How to Get Your SaaS Product Into AI 'Best Tools' Lists

What did ChatGPT and Claude cite when we asked for the best tools?

ChatGPT cited the vendors' own pages, and Claude cited roundup lists. On 14 October 2026 our content engine asked each assistant for the best tools in three software categories, listed under How we tested.

• ChatGPT checked each vendor by name. After at most a generic 'best [category]' search, it searched each candidate by name for pricing, integrations and product pages. It cited only vendors' own pages and a vendor's news release, never a review site or roundup.
• Claude read the roundups. It ran a broad 'best [category]' search per question and cited listicles, some published by a vendor in the same category. Every answer warned that many 'best' lists are vendor-written and told the reader to check each vendor's site or run a trial.
• Both matched tools to buyer situations. Each answer paired every tool with a situation, such as team size, card spend versus reimbursements, or in-app chat versus email support, then asked the reader for details such as their stack or channels.

How does an AI assistant decide which tools make its 'best tools' answer?

An assistant finds candidate names on third-party and vendor-written lists, then draws on pages about each tool to describe it.

Ahrefs analyzed ChatGPT responses to 750 top-of-funnel prompts across software, products and agencies. 'Best X' blog lists were the largest page type among its sources, at 43.8%, and brands placed high on third-party lists were more likely to appear in the answers.

DerivateX's May 2026 study of B2B SaaS citations covered 233 ChatGPT recommendations across 40 categories. ChatGPT cited the recommended tool's own site only 11.6% of the time, and 87.4% of cited recommendations credited a third party.

Nathan Mzumara's August 2026 analysis of 367 citations in 60 software buying answers found 65.9% pointed at a vendor's own website. The studies count different things: DerivateX counted a tool's own site backing that tool, Mzumara a page on any vendor's site.

Mzumara also found vendor pages were the primary source for checkable facts such as price, tiers, features and integrations.

Getting named and getting cited are separate events, as DerivateX puts it. In Ahrefs' experiment with self-promotional pages, an assistant used a promotional page as research and then recommended a competitor.

Which buyer prompts decide whether you make the list?

Shortlisting prompts decide it: best-of, best-for, head-to-head and alternatives queries. Each pulls a different kind of page, as the table below shows.

Arobis's guide to generative engine optimization says a brand cited in 'what is X' answers can still be missing from 'best X for [ICP]' prompts, which is where shortlists get made.

These prompts also send the assistant to the web more often. HubSpot reports Nectiv's finding that commercial-intent prompts trigger a ChatGPT web search 53.5% of the time, against 18.7% for informational ones. Words such as reviews, free, features and comparison are among the most common search triggers.

How do you get onto the third-party lists AI assistants cite?

Find the lists assistants already cite for your prompts and earn an accurate place on them.

1. Find the cited lists. Run your shortlist prompts and log every roundup cited. For clients we rank these as an earn-citation list of sites AI cites in your space that do not mention you yet, each with a play such as a guest post, partnership, listing or review.
2. Favor lists that get updated. Of 1,100 dated blog lists ChatGPT used as sources, 79.1% had been updated in 2025, Ahrefs found. A list that gets refreshed can add you at its next update.
3. Include small sites. DerivateX found domain authority did not predict which pages ChatGPT cited for software recommendations. Ahrefs found 35% of the 'best' lists ChatGPT cited were on low-authority domains, and 28% of ChatGPT's most-cited pages had no organic visibility.
4. Pitch with a reason. Tell the author which buyer situation you fit, with current pricing, integrations and plan limits.
5. Use the same description everywhere. Arobis describes a company that called itself an AI customer success platform on its site, a support automation tool on G2 and helpdesk software in a directory. Pick the category words buyers use and repeat them on every profile.
6. Show up where buyers talk. We draft channel-native posts for LinkedIn, X, Reddit, Quora, Medium, Substack, YouTube, G2 and earned media, for a person to post.

Never pay for a place on a list, and never buy reviews: Strivelabs notes that incentivised reviews violate most review platforms' terms. Our guide to the earned media sources AI engines cite covers the wider pitching method.

Should you publish your own 'best tools' list or comparison page?

Publish an honest comparison if buyers would already accept you as an answer to the query. If your own list ranks you first, weigh two signals: in our test Claude warned readers about vendor-written lists in every answer, and Ahrefs advises against manufacturing a ranking (quoted below).

Ahrefs tested this with 34 self-promotional pages on 5 domains, then analyzed 9,886 answers from ChatGPT, Gemini, Perplexity and Copilot. For its new conference brand, 82% of new mentions in previously empty slots came in answers citing its own pages. For its established tool, only 6% did, and 94% came from third-party content.

Ahrefs concludes that self-promotional pages seem to help most when your brand is already a believable answer. Its test before publishing: would your brand look like a natural recommendation even if you had not written the page? Query fit counts too (see the table above).

On ranking yourself, Ahrefs says, "Fill the gap accurately; don't manufacture a ranking."

DerivateX still recommends a current, well-structured 'best [category] software' comparison on your own domain. In its data, a vendor's 'best procurement software' page was the cited source behind recommendations for multiple brands within a single answer.

On that page, say where a rival fits better; our guide to comparison pages AI cites for a challenger shows how. Ahrefs warns that keeping comparisons accurate can become a burden without automation built in from the start.

Format counts as well. In DerivateX's data, every page ChatGPT cited for software used numbered or bulleted lists, 68% had a comparison table and 56% had an FAQ section.

What should an ICP-fit page say so AI can match you to a buyer?

An ICP-fit page should name the buyers you fit best and the ones you do not, in the words buyers use about themselves. Our test showed why: assistants answer 'best for' prompts by buyer situation.

