Service 02 — trace one number, account for every copy.
Data audits
We follow a critical figure from the system that made it to the person who acts on it. Every extract, transform, cache, spreadsheet, definition, and owner goes on the record. The result is a practical answer to what can be trusted, what costs too much, and what should disappear.
The work
An audit begins with one consequential number or dataset. Narrowing the question lets us inspect the real path deeply enough to expose the patterns affecting the rest of the platform.
01
Name the number, who reads it, what action it changes, and the consequence of it being stale, incomplete, or differently defined.
02
Walk backward through dashboards, semantic layers, marts, warehouse tables, stages, extracts, APIs, and source systems. Each copy gets a purpose, owner, freshness, and cost.
03
Compare definitions, grains, joins, keys, null behavior, row counts, freshness, access, and deletion behavior against the source and the decision it is meant to support.
04
Separate necessary boundaries from accidental duplication, then quantify storage, compute, egress, reconciliation effort, security surface, and decision risk.
What you leave with
Tangible, not theatrical
A record of where the data lives, why each copy exists, who owns it, how fresh it is, and which consumers depend on it.
A defensible path from origin to decision, with definitions, transformations, and boundaries written plainly enough to challenge.
The operational, financial, compliance, and decision consequences attached to each unnecessary or uncontrolled copy.
A prioritized plan to consolidate, retire, repair, or contract the highest-cost disagreements without breaking downstream work.
The xerocopy difference
A conventional audit often produces a maturity score and a longer shopping list. xerocopy produces a shorter list of truths. We ask whether every copy can defend its existence and whether a specific person can answer for it.
The recommendation is allowed to be deletion. A table can be retired, a pipeline replaced by a view, a metric definition narrowed, or an owner made explicit. The goal is not to make the diagram more impressive. It is to make disagreement more expensive to introduce.
If five copies answer the same question, you do not have redundancy. You have five chances to be confidently wrong.
Related work
Follow the problem
Start here
Send a paragraph. We will tell you whether this service fits, what a small first engagement could answer, and when we are not the right shop.
hello@xerocopy.sh →