Why Do Multiple Agencies Claim the Same Sale?

Start with the transactions the business recognizes.

Agree on the records used to establish orders, revenue or qualified opportunities. For ecommerce, that may be the order system reconciled with finance. For a longer sales cycle, it may be CRM opportunities with agreed qualification and revenue rules.

Specify treatment of returns, cancellations, taxes and discounts. Set a consistent reporting period and time zone. Preserve a distinction between booked orders, recognized revenue and collected cash when those measures differ.

These records establish the business total under the agreed definition. They do not automatically establish which marketing activity caused the result.

Ask each agency to explain its credit rules.

Google defines attribution as assigning credit to interactions along a customer's path. Different models allocate that credit differently. Read Google's attribution documentation.

For every report, request the conversion definition, attribution model, lookback window and whether view-through activity is included. Ask about modeled results, missing observations and delays in reporting.

A hypothetical customer may see a social advertisement, click a search ad and later buy after receiving an email. More than one platform may claim that purchase under its own rules. Adding all three reported revenue figures would overstate the business's sales if they include the same order.

Do not demand that reports from separate platforms always match. Instead, require the differences to be understood well enough to use each report appropriately.

Reconcile records where the data permits it.

Ask the measurement owner to compare transaction or lead identifiers using approved systems and appropriate permissions. Identify duplicate events, refunds and records excluded from one source but present in another.

Some reports use aggregation, modeling or privacy protections that prevent transaction-level matching. Label those limitations. An agency should not fabricate a neat reconciliation when the underlying data does not support one.

Review changes in tracking and definitions before interpreting a trend. If a report changes its lookback window, the apparent improvement may partly reflect the new counting rule. Preserve the old definition or annotate the break in comparability.

Evaluate business impact separately from platform optimization.

Platform reports can help teams manage campaigns within those platforms. Assess the overall marketing plan using the business's economics, with appropriate allowance for timing and other influences.

Incrementality asks what happened because of an intervention. Google describes conversion lift studies using treatment and control groups to estimate additional conversions. Such studies have availability and design constraints. Read about lift studies.

An experiment can strengthen causal evidence, but a small or poorly designed study may remain inconclusive. Ask a qualified measurement specialist what your budget and volume can support. Avoid replacing one overconfident dashboard with an overconfident experiment.

Agree on how the evidence affects agency evaluation.

Evaluate an agency against its contribution, responsibilities and dependencies. Creative work, technical implementation and media buying can all affect the same outcome.

Keep agreed delivery obligations visible alongside results. A team should be accountable for missed work it controls even when revenue is strong. It should also explain weak results without claiming that every external factor absolves it.

If compensation depends on attributed revenue, document the calculation, exclusions and change process in the agreement. Have the people responsible for finance and the contract review the formula before payment disputes arise.

Use this measurement agreement.

Complete this document with the agencies and the internal owner of the business records.

• We define the business outcome as [event], counted in [system] under [rules].
• We report the period in [time zone] and account for [returns, cancellations and adjustments].
• Each agency will disclose its attribution model, observation window and material limitations.
• [Measurement owner] will investigate discrepancies using [approved records and access].
• We will keep platform-attributed results separate from the reconciled business total.
• We will assess additional impact using [method], subject to [limitations].
• Any fee calculation will follow [contract reference], and changes require [approval].
• We will record definition changes and review this agreement on [date].

Use the agency-reporting guide for the rest of the reporting discussion. The purpose of reconciliation is to make decisions with a clear understanding of the evidence, including what remains uncertain.