A buyer asks ChatGPT: “Which customer onboarding software should a 20-person SaaS team use?” The answer names four vendors, gives each a quick pro and con, and links to two review sites and a competitor's comparison page. You're not on the list. The buyer never lands on your site, so your analytics never see them.
Here's the good news: you can find out why, and very often you can fix it. It rarely takes a site rebuild. Usually it takes a handful of specific facts the model was looking for and couldn't find on your pages. This guide shows which comparison questions to check, what being left off the list can cost, and what it costs to find out what's missing.
The questions where buyers pick a vendor
A comparison prompt is one where the buyer asks AI to choose, not to explain. They don't ask “what is a CRM?” They ask “which CRM is right for us?” That person already knows what they need and is narrowing the field. They're one step away from booking a demo.
You'll see five kinds of these questions most often:
| Type of question | Example | What the model looks for in sources |
|---|---|---|
| Best tool for a specific situation | “Best CRM for a manufacturer with 30 field sales reps” | who the product is for, what scale it handles, what it does for reps on the road |
| Head-to-head | “Pipedrive vs HubSpot for a small B2B sales team” | differences in price, limits, features and setup |
| Alternatives | “Cheaper Salesforce alternative for a 15-rep team” | price, scope, how easy it is to switch |
| Technical requirement | “HR and payroll software that integrates with NetSuite” | named integrations, which plan includes them, how data syncs |
| Budget and scale | “Help desk for 10 agents under $150 a month” | plan prices, seat limits, what's included |
A few more examples show how specific these questions get:
- “Which accounting automation tool works for a firm managing 80 small-business clients?”
- “Which e-signature tool integrates with Pipedrive?”
- “LMS for a 300-employee company that needs OSHA compliance training. What should we pick?”
- “Social listening tool for an agency managing 20 client accounts”
- “Which WMS can we go live with in under three months?”
- “Asana alternatives for a team that needs time tracking and invoicing”
You already have the best questions in-house. Your sales team hears them every day: “How are you different from X?”, “Do you integrate with Y?”, “What does it cost for 50 users?” Write them down in the buyer's words. Those are your first prompts to check.
How AI builds a vendor shortlist
To choose, the model reaches for sources that already compare: review platforms, listicles, competitor comparison pages, and documentation with hard facts on price, limits and integrations.
Vendors make the list when a source says plainly who the product is for, what it costs or how it's priced, what it integrates with and where it shines. “The all-in-one platform for modern teams” looks great on a banner, but it gives the model nothing to work with. “The Team plan is $12 per user per month, supports up to 50 people and syncs contacts with HubSpot both ways” gives it everything it needs to compare.
The biggest marketing budget doesn't always win here. Often the winner is simply the vendor with the clearest answer to the buyer's question. And that part is fully in your hands.
What missing the shortlist can cost
It depends on what you sell. Here are three example businesses. The numbers are assumptions, so swap in your own. In each case we assume, conservatively, that only a small share of buyers look at nothing but the AI shortlist.
A small subscription SaaS. A company sells an invoicing app for freelancers and small businesses at $59 a month. A buyer asks: “What's the best invoicing app for a freelancer that syncs with Stripe and QuickBooks?” Questions like this are common, say 500 a month. If one in ten people only considers the apps in the AI answer, that's 50 people. If one in five of them would have become a paying customer after the trial, you're missing 10 new subscriptions a month, or $590 in recurring revenue every month. After a year, that's 120 customers who would together be paying you about $7,000 a month.
A B2B SaaS with large contracts. A ten-person company sells fleet management software. An annual contract is about $8,000. A buyer asks: “Which fleet management software works for a company with 50 delivery vans?” These questions are rarer, say 40 a month, and one in ten people only considers the vendors in the answer. That's 4 people. If you usually win one deal in four, you're losing one contract a month. Over a year, that's 12 contracts and $96,000 in revenue going to competitors.
