Digital marketing due diligence fails when a team tries to pick one dashboard and call it the truth. The platform reports a conversion. Analytics reports an event. The CRM reports a record. Finance reports revenue. None earns the right to overwrite the others.
This is the working worksheet for the marketing slice of a pre-close review or a post-close operating reset. It requests the evidence, preserves the definitions, reconciles the differences, and leaves a decision record someone else can inspect. For the broader operating-partner frame, start with private equity portfolio operations for marketing. For the leadership engagement that can install a repair, see private equity portfolio marketing.
A number can be useful before it is reconciled. It cannot be presented as proof before it is reconciled.
scope the question before asking for data
Decide whether this is pre-close diligence or a post-close operating review. The evidence often overlaps. The decision does not. Pre-close work asks what can invalidate an underwriting assumption. Post-close work asks which constraint gets fixed first.
Record the company or business unit, decision being informed, review window, fiscal-calendar rule, currency, expected sales-cycle lag, data owners, and access restrictions. Ask for read-only access or minimized exports. Preserve stable row IDs and dates. Do not ask for passwords or a warehouse dump because the request feels comprehensive.
request the evidence chain
Pull the source records before comparing any headline. This table is the minimum request, not a promise that every system will contain every field.
| Evidence | Minimum fields to preserve | What it tests |
|---|---|---|
| CRM opportunity and stage history | Opportunity and account IDs, stage-entry dates, amount history, source fields, owner, close outcome, merge/deletion status | Whether pipeline and lifecycle claims retain their meaning over time |
| Metric definitions | Sourced, influenced, qualified, accepted, won, and paid-conversion definitions with owner and effective date | Whether a comparison uses the same words in both periods |
| Finance or billing records | Contract/account key, booked and recognized revenue, credits/refunds, currency, period | Whether a revenue claim can be linked to an approved financial basis |
| Paid-media platforms | Spend, campaign IDs, conversion-action name, count setting, goals, attribution window, date basis, change history | Whether an optimization event is being mistaken for a business outcome |
| GA4 and web analytics | Landing page, source/medium, event name, key-event setting, consent/modeling note, event date | What was observed and what consent or implementation may leave outside the report |
| Search Console | Query, page, clicks, impressions, device/country, property host, export window | Nonbrand demand, page dependency, and canonical-host questions |
| Forecast snapshots | Snapshot date, forecast, actual, segment/channel split, assumption and definition version | Whether past forecasts survived without hindsight replacement |
| Vendor and handoff records | Contract, scope, spend, owner, term, lead-acceptance outcome, response time, handoff process | Channel dependency and the operating system around it |
Google Ads reporting needs its own caution. Its conversion columns can represent primary actions and modeled conversions, while its date basis can be ad interaction or conversion time. Record the action, goal mapping, count setting, attribution rule, date basis, and finalization delay before comparing it with CRM pipeline. Google Ads attribution paths are platform evidence, not a complete customer journey. See the Google Ads conversion definitions, attribution-model changes, and attribution-path scope.
GA4 needs the same restraint. With consent mode, observed analytics data can be affected by the implementation and storage choice. Preserve the consent/modeling note rather than treating observed sessions as the full population. Export needed event-level evidence early because retention can constrain later explorations. See Google's consent-mode guidance and GA4 retention documentation.
make the metric contract explicit
Write the contract before comparing channels or periods. A missing definition is a finding. It is not permission to choose the most favorable interpretation.
| Measure | Definition | Source of record | Date and attribution rule | Owner | Change history |
|---|---|---|---|---|---|
| Marketing-sourced pipeline | |||||
| Marketing-influenced pipeline | |||||
| Accepted opportunity | |||||
| Won revenue | |||||
| Paid conversion used for optimization | |||||
| Forecast accuracy |
Keep sourced and influenced pipeline separate. Name whether revenue means bookings, recognized revenue, annualized contract value, or another approved basis. Record the selected attribution model and effective date. Request raw CRM objects, stage history, and the written definition separately from an attribution chart. HubSpot's current attribution documentation and property-history export guidance explain some of those limits.
