Skip to content

Generative Engine Optimization

Day 148: Commission the Question, Not Another Comparison Page

A marketing team has budget for one substantial asset. It can publish another comparison page using facts already available, or commission research into a question the category cannot yet answer well.

Original research sounds more defensible. It can also become an expensive content format in search of a purpose.

The commissioning test should come before choosing a method: would the business still pay to learn the answer if competitors could read the findings and no AI system ever cited them?

If the answer is no, the proposed study is probably distribution theatre. If the answer is yes, the business may be funding knowledge that improves a real decision, with publication as a separate choice.

Day 147: How Google May Expand One Buying Question

Google says AI Overviews and AI Mode may use “query fan-out”: issuing multiple related searches across subtopics and data sources while developing a response.[1] Its generative AI Search guidance defines fan-out as concurrent related queries generated to request more information and fetch additional relevant results.[2]

That product behaviour creates an important boundary for GEO work. The question a buyer types, the related searches Google may generate, and the answer and links it eventually displays are three different objects.

Google does not expose the exact fan-out path behind every response. A citation list is not a transcript of hidden searches, and a plausible subtopic is not proof that Google issued a particular query. But the documented possibility is enough to challenge a familiar planning assumption: one visible query does not necessarily describe the whole retrieval task.

Day 146: Every Pricing Model Has an Expensive Month

The following suppliers, tariffs and workloads are completely fictional. They are a pricing-risk illustration, not a market benchmark, client result, named-vendor comparison or ZSA price promise.

In this example, £0.012 per action looks smaller than £80 per user. But the useful procurement question is not which number looks cheaper. It is which pricing model creates the expensive month your budget can tolerate.

AI-assisted research may place unlike published price units beside each other during supplier discovery. No answer-engine output was tested for this post. The worksheet begins after discovery: it converts two fictional tariffs into budget exposure for one team.

Day 145: Red-Team the Claim, Not Just the Translation

Consider this fictional scenario: a regional sales lead pauses a launch after reading one sentence on the new local-language page. No real company, translation project, customer conversation, or market result is being reported.

The approved source says:

We can support rollout in selected European markets through approved delivery partners, subject to integration and data-location review.

The local page now says, when translated back into English:

Available across Europe with local implementation.

The sentence is shorter. It may sound more natural. It is also a different commercial promise.

“Selected markets” became “across Europe”. Partner delivery became local implementation. Two review conditions disappeared. A buyer, salesperson, search result, or answer-led system encountering the local page no longer receives the limits that made the source claim defensible.

For CMOs, Marketing Directors and founders, localisation quality cannot stop at fluency. The harder test is whether each language version preserves the claim the business is prepared to sell, deliver and defend.

Day 144: Decide What an AI Crawler May Do Before You Block It

“Block AI bots” sounds like one policy decision.

It can conceal several different business decisions:

  • Should this page be eligible for automated search discovery?
  • May its content be considered for model training?
  • May it support prompt-time grounding in a provider's products?
  • What happens when a user explicitly asks a product to visit it?

Those purposes are not interchangeable. Current first-party documentation from OpenAI and Google assigns different controls and semantics to some of them.[1][2][3]

A blanket allow or deny can therefore express a policy the business never meant to choose.

For CMOs, Marketing Directors, and founders, the question should come before the directive: which use of this public content do we want to permit, restrict, or investigate?

Day 143: A Brand Name Is Not a Unique Identifier

Two unrelated companies can trade under the same short name.

One may be a UK analytics consultancy. The other may be a US software vendor. Their domains, buyers, offers and legal identities are different. Yet a visibility report can search the shared name, find an answer that mentions it, and mark the result as positive.

That mention cannot be credited to the target until the report establishes which company the answer described.

This is not a hypothetical claim that a named answer engine has confused two real businesses. It is a practical failure mode any CMO, Marketing Director or founder can test: a string match can succeed while the commercial identity is wrong.

Before assigning target-company credit, sentiment or recommendation position, resolve the entity while retaining the observation in the study record.

Day 142: A Paid Directory Listing Is Not an AI Recommendation

Imagine a fictional directory proposal. It offers your company a paid profile in a category relevant to your buyers, prominent placement on a comparison page, and “AI visibility”.

The first two items can be specified as inventory. The third is a contingent claim.

The directory can sell space on its own property. It can define where a profile appears, how long it remains live, which audience can encounter it and whether an enquiry route is included. Buying that placement does not give the publisher contractual control over whether an independent answer engine retrieves the page, cites the directory, mentions your company or recommends it.

That does not make the listing worthless. It makes the purchase legible.

For CMOs, Marketing Directors and founders, the useful question is: which surface, audience and commercial route does the contract buy—and how does that inventory compare with the alternatives?

Day 141: Winning the Shortlist Is Not a Migration Plan

Consider a fictional company evaluating a multi-market customer-feedback programme. An AI-assisted comparison identifies a credible supplier. The product supports the required languages, connects to the relevant data sources, and gives regional teams a common place to analyse recurring concerns. The answer is accurate enough to justify deeper evaluation. The supplier remains suitable after the Marketing Director checks the underlying material.

The destination is clear.

The route from today's operation is not.

The company currently combines CRM exports, agency summaries, local spreadsheets and a monthly analyst review. Replacing that arrangement is not one purchase. It is a sequence: obtain access, agree shared definitions, preserve useful history, run old and new processes together, move planning meetings onto the new output, and decide when the former route can stop.

A comparison of steady-state capabilities cannot establish whether that sequence is feasible. For CMOs, Marketing Directors and founders, this is the commercial distinction that matters: a suitable destination does not automatically come with a viable transition.

Day 140: Sometimes the Best AI Answer Is ‘It Depends’

A buyer asks, “Which supplier is best for our AI visibility programme?”

The question sounds complete. It is not.

Best for a company that needs a one-off diagnosis or a managed monitoring service? Best for a team with authority to change its website or one that needs board evidence first? Best for a single market or a multilingual launch? Best when the problem is category confusion, product misinformation, weak comparison language, or inaccessible source material?

A confident list of providers may be fluent while answering a decision the buyer has not actually defined.

For CMOs, Marketing Directors, and founders, the aim should not be to make every buyer question produce an immediate brand answer. Some questions should produce a branch. Some should expose a missing input. Some should end with a better follow-up question.

“It depends” is not a weak answer when the decision genuinely depends on facts the buyer has not supplied. It is the start of a more useful one.

Day 139: Stop Sending Product Gaps to the Content Team

A company is absent from an AI-generated shortlist for a commercially important buyer question. The familiar response is immediate:

We need a page for that.

Sometimes it does. The company may offer exactly what the buyer needs but explain it badly in public. A clearer page can correct a genuine communication problem.

Sometimes the missing page is doing the business a favour.

The product may not support the requested capability. The evidence may not exist yet. The offer may deliberately exclude that use case. Or the query may describe demand the company has no reason to pursue. Publishing around those gaps would not improve the truth available to a buyer. It would create a more persuasive version of the wrong answer.

This is where GEO becomes a question of decision rights rather than content volume. Marketing can communicate an approved commercial truth. It cannot approve a capability, redefine the offer, or choose a market on behalf of the functions that own those decisions.