Why provenance is a delivery practice, not a deliverable

Most consultancies treat provenance as an afterthought — a references section appended to the final report, assembled during the last week of the engagement. We treat it as a practice that runs from day one.

The distinction matters. When provenance is an afterthought, the research team makes decisions without documenting why. Claims accumulate without source attribution. By the time someone asks "where did this number come from?", the answer is buried in a Slack thread or a half-remembered conversation.

When provenance is a practice, every claim is tagged to a source at the moment it enters the deliverable. The source tier (A through E) and confidence rating are assigned immediately. If a claim cannot be sourced, it goes into the open-questions register — not into the findings section.

What this looks like in practice

A researcher at Zhianrui does not write a finding and then look for a source to support it. The workflow is inverted: a source is encountered, evaluated, and classified; then a finding is drafted from the source. The deliverable is built source-up, not claim-down.

This is slower in the first week and faster in every subsequent week. The open-questions register fills gradually instead of erupting at the end. The reviewer reads a deliverable where every claim has a visible evidence trail — and where gaps are explicitly marked.

Why clients should care

If you are making a build decision, a regulatory filing, or a fundraising pitch based on a research deliverable, you need to know which findings are solid and which are provisional. A deliverable without visible provenance forces you to trust the author's judgement uniformly. A deliverable with provenance lets you calibrate your confidence per claim.

That calibration is the difference between a well-informed decision and an optimistic one.