run the reconciliation worksheet
Use the downloadable reconciliation worksheet CSV to record each test. It includes illustrative rows and blank reusable rows. Do not collapse the differences into an average.
| Definition | Owner | Period | Source | Observed value | Discrepancy | Confidence | Unresolved gap |
|---|---|---|---|---|---|---|---|
Run the work in this order:
- Align fiscal period, timezone, currency, revenue basis, and expected sales-cycle lag.
- Map stable IDs across campaign, session, contact, account, opportunity, and finance records. Record unmatched and duplicate rows.
- Rebuild the pipeline view from CRM stage and amount history. Do not trust only the current field value.
- Separate media cost, agency cost, tooling cost, and other material cost so the decision can choose the right total.
- Trace optimization events to accepted opportunities and financial outcomes without assuming they should match one-for-one.
- Preserve the report/query name, filters, extraction date, row count, and omissions for every conclusion.
a synthetic worked example
All values below are illustrative. They are not a client result or an industry benchmark.
For one period, Google Ads reports 42 primary conversions on ad-interaction date. GA4 reports 31 consent-observed key events on event date. The CRM contains 37 created inquiry records, 22 with the agreed paid identifier, five accepted opportunities, and one finance-linked win.
The worksheet does not average 42, 31, and 37. It records each definition, population, and date basis. It matches stable IDs where possible, explains the unmatched records, and concludes only that five accepted opportunities and one finance-linked win survive the stated reconciliation. Platform conversions remain useful for optimization. They are not sourced pipeline proof on their own.
use readiness states, not a score that pretends to be a benchmark
Assign each material claim one of four operating states:
| State | Meaning | Required next move |
|---|---|---|
| Missing | The source, owner, definition, or period is unavailable | Request or preserve the evidence before drawing a conclusion |
| Documented | A definition and source exist but have not been tied across systems | Test the join, date basis, and coverage |
| Reconciled | The stated population and period agree across the approved evidence chain | Record limitations and the operating decision |
| Independently reproduced | A separate reviewer can rerun the documented method and reach the same result | Use it as a decision-ready baseline |
These are evidence-readiness labels. They are not a performance score, a percentile, or a valuation opinion.
keep a decision register and 30-day plan
The last page should name the decision, not decorate the audit.
| Decision or claim | Evidence and period | What reconciled | Limitation | Owner | Next test and date | Verification receipt |
|---|---|---|---|---|---|---|
In the first 30 days, confirm scope and access; preserve definitions and source extracts; map identities and run the reconciliation tests; then rank unresolved risks by revenue relevance, evidence quality, time to test, and reversibility. Close with the first operating actions, each with an owner and a receipt. In the register and CSV, record the rerun, approved extract, or other observable result in the verification-receipt field; leave it blank until that check actually passes. The result should survive handoff to the CFO, CRO, operating partner, or future buyer without needing the original analyst in the room.
where this worksheet fits
This resource is the deeper working layer for the portfolio-operations playbook. It does not replace commercial, legal, tax, or full quality-of-earnings diligence. It gives the marketing and revenue evidence a structure before someone turns it into an underwriting claim.
If the evidence is already present but the reporting still fails to reconcile, that is a growth constraint worth fixing. Start a growth diagnostic to identify the repair, owner, and verification path.
sources and boundaries
- Google Ads conversion definitions, attribution-model changes, attribution paths, and change history: platform behavior and limits cited above.
- Google Analytics consent mode and data retention: observed-data and retention limits cited above.
- HubSpot attribution reporting and property-history exports: CRM reporting and history limits cited above.
- The process, order of operations, and readiness labels are operator judgment. The synthetic example is illustrative. Neither is a benchmark, valuation opinion, or promise of performance.