SaaS GEO Playbook Whitepaper
SaaS GEO: Support Buyer Research and Evaluation
Make SaaS product information easier to evaluate. Read the scope of Forrester, 2X and 6sense findings, then plan comparison content, evidence and measurement.
v1.1 · October 2026 · Tenten GEO Research
CH.01
Help Buyers Check Product Information
Buyers may research products before speaking with sales. Useful content answers their questions and gives them evidence they can check, whether they arrive from search, an AI answer or a colleague.
Forrester’s January 2026 summary draws on its 2025 survey of nearly 18,000 global business buyers: 94% reported using AI in the buying process. Respondents also checked AI output with peers and experts. This is not evidence that the entire pre-contact journey happens inside an AI chat.
The often repeated “70%” describes an older estimate of when buyers first contacted sellers, not time spent using AI. In 6sense’s 2025 study, that point moved from about 69% of the journey in 2024 to 61%. Vendor conversations still matter. Make your website useful before the conversation and specific enough to support the questions that follow.
What to do
Start with questions your sales and support teams already receive. Explain who the product fits, its limitations and how to verify its capabilities. AI visibility is one way to assess distribution; it does not replace product clarity or human evaluation.
CH.02
Read the 2X Visibility Study in Context
A vendor benchmark can suggest questions to investigate. It cannot tell you that every company in your market has the same visibility problem.
2X’s April 2026 release reports a proprietary audit of 70 B2B companies. It classed 4.3% as having a healthy discovery funnel; the other 95.7% appeared primarily in queries where buyers already knew their names. “Primarily” does not mean never mentioned in early discovery.
How 2X Classified Its 70-Company Sample
2X AI Visibility Index, April 7, 2026 · https://2x.com/press-release/2026-2x-ai-visibility-index-b2b/
Publisher-defined categories in this sample, not a population estimate. The release does not provide the full prompt set or repeat-run protocol; no causal effect of a website change is established.
For your own baseline, separate category questions from branded questions. Record the platform, interface, language, date and exact prompt. Keep brand mentions, cited URLs and recommendations as separate observations. A missing mention in one answer is not proof that a brand is ineligible.
- →Use existing customer questions to choose a manageable sample.
- →Keep successful responses, errors and unavailable answers distinct in the denominator.
- →Compare repeated observations before assigning a content fix; preserve unknowns.
What to do
Use the study to motivate a baseline, not to label your business invisible. Prioritize gaps you can reproduce and connect to a real buyer question.
CH.03
Make Product Comparisons Useful
Forrester’s February 2026 summary reports AI use for product comparison (55%) and product research (54%) in its 2025 buyer survey. These uses include private workplace tools; they are not all public search or necessarily before sales contact.
A comparison page should help a buyer judge fit. Name the versions, date the checks and link to evidence. The examples below describe content quality, not observed AI rankings or named competitors.
| Buyer question | Useful information | What remains unclear | How to verify |
|---|---|---|---|
| Who is it for? | Supported team sizes and actual use cases | “For every business” without examples | Product documentation or a scoped trial |
| What does it cost? | Plans, billing units and important exclusions | Missing units or unexplained custom pricing | Current pricing or a written quote |
| What can it connect to? | Supported integrations and known constraints | A logo with no implementation details | Integration documentation and ownership |
| What evidence is available? | Dated reviews and permission-cleared cases | Untraceable ratings or unattributed results | Original source, sample and method |
| What requires a conversation? | Security, migration or procurement constraints | Unknown requirements presented as solved | An appropriate specialist or evaluation |
Tenten’s editorial checklist for a useful comparison; no citation or sales outcome is implied
Custom pricing can be legitimate. Explain what affects the quote and what the buyer should prepare. Do not invent a public price or disclose confidential terms to satisfy an assumed AI preference. Clear information helps evaluation without guaranteeing a recommendation.
What to do
Choose the comparison questions that matter to your buyers. Publish verified answers, include material limitations and label anything that still requires confirmation.
CH.04
Support the Buyer’s Internal Decision
Forrester reports that 47% used AI to develop a business case. That describes a task, not an obligation to publish private customer data or a promise that an AI answer will endorse a vendor.
Give the buyer material they can check and share appropriately: scope, cost assumptions, implementation dependencies and approved evidence. If no measured return is available, explain the evaluation method rather than inventing a result.
| Internal question | Evidence to check | Useful material |
|---|---|---|
| When might the investment pay back? | Costs, assumptions and measurement period | A documented model or an approved measured case |
| What time or cost could change? | Comparable baseline and calculation method | A results page that separates estimates from observations |
| Is the use case comparable? | Industry, scale and relevant constraints | A permission-cleared case with its context |
| What does rollout require? | Responsibilities, dependencies and risk | Current rollout, support and security documentation |
| How do alternatives differ? | The same scope and dated product versions | An evidence-linked comparison with limitations |
Tenten’s suggested evidence structure; the buyer must still validate fit and assumptions
Keep core public facts accessible as text where appropriate. PDFs can be indexed; gated material depends on access, and video may need supporting text. Choose formats for the reader and confidentiality requirements. Do not assume every crawler or private AI tool can retrieve every file.
