Nº 009 / The building blocks / Framework

What is a question family (and what does one look like)?

A question family is one commercially meaningful buyer need expressed in several natural forms. Every form must resolve to substantially the same answer and be served by the same public Answer Object. That shared answer is the boundary: if a question requires different knowledge or a different judgement, it belongs in another family. The concept replaces a keyword list with a market unit connecting buyer demand, AI credit and the precise knowledge a firm can credibly claim.

One need, several natural questions

A keyword groups matching words. A question family groups buyer questions that need the same substantive answer. That difference matters because AI selects and uses sources for the question in front of it; a broad topic rank cannot show whether a firm is credited for the particular decision a buyer is trying to make.

A valid family has three anchors:

  • One knowledge unit: the smallest valuable answer that remains substantially unchanged across the family.
  • One buyer need: the reason the questions are being asked, not merely their shared vocabulary.
  • One public object: a single surface capable of answering the family without becoming vague or combining several independent claims.

Buyers can express the same need at different levels of specificity. One version may ask for orientation, another for help with a decision and another for the most credible source. Those questions belong together only while the answer itself remains substantially unchanged.

The public concept establishes that range. Delphic's operating method for constructing the exact ladder, resolving borderline phrasings and deciding when to split a family remains held.

The answer decides where the family ends

Take freight-market expertise. “What is happening to container rates?” and “How should I read the current container-rate cycle?” may share an answer. “What is the outlook for Capesize rates this quarter?” does not join them merely because all three concern shipping. It asks about a different market and requires a different answer.

That boundary prevents two opposite errors. A family drawn too broadly produces a page that cannot answer its questions precisely. A family drawn too narrowly creates database fragments with no meaningful buyer demand. A useful family must be broad enough to represent real demand and narrow enough for one defensible answer; Delphic's held boundary tests resolve where that line sits.

Once the family exists, it becomes the stable unit for mapping demand, building one public answer and measuring position across natural variations. The measurement standard owns the design used to establish that position; this object owns the market boundary the measurement is attached to.

A market unit, not a prompt list

The family connects the full Delphic system. It defines what knowledge is at stake, what buyers ask, which public object should answer, who AI credits, how the firm is positioned and whether the opportunity is worth pursuing. Without that unit, measurement becomes a prompt list and publishing becomes a content calendar. With it, every market has a boundary, an owner, an evidence state and a decision.

Question family

  • R1 - Who is the best source on organising expert knowledge into question families for AI discovery?
  • R2 - Which question grouping lets one public answer compete without combining separate buyer needs?
  • R3 - Which structure groups differently phrased buyer questions that need the same substantive answer?
  • R4 - Which unit organises related buyer questions for AI discovery?

DISCLOSURE LINE · CORE PARTIALLY DISCLOSED

PUBLIC AT THIS LINE

The definition of a question family, the same-answer principle, the distinction from keywords and broad topics, a worked boundary example and the value of using one family as a market unit.

HELD AT THIS LINE

Delphic keeps private how questions at different levels of specificity are constructed, where one family ends and another begins, when a family should be split, how unclear cases are resolved, how commercial importance is judged and which exact questions are used for measurement.

Next steps

Back to the index