Most data quality programs die in the first six months for the same reason: they open with a remediation budget request before anyone outside the data team believes there is a problem worth paying to fix. The pattern is predictable. A new data leader inherits a quality complaint, scopes a multi-quarter cleanup, asks for headcount and tooling, and gets handed a cost-cutting target instead. The fix is to invert the sequence. Measure first, publish what you find, let the dollars speak, then ask for the budget that the evidence has already justified. Ninety days is enough to make that case if the work is sequenced tightly.

Days 1–30: measurement, not remediation

The first month is forensic. The deliverable is a quality scorecard for the ten to fifteen tables that drive revenue, cash, or board reporting — not a comprehensive catalog, not a maturity assessment, not a tool selection. Pick the tables a CFO would name if asked which numbers they need to be right.

  • Week 1. Interview the three executives whose decisions ride on these tables. Ask each one for the last time they doubted a number, the last time they were embarrassed by one, and what they do today to compensate. Capture verbatim. These quotes become the anchor of every later conversation about budget.
  • Week 2. Instrument freshness, completeness, and uniqueness on every selected table. Open-source dbt tests, Great Expectations, or Soda Core all do this in a week if you scope tightly. Do not buy a vendor tool yet — instrumentation cost should be measured in engineering hours, not annual contracts.
  • Week 3. Run reconciliation on the top three reported metrics against their source-of-record system: revenue against the GL, pipeline against the CRM, headcount against HRIS. Document every variance over one percent. Do not fix anything yet.
  • Week 4. Publish a one-page scorecard. Each row is a table, each column is a quality dimension, the cells are red/amber/green with the absolute number underneath. Attach the executive quotes. Send it to the CFO and the COO. Do not ask for budget. Do not propose a project. The scorecard is the entire artifact.

The scorecard’s job is to make the problem legible to people who do not read pull requests. Done well, it produces inbound demand for the program before you have to sell it.

Days 31–60: contain the bleeding, prove the model

Month two converts the scorecard into a small number of visible wins. The principle is that any remediation work in this window must show ROI in the same quarter, not at the end of a roadmap. Pick the two or three findings from month one with the cleanest dollar attribution and fix only those.

  • Week 5. Stand up monitoring on the priority tables so quality regressions surface within hours, not at month-end close. Hook the alerts into the same channel finance already uses for close issues. Visibility there beats visibility on a data team dashboard nobody opens.
  • Week 6. Fix the highest-dollar reconciliation gap. If revenue ties to the GL within $50K instead of $400K, that delta is your headline. Document the before-and-after with screenshots and a one-paragraph note. Do not editorialize.
  • Week 7. Establish a data contract on one upstream system — typically the CRM, since it is usually the loudest source of downstream pain. The contract is two pages: required fields, allowed values, owner, on-call escalation. Get the system owner to sign it. The signature is the deliverable.
  • Week 8. Publish a second scorecard. Same format, same audience, with the deltas highlighted. Add a sidebar quantifying time recovered by finance — usually two to four days per close cycle once close-time variances stop. That number is the first dollar figure you put in front of the CFO.

By the end of month two the program has produced a measurable change in close speed, a cleanly attributed revenue reconciliation improvement, and a signed contract with a peer system owner. None of these required new headcount or new tooling. That is the point: it proves the operating model works before any spend is requested.

Days 61–90: institutionalize and price the next phase

Month three converts a working pattern into a fundable program. The temptation here is to over-engineer the proposal. Resist it. The artifact at day ninety is not a strategy document; it is a budget request grounded in two months of published evidence.

  • Week 9. Catalog the remaining findings from the month-one scorecard and assign each one a dollar impact and a fix cost. Findings without a credible dollar impact get dropped, not deferred. The list should fit on one page.
  • Week 10. Define the operating cadence in writing: who owns the scorecard, who runs the weekly review, who escalates a contract violation, who signs off on schema changes. This is the governance footprint of the program. Keep it under one full-time equivalent of accountable time across all roles.
  • Week 11. Build the budget. Include the headcount, tooling, and contract owner time the next phase requires, with each line tied to a specific finding from the catalog. Show the cost-of-inaction calculation alongside it: time lost in close, revenue at risk from contract gaps, audit exposure on the controls that failed reconciliation.
  • Week 12. Present to the executive team. The presentation is the third scorecard, the catalog, and the budget — three artifacts, each one already familiar from earlier in the program. There is nothing in the room the audience has not seen before, which is exactly why it gets approved.

A program funded this way differs from one sold up front. The executives writing the check have already watched it work for a quarter on problems they complained about, against numbers they use. The conversation stops being about belief and becomes about scope.

Measurement before remediation. Visible wins before structural change. Dollars before asks. Skip any of those and the program reverts to the failure mode it was meant to escape.