Read. Choose. Prove. Spread.

We start with the buyer’s side: the questions they ask, the alternatives they actually weigh — often manual work, conventional automation or waiting — and the evidence you already hold, from test results and pilot data to customer feedback.

We also sample what AI assistants say about you and your category, for a fixed set of buyer questions run several times on agreed platforms. We record the answers verbatim and keep three things apart: sources the assistant cited, sources we infer may have influenced it, and provenance nobody outside the provider can see. AI descriptions can contain errors and unsupported statements, so every answer is checked against your approved facts.

In parallel we check whether your site is readable: whether essential information is in the HTML a crawler receives, whether access is permitted at every layer, and whether your company details are consistent.

This is the positioning decision, and it is where most of the difficulty lives. The output is not a tagline. It is a decision about the customer, the problem, the alternative and the differentiation.

We hold candidate positions to three tests. Is it true — could a customer confirm it under questioning? Does it matter to the buyer — does it change a decision they actually make? Is it supported — by evidence an outsider could check, and clear about which alternative it beats?

A position does not have to be untrue of every competitor. A relevant advantage, or a combination of capabilities nobody else offers this buyer, is usually enough — and far easier to defend than artificial uniqueness. Likewise, a recognised category with clear differentiation is often more useful than a new one, because buyers already know how to budget for it.

Every important claim goes into a register with a status. Something shown under controlled conditions is not the same as something a customer has confirmed, and neither is the same as something on the roadmap. The register makes the difference visible to everyone who writes about the company.

Then the material gets built: the application page, the sales narrative, the buyer FAQ. Each states claims at the strength the register allows — with their conditions — and nothing more.

Occasionally this stage produces the answer nobody wants: the claim is not yet true. When that happens we say so. The work then becomes making it true, or stating it more modestly — which is a better use of a quarter than publicising something buyers will not believe.

Independent sources tend to carry more weight than self-description, with buyers and, as far as anyone can tell, with AI systems. So the plan aims to get the chosen position repeated accurately by partners, customers, publications and the catalogues that list your category.

Some of that is conventional: coverage, contributed writing, partner material. Some is not: publishing research or documentation that other people want to cite. We scope it to the relationships and resources you actually have.

What we will not do is buy it. Purchased mentions and syndicated filler are visible to anyone paying attention, and they are a liability with a delay on it.