When buyers use AI to build a shortlist, can your sales offer survive the comparison?

Translate the shopping story into a sales question

In a LinkedIn post shared with us, Olivia Moore described watching an AI shopping agent compare winter coats and select a retailer. Her reported test highlighted access failures, stock, price and pickup. Those observations are specific to that shopping journey, which we have not independently reproduced.

For a service business, the equivalent question is not whether an assistant can buy a consulting engagement autonomously. It is whether the buyer can discover the service, understand its boundaries, verify relevant evidence and reach the right next conversation.

This is an application of the idea to sales, not proof that every B2B buyer uses agents or that ecommerce ranking behavior carries across to services.

Publish the facts a buyer needs to qualify you

Make the buyer, problem and deliverable explicit. Explain the starting price or what determines a quote, the usual engagement structure, required customer input, exclusions and the next step. If timing depends on capacity, say so instead of presenting a permanent delivery promise.

Attach evidence to the claim it supports. A case study should distinguish your role, scope, period and measured result. An employer's brand name alone does not prove you can deliver every outcome associated with it.

For an illustrative CRM follow-up service, useful facts include which platform is supported, whether implementation is included, how many workflows are covered, who approves messages and what handoff the client receives. “AI-powered growth transformation” answers none of those questions.

Test the shortlist without leading the assistant

Use a fresh session and realistic buyer constraints. Run a non-branded prompt first. A separate branded query can verify facts but cannot show that the assistant would have found you independently.

Prompt to try

Find providers for a small B2B team that needs to repair lead follow-up in its existing CRM. Compare scope, evidence, pricing transparency, implementation responsibilities and the next step. Link to public sources. Mark unknowns as unknown. Do not contact providers or submit forms.

Repeat across a few sessions and record the product used, date, region, exact wording and cited pages. If your offer is absent, investigate before rewriting everything. Your page may be new, hard to retrieve or a poor fit for the query.

Ask which source supports a capability or limitation. Treat the assistant's stated reason for ranking a provider as a hypothesis, not an exact account of its internal scoring.

Make comparison fair and useful

Create a page or sales asset that explains when coaching, an audit, implementation or an ongoing engagement fits. Compare observable scope and operating requirements. Give buyers reasons to choose another path when yours is unsuitable.

Avoid fake “independent” rankings, fabricated reviews and testimonials that imply work you did not perform. Buyers need evidence they can check, not a larger volume of unsupported claims.

Google's current AI search guidance retains conventional SEO foundations rather than requiring a special AI ranking format. Clear public pages and accurate evidence are sensible work even when an assistant never cites them.

Design the handoff to a person

The next step should say what happens after the buyer clicks: a fit conversation, a brief review, an application or a paid session. Make the required information proportionate to that step. A forced account, unexplained form or unclear meeting purpose creates avoidable friction.

When a prospect arrives with an AI-generated summary, ask what they understood, which alternatives they considered and what remains uncertain. Correct material mistakes directly. Do not pretend a misleading summary is accurate because it makes the sale easier.

A useful reply might be: “The review includes the diagnosis and prioritized recommendations. Implementation is separate. If the main problem is slow lead routing, the first conversation should confirm your current process and who owns it.” That protects the relationship and improves qualification.

Bring a better question to the discovery call

Ask the buyer how they built the shortlist and which information influenced the decision. Record their answer in an existing source or research field; avoid treating self-reported attribution as perfect tracking.

Then move from the tool to the problem: what breaks today, who experiences it, what has been tried, and what would count as improvement? An assistant can prepare a comparison, but it cannot supply the buyer's actual internal priorities unless someone gives them accurately.

For practice, use the AI playbook to prepare questions and the sales practice space to rehearse a clear next-step conversation.

Keep a small evidence log

Use one row for each test: buyer scenario, assistant, date, recommendation, cited page, accurate facts, missing facts, visible blocks and whether the buyer can reach the intended next step. Keep branded and non-branded tests separate.

Track meaningful sales outcomes separately: relevant enquiries, qualified conversations, agreed next steps and conversion. A mention in an answer is not a lead. A new lead is not evidence that one content edit caused a sale.

Your next action is to repair one ambiguity in a real offer and test whether both an assistant and a prospective customer can explain it more accurately. For the broader technical audit, read Waymaker Agency's AI discoverability guide.

Sources and editorial note

Google Search Central: AI features and your website

Inspiration: a reader-supplied LinkedIn excerpt by Olivia Moore (profile). The original post permalink was not supplied. Sources checked October 20, 2026. The shopping observations are attributed to Moore; the sales exercises are our editorial recommendations, not measured client results.