Audit before you subscribe: see where ChatGPT and Claude put your SaaS, once, for free

The situation

The shortlist is the new first page

OpenAI and Claude answer a buyer question with a short list of tools. Across 4,500 answers in our study the mean list was 5.4 brands long. A buyer who asks "which CRM for a 10-person startup" reads those names, opens two or three, and rarely asks again.

Most SaaS founders find out they are missing from that list when a churned customer mentions how they found the competitor. Monitoring platforms sell a dashboard that watches the list every week for USD 95 to 499 a month. The first question is cheaper than that: are you on the list today, for the questions your buyers actually ask, and who is there instead of you?

What to measure first

Four numbers before any subscription

The audit reads your own site to derive the questions, then asks each model and keeps the raw answers as evidence. Measure these first; everything else is a dashboard around them.

1. 01

   The buyer questions, derived from your own site

   Up to 12 questions read from your own pages, in the language your site is written in ("best [category] for small teams", "[you] vs [competitor]", "alternatives to [competitor]", "[category] for [use case]").

2. 02

   Recommendation counts with their N

   For each question, whether each model named you, shown as "named in X of N answers" per provider, never as a single score. The verdict is one of three bands and comes from those counts.

3. 03

   The competitors named instead of you

   The brands that fill the slots you are missing, counted across the measured answers, so you know who you are actually losing the shortlist to, per model.

4. 04

   The pages you are missing

   Which questions have no page on your site that could answer them, which is usually the reason a model cannot name you. This is the list the publishing plan is built from.

What you get

What the free audit returns, and what the paid pieces add

The first run is free, with no signup and no card, and takes about a minute. The two paid pieces are one-time; there is no subscription.

One-time audit vs monitoring platform

Same job once, or a dashboard every month

The platform facts below were read from each vendor's pricing page on 2026-08-23; prices change, so re-check before buying. Asked Thrice's column comes from our own pricing.

The honest difference: a platform watches a fixed prompt set every week and charges for the weeks you do nothing; the audit measures once, gives you the plan, and lets you re-run when you have published something. Checked 2026-08-23.

• Otterly.AI
• Peec AI
• Profound

When to buy a platform anyway

Buy the subscription when these are true

A monitoring platform is the right buy for some SaaS teams. The audit is not a replacement for it in these cases; it is the diagnosis you run first so the subscription starts with a baseline.

• ✓Someone on the team acts on the data every week: publishes, measures, adjusts. A dashboard nobody opens is a cost.
• ✓You need to track more than 12 questions, or 100 or more prompts across several product lines.
• ✓You sell in several countries or languages and need the same prompt set measured per market.
• ✓You need seats, roles and alerts for a marketing team, or a reporting cadence a client or a board expects.

Repeatability

One run is a sample, so the audit shows counts with their N

In our repeatability study (4,500 calls, 30 categories, three model APIs) the same buyer question asked 5 times to the same model returned the same set of brands in 15 percent of cases. The first-named brand held in 75 percent of runs; the rest of the list moved. So a single run is a sample, not the answer: read every count with its N, re-run before you call a change real, and treat any tool that reports a 0-100 score from one run as printing noise.

That is also why there is no score on any screen: the verdict is three bands plus counts ("named in X of N answers"), and the methodology page lists the exact models, temperatures and run counts behind every number.

• How repeatable are AI recommendations (the study, raw answers downloadable)
• Methodology: models, parameters, what we do not claim
• How many runs does a measurement need

FAQ

Questions SaaS founders ask before running it

How long does the free audit take and what do I need?

About a minute for the first measurement. You enter your site, your category and the competitors you know about; the questions are derived from your own pages. No signup, no card. The models measured are OpenAI and Claude, through their APIs.

Is this the same as what the monitoring platforms measure?

Same kind of question (does a model name you), different scope and price. Peec AI Starter is USD 95/mo for 50 prompts and 3 AI models; Profound Starter is USD 99/mo billed yearly and ChatGPT only; Otterly Lite is USD 29/mo for 15 prompts (all as of 2026-08-23). The audit measures up to 12 questions from your site across 2 model APIs, once, for free, and the full report is USD 9 one time.

Why no 0-100 score?

Because the same question to the same model returned the same set of brands in 15 percent of cases in our 4,500-call study. A score to one decimal from one run is noise. You get counts with their N and a three-band verdict, and you can re-run to see whether a count holds.

Does the audit read the consumer ChatGPT app?

No. It measures the model APIs (budget-tier models, named on the methodology page), with no web browsing. Other people have measured gaps between the logged-in app and the API; we measure the API and say so.

What do I do with the result?

Ship the pages the plan lists first, then re-run the audit. The lifetime pass at USD 29.99 covers every re-run, so measuring again after publishing costs nothing extra. If you find you want someone watching it weekly, that is the moment a platform earns its subscription.