HubSpot says a product page that explicitly states a use case will outperform one that claims to serve all businesses. It suggests sections such as 'Best for freelancers' or 'Ideal for enterprise'.

Put these on the page, each in a plain sentence:

• Company profile. Industry, business model and region of your best-fit customers.
• Team size and stage. The headcount band and funding stage where it fits, in the buyer's own terms.
• Sales motion. Product-led self-serve, sales-assisted or enterprise procurement.
• Stack. The CRM, accounting, warehouse or support tools it connects to, named exactly.
• Who it is not for. The buyers another tool serves better, and why.

We research each buyer persona from the client's market (role, company type, goals, pain points and decision criteria), because each persona gets a different answer, and the same personas brief the pages we write.

Do use-case, integration and template pages help?

Yes, when each use case, integration or template has its own indexable page. Ahrefs found landing pages are commonly cited when ChatGPT recommends software. When a brand's own site was the first recommendation, an internal landing page was more likely to be cited than the homepage or the brand's own list.

Google says AI Overviews and AI Mode may use query fan-out, running related searches across subtopics. A page per subtopic gives those searches something to find.

• Use-case pages. Say what the buyer gets done, for which team, and what it takes to set up.
• Template pages. Give each template a crawlable page saying what it builds and what it needs: data, plan and integrations. Templates that exist only inside the logged-in app cannot be found by an assistant's search.
• Integration pages. In our test, ChatGPT cited a vendor's help article on its QuickBooks Online setup. State what syncs, in which direction and on which plan.

How much do G2 and Capterra reviews matter?

Review profiles appear to work as an eligibility check for an assistant's picks. A study summarized by Strivelabs found 100% of tools ChatGPT recommended had Capterra reviews and 99% had G2 reviews, which Strivelabs reads as an eligibility gate.

The citation data, what to ask reviewers for and what to stop spending on are in our analysis of G2 reviews and AI recommendations.

Which site basics can keep you out of AI answers?

A crawler block or an unindexed page can keep an engine from using your pages, and hidden pricing is the third thing to check.

If your robots.txt opts out of OAI-SearchBot, your pages drop out of ChatGPT's search answers, though they can still appear as navigational links, per OpenAI's bot documentation.

To be a supporting link in Google's AI Overviews or AI Mode, Google's AI features documentation says a page must be indexed and eligible to show with a snippet. No special schema.org markup is required. Our post on the robots.txt mistake that blocks AI crawlers shows how to test a CDN or firewall rule as well as the file itself.

HubSpot adds that a product whose pricing is hard to surface, such as contact for pricing with no ranges, is at a disadvantage. Our guide to pricing pages for AI answers covers what to publish.

How do you know whether you made the list?

Run a fixed set of buyer prompts on each engine, repeat the runs, and log whether you appear, where, and which source was cited. HubSpot suggests manually running 10 to 20 queries in ChatGPT each week that match your target buyer's language.

Repeat because answers move. In Ahrefs' experiment, only about 54% of cited URLs in Google AI Overviews carried over between consecutive checks, and the longest unbroken citation run lasted 52 days.

We count answers where an assistant recommends the brand separately from answers that only mention it in text or only link to it as a source.

Our AI visibility platform runs ChatGPT, Claude and Gemini with web search by default, can add DeepSeek and Grok models, and repeats every prompt on each engine. For the full setup, see how to measure your AI visibility.

How long does it take, and how do you choose help?

In our client work, the first change you can measure tends to show up after 60 days or more. Nobody controls what an assistant says, and we do not guarantee results or promise a specific ranking, citation or place on any list.

OnlyAEO is an AEO platform that tracks where AI names your brand and creates the content to change it. Our team can run it for you on a managed plan, and your team reviews and approves every piece before it goes live.

For list placement, the relevant parts of the platform and its managed plans are help getting mentioned on trusted sites and listings AI reads, fact-checked articles on your own site, monthly AI visibility analytics and a website health check.

Our published FastTrackr AI case study starts at 0.1% AI visibility and #338 of 458 brands, and ends roughly 100 days later with the brand at #3 among the most-cited in its category. That program covered more than list placement, so it does not isolate any tactic in this article.

When you compare partners, look for:

• Reports by prompt and engine. The checks described above, run before and after the work.
• A named list of target sites. Something like the earn-citation list in step 1 above.
• No guaranteed spots. For the reason given above, treat any promised placement as a red flag.

How we tested

Our content engine put a buyer question to ChatGPT and to Claude Sonnet, each with web search on, in each of these categories: expense management tools for a growing startup, product analytics for a B2B SaaS company, and help desk software for a small SaaS support team.

The engine saved every search each assistant ran, every page it cited and the full answer.

We did not repeat the runs, so the results show what the assistants did on that day and nothing more.

More questions about AI 'best tools' lists

Should a comparison page put the year in its title?

DerivateX found 78% of the pages ChatGPT cited for software carried the current year in the title or headline. Change the year only when you actually refresh the page.

Do ads on ChatGPT change how tools get into answers?

OpenAI's bots documentation lists OAI-AdsBot, which checks the safety of landing pages submitted as ads on ChatGPT. Ads go through their own review, so treat them as a separate paid channel.

Is Reddit worth the effort for SaaS shortlists?

Reddit was the most-cited single domain in DerivateX's study, with 16 citations. Mzumara's corpus gave community content no citations at all, so check whether Reddit threads appear for your own prompts before you invest.

Does any one site dominate the citations in software answers?

No. The most-cited domain in Mzumara's corpus of ChatGPT software buying answers held only 4.9% of its citations, so no single roundup or publication dominated that corpus.