SEO and GEO for SaaS and B2B — How to Get Recommended When the Customer Asks AI "Which Tool Should I Pick"

Before a CTO books the first demo, they ask ChatGPT: "which infrastructure monitoring tool for a 20-person team, AWS stack, budget up to $500 a month". They get three names with justifications. If your product isn't among them, you just lost a deal you'll never hear about — because B2B buyers spend a combined 17% of the purchase journey with the sales reps of all considered vendors (Gartner), and the rest on silent research. And that research has moved into AI: per G2's 2025 research, GenAI chats are now the number one source influencing vendor shortlists — 17.1% of mentions, ahead of review sites (15.1%), vendor websites (12.8%) and peer recommendations (8.9%) — and about half of software buyers start their research with an AI conversation.

GenAI chats are now the number one source influencing B2B vendor shortlists — 17.1% of mentions, more than review sites (15.1%) and vendors' own websites (12.8%) — and about half of software buyers start their research with an AI conversation (G2, 2025). Buyers spend a mere 17% of the purchase journey with sales reps — the decision largely forms before anyone fills in a form. How to make the models recommend your product in that invisible phase: comparison pages, quotable pricing, G2 and communities, and category-level SoV measurement.

This post closes my verticals series — after local SEO and e-commerce — and answers the question worth the most in B2B: what decides whether a model names your product in response to "which tool should I pick" — and how to get onto that list. The spoiler that frames the whole strategy: your own website is just one of several sources models assemble a recommendation from. The rest is reviews, rankings and conversations — places where you win differently than with Google rankings.

The B2B funnel has flipped — numbers, not vibes

Three consequences of these numbers that change planning. First, the invisible phase has grown: if a buyer spends 5–6% of their time with any one sales rep and does the research in AI, most of the contest happens where neither the CRM nor form analytics can see. I described the symptoms in the Share of Voice post through the story of a SaaS marketing lead whose sales team kept reporting "I checked in ChatGPT and it recommended someone else" — with TOP 3 Google rankings and traffic looking normal. Second, the traffic that does arrive from AI is worth more: the buyer has done the basic research, so they convert several times better (on the order of 4.4× in Semrush's measurements) — and in the B2B data I collect for clients, traffic from Claude stands out as the best-converting AI source. Third, the classic "fewer sessions = fewer leads" arithmetic stops working — the full argument and the new KPI set are in the zero-click era strategy.

Where models get their tool recommendations

When a model answers "which tool for X", it doesn't invent the list — it assembles it from sources that are surprisingly repeatable in SaaS categories:

1. 1.Review platforms — G2, Capterra, TrustRadius. Structured ratings, categories and comparisons are a ready-made recommendation base for models; they're cited regularly for "best tool for X" questions.
2. 2.Rankings and "best X for Y" roundups — comparison articles in industry media and strong blogs. These most often supply the "why this tool" justifications.
3. 3.Communities — Reddit, forums, groups. "What do you recommend for X" threads weigh disproportionately to their reach; how to show up there without blowing up is in the Reddit, forums and UGC post.
4. 4.Your own domain — but not the homepage: the specifics. Comparisons, pricing, documentation, case studies. Pages an answer can be cut from.
5. 5.Wikipedia/Wikidata and the Knowledge Graph — the identity layer: who the company is, since when, in which category. The foundation from the entity building post.

The strategic conclusion: visibility in a SaaS category is a portfolio of sources, not one domain. A company with a great blog but an empty G2 profile and zero community presence hands the models two of the three main votes about itself.

The bottom-funnel that wins recommendations

In B2B the classic content investment order flips: before you build hundreds of guides, build the pages models assemble decision answers from.

"X vs Y" and "alternatives to X" pages. The buyer compares — and so does the model. A comparison page written honestly (a feature table, prices, "when to pick them, when to pick us" scenarios) is quotable precisely because it isn't a puff piece; admitting a competitor wins in a specific scenario raises the credibility of everything else. Build them for your 3–5 most-compared competitors and for "alternative to [category leader]" phrases — the shortest route into answers about your category.

Pricing as quotable content. To "how much does X cost and is it worth it", the model will answer from someone's page — the only question is whose. An open price list with tiers, limits and an FAQ gets cited; "contact us for a quote" is invisible, so the answer about your prices gets assembled by your competitor or a stale forum thread. If your business model requires custom quotes, publish at least ranges and rules ("from $X per seat, discounts from N licenses") — you control the answer's content without giving up the sales process.

