Skip to content

2026

Day 159: Build the Demo Around the Moment the Buyer Needs

A buyer arrives with one task: see how the product handles an exception.

The usual product demonstration makes her earn that answer. First comes the company story, then the dashboard tour, then the standard workflow. The relevant sequence appears six minutes later—if she is still watching.

A more useful commission treats the demonstration as a set of task-specific moments. The recording still has a coherent narrative, but each meaningful segment also has a label and a destination. A researcher can start where her question starts.

This is an annotated storyboard for a fictional B2B product, not a published ZSA experiment or a report of observed search behaviour. The timestamps are invented positions.

Day 158: A Canonical Tag Is Not a Syndication Contract

A media partner offers to republish your research in full. The audience fits. The reach is attractive. The partner promises to add a canonical tag pointing back to the original.

Marketing approves the deal because the source URL appears protected.

That protection is weaker than the contract assumes.

Google describes rel="canonical" as a strong signal, not an instruction that must be obeyed. Its current troubleshooting guidance goes further for syndicated content: the canonical link element is not recommended as the way to avoid duplication by syndication partners, because the pages are often very different. Google says the most effective solution is for partners to block indexing of the syndicated copy.[1][2]

This turns a technical-looking checkbox into a commercial distribution decision.

Day 157: An AI Adoption Headline Is Not a B2B Channel Plan

Consider this invented survey statement: “Most enterprise knowledge workers use generative AI for work tasks.” It is not an actual survey finding, and no budget decision reported here took place. It is a test sentence for a common leap in channel planning: if AI use is widespread at work, should a B2B company move acquisition budget into generative engine optimization?

Not from this statement alone. It describes neither the specific buyers a company needs to reach nor the activities those buyers perform while choosing suppliers.

Day 156: Commission Genuine Customer Reviews, Not Synthetic Advocacy

Consider a fictional proposal for a UK consumer-facing review campaign. An agency promises to “scale public brand sentiment and review velocity for AI answer engine discovery”. Its plan includes social-media recruitment, free samples tied to five-star ratings, and consumer-style testimonials generated under simulated profiles.

This is not a delivered campaign or an enforcement finding. It is a procurement test: which conduct should a marketing leader reject, and what should they request instead?

The legal boundary matters. The Competition and Markets Authority’s CMA208 guidance explains the UK prohibition on submitting or commissioning fake reviews and concealed incentivised reviews. It also distinguishes those practices from encouraging accounts of genuine experiences without predetermining their content or sentiment.

The red lines below are ZSA’s proposed procurement standard, informed by that guidance. They are not legal advice.

Day 155: The Same Question Is Not the Same Research Task

A buyer can ask exactly the same supplier question in two ChatGPT chats and mean something different in each.

The difference may sit several turns earlier. One conversation may contain only a broad project brief. Another may already establish a deployment restriction that the final message does not repeat. Reading only the last question discards part of the request.

This matters for Generative Engine Optimization (GEO) because a prompt is not always an isolated unit of demand. Sometimes the useful object of study is the sequence that produced it.

Day 154: When Accurate AI Research Still Leaves a Buying Committee Stalled

Imagine a fictional buying committee choosing a customer data platform. This is not a client case or a test of any answer engine. We will stipulate that each executive's AI-assisted research is accurate, the supplier facts are fixed, and every person has asked for a comparison against explicit criteria.

The committee shares one requirement: whichever platform it selects must keep regulated customer data within the company's approved control model. Beyond that, its objectives conflict.

Marketing wants the fastest route to using unified customer data in campaigns. Finance wants predictable recurring expenditure and would rather avoid an additional implementation fee. Operations wants to protect limited engineering capacity as well as the agreed data controls.

Their research has not failed. It has helped each person compare the market against a legitimate brief. The stall begins when three sensible briefs produce three different preferences.

Day 153: Before Calling It a GEO Experiment, Defend the Comparison

Consider a hypothetical proposal. No agency study, answer-engine test or client result is being reported here.

The proposal calls itself a controlled GEO experiment. A company would revise an analytics-service page and a shared overview page, leave its advisory-service page unchanged, then compare recorded answers to questions about both services. If analytics visibility rises more, the proposal would attribute the difference to the edit.

The comparison has not yet earned that conclusion.

Leaving the advisory page unchanged does not establish that advisory answers were unaffected. If those answers consult the edited overview page, the nominal comparison may also be exposed to changed material. If they do not, that pathway may not matter. The study must state and investigate the exposure assumption rather than treating “we did not edit this page” as proof of no intervention.

CMOs, Marketing Directors and founders should ask whether the design supports the causal sentence beneath its chart.

Day 152: Run the Channel-Loss Tabletop Before Buying More Reach

Imagine your strongest answer-led discovery route sends no qualified introductions for a quarter.

The cause is deliberately unspecified. The platform may change how it assembles answers, buyers may change where they research, or your category may appear less often. The exercise is not a forecast. It is a tabletop question for the next budget decision:

Which routes could still bring a relevant prospect into a commercially useful conversation?

Counting logos in a channel plan will not answer it. Five platforms can still behave like one point of commercial dependence if they rely on the same route into the market. Resilience comes from usable routes with different failure dependencies, not platform count alone.

Day 151: Define the Paid Intervention After the Answer

A prospective buyer asks for guidance on a difficult business problem. An AI response explains the common options, gives them a checklist and links to useful sources. They arrive at a consultancy site already understanding the basics.

That does not prove the consultancy has been replaced. It does expose a weak offer if the paid proposition is only “we know the explanation”.

Google describes AI Mode as capable of answering questions that might previously have required multiple searches, including comparisons, with follow-up questions and links for further exploration.[1] That is a documented Google capability, not evidence that every buyer receives a sufficient answer, trusts it or stops buying advice.

The commercial design question is narrower: after the general explanation has been supplied, what is the buyer paying the provider to judge, change or deliver?

Day 150: A Supplier Page Is Not an Instruction to Your Research Agent

A supplier is allowed to explain its offer. It is not allowed to redefine the buyer's research task.

That boundary becomes important when a CMO, Marketing Director or founder delegates supplier research to an AI agent. The agent may read product pages, documentation, comparison material and third-party sources. Those pages contain evidence the buyer wants assessed. They may also contain language that tries to redirect the agent, change its priorities or obtain information the buyer never meant to disclose.

The commercial risk is not simply that the answer contains a bad fact. It is that an external source starts acting like an instruction from the buyer.