When two agencies report different leads, revenue, or return, the fastest reaction is to ask which dashboard is right. That question is premature. The reports may use different conversion definitions, time zones, attribution windows, scopes, filters, ownership rules, or CRM joins. Diagnose the data contract and lineage before changing providers again.
1. Freeze the decision and the reporting period
Write what the organization needs to decide: renew a provider, allocate budget, explain a variance, approve a handover, or repair measurement. Freeze a comparable period, account scope, currency, time zone, and outcome maturity window. Save the original exports and report configurations.
Do not reconcile changing date ranges while campaigns, tracking, or CRM stages are moving. A frozen period gives the team a common specimen even when the conclusions remain uncertain.
2. Build a metric dictionary
List every disputed metric: impression, click, session, event, key event, conversion, lead, accepted lead, opportunity, revenue, cost, and return. Define numerator, denominator, source, filter, timestamp, deduplication, owner, and refresh schedule.
Use one vocabulary when comparing reports: event, key event, advertising conversion, lead, and revenue should each have their own definition. See Google Analytics’ conversion and key-event guidance. A bare “conversion” without a product and counting rule is not a shared metric.
3. Trace data lineage from source to report
Draw ad platform, tag or SDK, analytics property, call or form system, CRM, warehouse, spreadsheet, dashboard, and agency export. For each handoff record field mapping, transformation, filter, join key, time zone, and last change. Identify the system of record for every business fact.
Compare raw rows before comparing totals. A small sample with IDs, timestamps, source, campaign, status, and value can reveal where two pipelines diverge. A polished dashboard cannot prove that its upstream rows are complete.
Ask both agencies for the query, filter, calculated field, and refresh timestamp behind the number. Store the files in a shared evidence register with a version and owner. If a provider cannot expose the row-level definition because the report is a proprietary screenshot, mark that metric as not reproducible.
4. Compare attribution and counting rules
Record first touch, last touch, data-driven or custom model, lookback windows, direct-traffic treatment, view-through rules, conversion counting, cross-device behavior, and imported offline outcomes. Two agencies can both calculate their report correctly under different models.
Google’s attribution overview explains that attribution assigns credit across touchpoints; credit allocation is not the same as causal proof. Keep model output, observed outcomes, and business decisions in separate columns.
5. Check campaign and channel scope
Verify accounts, properties, campaigns, locations, devices, search types, brand filters, exclusions, agency-managed channels, and start dates. A provider may include a channel that another excludes or use a different definition of active campaign.
Check whether the handover left duplicate tags, old UTMs, legacy conversion actions, or overlapping audiences. Preserve original source names and map them to a controlled taxonomy instead of renaming history in place.
6. Reconcile CRM and revenue outcomes
Join report records to CRM leads, accepted stages, opportunities, closed outcomes, and revenue. Test duplicates, recycled leads, existing customers, late-stage imports, refunds, cancellations, and records without an owner. Define when a result becomes mature enough for the comparison.
Agency reporting may end at a platform conversion while the owner cares about qualified pipeline. Salesforce’s Account Engagement campaign reporting illustrates why campaign metrics and opportunity outcomes can appear in different reporting layers. Join them deliberately rather than assuming the dashboard has done so.
7. Review the handover and ownership contract
Record who owns accounts, tags, audiences, naming, report definitions, credentials, exports, data retention, and change approval. Ask for the previous provider’s field map, conversion dictionary, query logic, and exclusions. If these were never delivered, list the missing evidence instead of blaming the new report.
Separate implementation ownership from analytical interpretation. An agency may be responsible for campaign execution while the owner retains the definition of qualified lead and revenue. Write the boundary into the operating record.
Review access logs and change history where available. A new agency may inherit a tag, conversion action, or dashboard that another team still edits. Conflicting reports can be a consequence of conflicting control, so the remedy may be an access or approval change rather than another visualization.
8. Use a reconciliation matrix
| Control | Pass evidence | Hold signal | | — | — | — | | period | same dates, time zone and maturity window | reports mix live and mature cohorts | | metric | definition, source and denominator match | “conversion” means different events | | lineage | sample IDs survive each handoff | dashboard cannot expose source rows | | attribution | model and counting rule are declared | model output is called causality | | scope | accounts, channels and filters match | one provider includes hidden traffic | | outcome | CRM and revenue join is documented | platform lead equals pipeline by assumption | | ownership | dictionary, access and rollback are assigned | no one owns the contract |
Keep both reports, raw extracts, mapping tables, query versions, reviewer, exception owner, and decision date. A reconciliation pass should explain unresolved differences, not erase them.
9. Choose one controlled repair
Select the smallest issue with the highest confidence: align a metric definition, remove a duplicate tag, restore a source field, correct a time zone, repair a CRM join, or rebuild one dashboard. Run the change against a frozen cohort and compare both providers’ outputs before changing budgets or ending a contract.
Set a follow-up date for the reconciliation record and assign an owner for unresolved differences. A disagreement that remains open should be visible in the next report with its scope and evidence, not silently converted into a new baseline.
The durable artifact is an Agency Reporting Reconciliation Record linking metric dictionary, lineage, attribution, scope, CRM outcomes, ownership, exceptions, and rollback. It turns conflicting reports into a governance decision that remains useful after the next provider handover.
How did this article land?
Choose one reaction. You can change it anytime.