What to do
Review the material with the teams responsible for product, finance and customer permissions. Publish what is accurate and approved, then keep its owner and review date visible.
CH.05
Distinguish Familiarity from AI Influence
6sense’s 2025 research describes strong early preferences in B2B buying. It does not establish that an AI answer created the shortlist or caused the purchase.
The main survey includes nearly 4,000 buyers of software, services and physical goods, with purchases of at least US$25,000 in the prior two years. A separate companion survey has 766 respondents. This is a broader B2B sample, not a SaaS-only experiment.
Two Different Observations in 6sense’s 2025 Study
6sense, Buyer Experience Report 2025 · https://6sense.com/science-of-b2b/buyer-experience-report-2025/
Separate reported outcomes, not complementary shares. The detailed first-contact result is 77.0% (n=3,744); the overview rounds related preference findings to about 80%. Contact order is associated with prior preference, not proof that speaking first causes a win.
Treat AI citations as one observable part of discovery. Prior product experience, colleagues, reviews and vendor interactions can all affect consideration. Your own CRM and buyer interviews are needed to understand which influences mattered in a particular opportunity.
- →Use sales questions to map discovery, comparison and evaluation needs.
- →Maintain useful product and category content with sources and limitations.
- →Track AI mentions and citations with explicit sampling rules.
- →Keep observed referrals, buyer-reported influence and analyst inference separate.
What to do
Invest in accurate product information and credible relationships. Use visibility measurements to find questions to improve, without treating a citation as a place on a real buyer’s shortlist.
CH.06
A 90-Day Working Plan
This is Tenten’s suggested planning sequence. Adapt it to capacity, access and review requirements. The deliverables are a baseline, clearer content and a measurement record; AI citations and revenue are outcomes to observe, not guaranteed milestones.
Weeks 1 to 2
Define the questions and baseline
Choose questions from actual buyer research. A 100–300-prompt bank is an optional planning range, not a statistical minimum. Record the product interface, language, response status, mentions and cited URLs; API results must not be treated as the consumer interface.
Weeks 3 to 4
Resolve verified access and content gaps
Check indexability, readable public content and applicable structured data. Review each provider’s search and training controls separately. llms.txt is optional and Google Search does not use it. FAQ markup is not a prerequisite for AI visibility; Google retired FAQ rich results in 2026.
Weeks 5 to 8
Improve comparison and evaluation content
Resolve the most consequential unanswered questions. Link claims to sources, state product limits and secure approval for customer evidence. Reuse useful existing pages instead of publishing near-duplicate pages for every prompt.
Weeks 9 to 12
Review observed changes and next priorities
Compare like-for-like samples, note platform changes and distinguish missing data from zero. Review recognized AI referrals and qualified inquiries with consent and deduplication. A before/after change is observational unless the study design supports a causal claim.
| Work item | Completion evidence | Boundary |
|---|---|---|
| Question sample | Documented questions, platforms and collection conditions | Does not represent every buyer or answer |
| Access and identity | Verified blocking issues resolved and public facts consistent | Eligibility is not a citation guarantee |
| Comparison content | Material buyer questions answered with dated sources | Custom terms may still need discussion |
| Evaluation evidence | Approved evidence with assumptions and limitations | Estimated returns are not realized returns |
| Measurement record | Visibility, recognized referrals and qualified inquiries separated | No automatic causal revenue attribution |
Tenten’s delivery checklist for the proposed work period; adjust it to the agreed scope
The next decision
Choose one important buyer question your site does not answer well. Fix the missing information, verify the main reading path and record what changed. Expand only when the next gap is clear.
Sources & methodology
Updated October 5, 2026. Primary sources: Forrester’s January 21 and February 12, 2026 public summaries of its 2025 buyer survey (https://www.forrester.com/blogs/state-of-business-buying-2026/ ; https://www.forrester.com/blogs/zero-click-is-only-half-the-ai-story/); 2X’s April 7, 2026 release (70-company proprietary audit); and 6sense’s 2025 Buyer Experience Report (main survey nearly 4,000; separate companion n=766). The latter two source URLs appear with their charts. Samples and questions differ; they do not establish that AI causes shortlist entry or sales. Paywalled reports and raw respondent data were not accessed. Google guidance: https://developers.google.com/search/docs/appearance/ai-features . The 90-day sequence, prompt range and editorial tables are Tenten planning suggestions, not experimentally validated thresholds or outcome promises.
© 2026 Tenten GEO · Free to cite, please credit the source.
Discuss your AI search priorities
Request a 30-minute benchmark interpretationSend a booking request. We’ll email you to confirm a time within 24 hours.