A regular online store. A shop sells coffee machines, and the average order is $400. A buyer asks: “What's the best coffee machine under $500 for a 10-person office, and where can I buy it?” In e-commerce, volume does the work: there could be 2,000 similar questions a month. If just 5% of buyers only shop where AI points them, that's 100 people. If one in ten of them would have ordered from you, you're missing 10 orders, or $4,000 a month.
| Business | Value of one sale | What may be slipping away |
|---|---|---|
| Small subscription SaaS | $59 a month | about 10 customers a month, about $7,000 a month after a year |
| B2B SaaS with large contracts | $8,000 a year | about 1 contract a month, $96,000 a year |
| Online store | $400 per order | about 10 orders, $4,000 a month |
Revenue grows differently in each case. A small SaaS wins on customer count, and every customer pays monthly, so the loss compounds over time. A SaaS with large contracts needs one good conversation for the audit to pay for itself many times over. The store sees the most questions, and what counts there is hard product facts: price, specs, stock and delivery time.
The exact figure matters less than the pattern. Whoever asks a comparison question is actively choosing a vendor. That's why getting back onto even one of these shortlists can move revenue more than hundreds of mentions on broad questions like “what is a CRM?”
What it costs to find out what's missing
One audit is one buyer question checked in one model. A $25 starter pack with no subscription covers roughly 4–7 audits, depending on the model. Premium at $99 per month covers roughly 20–35 audits a month. See current pricing. So your five most important comparison questions fit into a single month of Premium, even if you check each one in all four models.
For a small SaaS, a month of Premium pays for itself with two new subscriptions. One $8,000 contract is worth dozens of times more than a month of audits. For a store, a month of Premium pays for itself with a single $400 order. All it takes is one edit that puts you back on a list someone actually buys from.
And if the audit shows that every winning source is an independent review site? That's valuable too. You learn it in an afternoon, not after a quarter of writing content that never had a chance.
Budget separately for writing, fact-checking and publishing. For the full picture, see AI brand visibility audit pricing.
What you get from a comparison-prompt audit
For each question, the audit shows:
- The answer and the shortlist the model built, separately for ChatGPT, Gemini, Claude and Grok. Each model may recommend someone different.
- The sources behind the answer, so you see at a glance whether reviews, competitor pages or documentation are winning.
- A comparison with your page: which facts, names and conditions the winning source states that your page leaves out.
- A clear move for each gap: keep, rewrite, add an answer, get evidence, create a supporting asset or park the prompt.
Every recommendation links back to the question, the model's answer, the source and the passage on your page. Whoever writes the fix knows exactly what to change and why.
Two examples: from question to page change
Head-to-head. Say a buyer asks: “Competitor A vs your product for SaaS teams on a self-serve plan.” The model favors Competitor A and cites its comparison page. That page lists the plan price, the number of in-app guides and the Segment and HubSpot integrations in one table.
Your pricing page has the same plan, but the guide limit is hidden in a tooltip and Segment isn't mentioned at all. The audit flags three gaps: the limit, the integration and who the plan is for. The first move is to rewrite the plan description and add a short table with facts your product owner confirms. That's enough to start.
Alternatives. A buyer asks: “cheaper alternative to Competitor B for an accounting firm.” The model cites an industry ranking and the competitor's pricing page, which shows a price range. Your site only says “contact us for a quote.” The audit shows the model has nothing to compare because it doesn't know your price. The recommendation: publish a starting price or a sample cost for a typical firm, and say clearly what size of firm the offer fits.
Both are single-page edits, not content campaigns. For a longer walkthrough, see from prompt to page change.
When to do something else first
An audit pays off most when you can act on it. Start somewhere else if:
- you can't edit the pages that should answer the comparison,
- you don't yet know what buyers ask late in the deal (talk to your sales team first),
- every winning source is a third-party review and you have no plan to earn reviews,
- your product genuinely doesn't meet the condition in the question. Pick a question where you have a real edge instead.
For the wider problem of a better product losing to a cheaper one, read better product, customers choose cheaper.
Start with one question
Ask your sales team which comparison they hear most often late in deals. Run it in the model your buyers use, look at the sources and fix the first gap. Then check the same question again.
Audit a comparison prompt in TextRank. For the full process, follow the five steps to be visible in AI search, and compare tools in EchoRanked vs Peec AI vs TextRank.