Case studies with numbers. "A company like mine achieved Y in Z months" is the most-cited proof in B2B recommendations — provided the case has concrete numbers, an industry and a scale, not "significant efficiency gains". The writing for retrieval pattern applies here in full: the result in the first sentence, context after.

Documentation and changelog. An underrated citation source for technical questions ("does X integrate with Y"). Public, indexable documentation answers dozens of decision questions marketing will never cover.

Integration pages as programmatic. If the product connects to dozens of tools, every "[your product] + [tool]" pair is a separate decision intent — and a ready-made matrix for programmatic SEO: the integration data lives in your own API, the template answers "how do I wire this up and what do I gain", and the long tail of such queries usually has no competition at all. It's the same pattern Zapier built tens of thousands of pages on — at a single SaaS's scale, a few dozen solid ones are enough.

G2 and Capterra — a profile isn't enough

Since review platforms are the second vote about you, treat them as a channel, not a business card: a complete profile in the right category (models inherit these platforms' taxonomy — a wrong category means wrong recommendation contexts), systematically — and within the platforms' rules — asking customers for reviews after implementations, and replying to negative reviews, because those enter the same corpus models build sentiment from. Freshness matters: a category where a competitor collected 40 reviews this year and you collected 40 three years ago reads, in answers, as "once popular, now fading".

Topical authority — the entry ticket, not the strategy itself

None of the above excuses skipping topical authority — content clusters around the problems your product solves give models a reason to treat the domain as expert and feed the top of the funnel. But in B2B the order is the reverse of blog-guide orthodoxy: decision pages first (comparisons, pricing, case studies), educational clusters second — because the first dollar comes from "what should I pick" questions, not "what is this".

Measurement: SoV for your product category

Measuring "does AI recommend us" is more concrete in B2B than anywhere else, because decision questions are countable. Build a set of 30–50 prompts around the category (with variants: company size, stack, budget, industry) and measure three things monthly: Share of Voice (in how many answers you appear vs competitors — methodology in the SoV post, tools in the monitoring comparison), position and context (recommended as first choice or mentioned with a caveat) and citation sources (which domains feed your category's answers — a ready-made action list). Plus one CRM change: a "how did you hear about us" field with a "ChatGPT/AI" option — the cheapest measurement of the funnel's invisible phase there is, and the natural closing of the lead generation automation that qualifies those leads downstream.

Closing the series: three verticals, three different games

Finally, the thing that sorts the strategy if you operate several models at once. Locally you win with the Business Profile, reviews and a consistent place entity; in e-commerce — with the product feed and structured data; in B2B/SaaS — with a portfolio of recommendation sources and decision content. A different funnel, different citation sources, different KPIs. Only the foundation is shared: an unambiguous entity, content written for extractability, and measuring what the models say — before falling sales tell you instead.

A step-by-step implementation plan

1. 1.The baseline category audit — 30–50 decision prompts across 3 models; record SoV, contexts and cited sources. It's the map for everything else.
2. 2.Build comparison pages for 3–5 competitors and "alternative to [leader]" phrases — honest tables, selection scenarios.
3. 3.Publish pricing or at least ranges and rules — the answer about your prices should come from you.
4. 4.Sort out G2/Capterra — the right category, fresh reviews after every implementation, replies to negatives.
5. 5.Enter the category's communities — mention monitoring and expert answers where buyers ask "what do you recommend".
6. 6.Rewrite case studies around numbers — result, industry, scale in the first paragraph; delete "significant improvements".
7. 7.Add "ChatGPT/AI" to the CRM source field and tie it to AI traffic analytics.
8. 8.Measure monthly, act quarterly — SoV and citation sources decide whether the next sprint goes into content, reviews or communities.

---

I build SaaS and B2B visibility in AI answers end to end: from the category SoV audit, through decision pages and review strategy, to monitoring and reporting. I do this as part of AI optimization (GEO) and AI consulting. I teach it in the SEO & GEO course. Get in touch — I'll start by checking who the models recommend in your category and which sources those answers are built from.

Worth reading next:

• Zero-click and declining organic traffic — a brand strategy
• How to measure brand Share of Voice in AI models
• Reddit, forums and UGC — why AI models cite communities
• B2B lead generation automation with AI