Three ways of working, and the reasoning behind each.

The answer engine audit

Three weeks, fixed fee, no writing. This is where nearly every relationship begins, and a fair number of them pause here quite deliberately while the findings get argued about internally.

We build the question set from your sales calls, your support tickets, and the searches that already convert, then run it across the major engines and record who gets named in each answer. The output is the coverage picture, a ranked list of the gaps that sit closest to a purchase decision, and an assessment of why the pages you already have are not being retrieved.

That last part is usually the uncomfortable one. Most companies discover they have written plenty about the topic and buried the answer four hundred words down, underneath a preamble nobody asked for.

The audit is deliberately sold on its own. If the finding is that your existing pages need restructuring rather than replacing, that is a cheaper answer and you should hear it before you commission anything.

Editorial retainers

A standing arrangement, usually two pieces a month, occasionally three when the subject matter is already well documented internally.

The value of a retainer is not volume. It is that we stop starting from zero. By the second month we know your product, your objections, and which of your engineers explains things well under questioning. Reporting gets faster because the context is loaded, and the pieces begin to reference one another, which matters more than it sounds. Corroboration across your own body of work is one of the signals an engine weighs when it decides whether a claim is safe to repeat.

Retainers include a quarterly re-run of the question set. We look at which prompts you now appear in, which you lost, and what the next quarter should target. Movement is reported honestly, including the quarters where a piece did not land.

Original research

A practitioner survey, a benchmark run under a common workload, or an analysis of a dataset you already hold and have never published.

These take eight to twelve weeks and most of the effort lands before drafting. We design the instrument or the test, run it, analyse the results, and only then decide what the piece argues. Occasionally the data refuses to support the thesis anyone hoped for. When that happens we say so and write the piece the evidence allows, which in our experience earns more credit than the original thesis would have.

Research is the format with the strongest pull on citations, for a straightforward reason. A model can paraphrase an opinion from anywhere, but a figure with a documented method has exactly one source, and naming it is the only safe way to use it.

Output is the written piece, a methodology appendix, and the underlying data in a form a sceptical reader can inspect.

Turning citations into conversations

Included in retainers, available alongside an audit, never sold as a standalone growth service.

Readers arriving from an answer engine have had the basics settled already, so a generic demo request wastes the position you paid for. We match the next step to the question the reader arrived on, write the page that receives them, and where it helps, write the follow-up sequence that references the piece rather than pretending the visit never happened.

We do not run your ad accounts, build your CRM, or promise attributed revenue. Answer engines strip referrer data often enough that clean attribution is not currently available to anyone honest.