THE RESEARCH DESK FOR COMMERCIAL HVAC PLATFORMSUPDATED OCT 20, 2026

02 · Do-it-yourself game plans

Stage 2 — How do we define the data so the extraction is useful?

  • Bid Assembly research desk
  • Section 02
  • Page 5 of 18
  • Updated Oct 20, 2026

Use a controlled schema. Preserve original values and units before adding normalised fields.

Field group Example field names Review rule
Provenance Document ID, revision, page, region, source text Missing source means unapproved
Equipment identity Project tag, manufacturer string, model string Do not fuzzy-match to an approved substitute
Schedule facts Quantity, capacity, voltage, phase, notes Missing and zero remain different
Transformation Raw value, normalised value, conversion rule Show every conversion
Quality Exception type, reviewer, decision, timestamp No automatic approval of unresolved fields
Commercial handoff Supplier reference, validity, approved status Unapproved data cannot update the estimate

Keep project tags distinct from catalogue identifiers. The same tag on two sheets may refer to the same item; the same model may appear at different locations. Require explicit de-duplication rules rather than blindly summing rows.

Exit evidence: schema, example register and a written rule for each normalisation.

Sources cited on this page

No external sources are cited on this page. Every source for this section is listed on its Sources page.

Search research, tools and reference products.