01
Transfers
The system
Question mapping, gain scoring, entity work, nightly sampling. The method is category-independent.
Identical on day one, everywhere.
Industries
The categories where we already know the questions
Answer engineering depends on knowing what buyers actually ask, and that knowledge is category-specific. These are the industries where we hold a live prompt panel, a competitive map and clients already being named.
6
categories with an active nightly prompt panel
2,000+
category questions mapped across those panels
Day 1
we start from an existing map, not a blank page
Why category knowledge is the whole job
I
The question space is the strategy
Two companies in different industries need entirely different answer surfaces. Knowing the fifty questions that decide purchases in a category is worth more than any tactic.
II
Buying behaviour dictates the engine mix
Enterprise software buyers research in ChatGPT and Reddit. Consumers land in AI Overviews. Students ask comparative, fee-shaped questions. The same budget goes to different places.
III
Regulated categories change the rules
Finance, health and education carry accuracy obligations that shape what can be published and how a model must be allowed to quote it. We would rather say this upfront than discover it in month three.
Honestly
If we have never worked in your category, we will tell you what the first month costs to learn it, or tell you we are the wrong firm.
A panel takes about four weeks to build from scratch. Where one already exists, you skip that and start from the map.
Existing panels
Four weeks to build new
Straight answers on fit
Where we hold live panels
Open a category to see how buyers there actually search.
Buyers compare programmes on cost, eligibility, placement rate and credibility, and they overwhelmingly ask comparative questions rather than branded ones. Answer engines are already the first stop for prospective students, and most institutions are invisible in them. Futurense went from zero to 40K+ monthly organic clicks and is now named by all five engines on its core question.
Comparative queries
Placement data
Programme × outcome pages
The highest-value questions are the ones vendors avoid: alternatives to a competitor, real pricing, where the product falls short. Buyers research in ChatGPT and verify on Reddit. Integration and comparison pages carry disproportionate pipeline, and honest ones outperform defensive ones consistently.
Alternatives pages
Integration matrix
Reddit verification
Purchase decisions hinge on trust signals that live outside your site: reviews, forum threads, and whether an engine is willing to vouch for you. CARS24 holds five out of five engines on its core buying question, defended nightly, because the off-site consensus was worked as hard as the on-site content.
Trust signals
Off-site consensus
Location × category scale
Engines are cautious about naming financial brands and lean heavily on regulated sources and third-party corroboration. Entity strength and factual precision matter more here than anywhere else, and factual drift in a model answer is a compliance problem, not just a marketing one.
Entity precision
Drift monitoring
Regulated sourcing
Quality thresholds are the highest of any category. Named clinical reviewers, cited primary sources and unambiguous provenance are prerequisites, not enhancements. Done properly this is a durable advantage, because most competitors cannot clear the bar.
Clinical review
Primary sourcing
Provenance schema
Buyers are choosing people, so Person entities, published thinking and third-party mention carry more weight than page count. A small library of genuinely expert content plus strong practitioner entities beats a large blog, reliably.
Person entities
Published thinking
Practitioner positioning
What transfers, what does not
Being clear about this saves everyone a wasted quarter.
02
Transfers
The instrumentation
The prompt harness, the warehouse, the reporting. Built once, pointed anywhere.
No ramp-up cost.
03
Does Not
The question panel
Every category needs its own, built from its own buyers. Roughly four weeks of work.
Priced separately and honestly.
04
Does Not
Editorial credibility
We can structure and interview, but the expertise has to come from your people or specialists we recruit.
We will not fake domain authority.
Starting in a new category
- 01 Fit conversation What you sell, who decides, and whether the answer layer is even where your buyers are. Day 1
- 02 Question harvest Search data, sales calls, support tickets, forums, and direct prompting of the engines. Week 1-2
- 03 Panel build The prompt set, scored by value and contest, running nightly. Week 3-4
- 04 Baseline and map Where you stand, who holds what, and the first thirty questions to take. Week 4
- 05 Execute The standard system, now aimed at a map specific to your category. Month 2+
Next
Tell us your category
If we hold a panel there, you will see real readings on your questions within a week. If we do not, we will tell you what it takes to build one, or that we are the wrong firm.