves scope and client-facing wording.
After purchase
- 01
Kickoff: confirm the agency, client context, questions, and working scope.
- 02
Collection and QA: the agreed prompts are asked of every engine on the panel, gaps are filled by hand, and a named reviewer reads and signs off every record.
- 03
Handoff: you receive the verified evidence ledger, the prioritised remediation backlog, and one expiring client-report link you can forward to your client.
Not sure where to start?