How do you build a marketing source of truth that finance will sign off on? I Fospha Academy

A marketing source of truth is the one set of numbers marketing and finance both agree to run on. Building it takes a written agreement with five parts: a shared dictionary of metrics, open settings, proof sized to the stakes, a revenue bridge that closes, and finance in the room before budgets are set.

Why this matters

A single set of numbers stops teams arguing over conflicting figures and aligns them on real business impact. Getting there is harder than it sounds. Most source-of-truth projects fail for one of two reasons. Finance doesn’t trust the number, or the team can’t run the week on it. A real source of truth passes both tests. Finance can verify it, and marketing can use it every Monday morning.

The first test gets the attention. In a 2026 survey of 1,397 senior marketing and finance executives, Bain & Company and Google found that more than half of both groups see direct revenue impact as the most important way for marketing to earn standing with finance. Yet only 41% of the marketers surveyed felt equipped to tie their performance to business outcomes.

The second test gets less, and it’s where we see enterprise teams stall. Even brands with a strong marketing mix model (MMM) and a test program keep making their weekly calls from last-click and platform reports, because those update daily and everyone knows how to read them. A number that’s trusted in the quarterly review but ignored on Monday isn’t a source of truth.

Both tests come down to a written agreement. Bain found that leaders set a measurement framework shared by marketing and finance before budgets are allocated. Companies with strong marketing and finance relationships were almost 1.5 times more likely to lead their sectors, though that’s a correlation, not proof of cause.

Here’s the five-part agreement we’d put in place:

1. Build one dictionary

Write down every metric once, with one definition, one data source and one owner. Keep the list short. We’d cap it at ten metrics, because a long list is a sign nobody has decided what matters.

The test is simple. If two people can’t define a metric in a single sentence, it isn’t ready for the agreement. Retire a metric rather than redefining it, and date every change.

Start with the words that cause the most arguments. “Revenue” can mean gross sales, sales after returns, or what the ledger books. “New customer” can mean a first order on the site or a first order ever. “Spend” can include or exclude agency fees and tax. Retail brands face one more decision. Does revenue include retailer and marketplace sales, and does it count at sell-through or at wholesale shipment?

‍

2. Keep the settings open

Every number in the agreement should arrive with its settings, and with how it differs from last click. Finance trusts what it can question. A number nobody can interrogate gets overruled in the room, however good the model behind it.

Keep two logs.

A settings log. Attribution windows, how returns and discounts are treated, how promotions and seasonality are handled, when models were last refreshed, and any data source added or dropped. Record the date, the person and the reason for each change. When a number moves, everyone can see whether the market changed or a setting did.
A calls log. Record each budget decision, the result you expected, and what happened. After a few quarters, this is the strongest evidence that your numbers deserve trust, because it shows finance how often your forecasts held.

Both logs work with any measurement method, whether MMM, multi-touch attribution, incrementality tests or platform reports. Our lesson on triangulating the three shows how they fit together.

‍

3. Size the proof to the stakes

Decide in advance how much evidence each kind of decision needs. The more expensive a mistake is to reverse, the more independent the evidence should be. These thresholds are examples, so set your own.

Decision size	Example	Evidence you agreed to
Small and reversible	Move under 5% of monthly budget between campaigns, undoable within a week	The agreed measurement view, sense-checked against the blended revenue trend
Medium	Shift budget between channels for a quarter	The agreed view, reconciled to finance’s revenue, with settings logged
Large or hard to reverse	Set the annual budget, or enter a new channel or retailer program	Two independent lines of evidence that point the same way, plus finance sign-off before spend

“Independent evidence” can be a model, an incrementality test, a natural experiment or a historical comparison. The agreement sets the bar, and any method that clears it counts. Platform reporting keeps its job in daily optimization inside each platform. The ladder applies to decisions that cross platforms or budgets.

‍

4. Build a revenue bridge that closes

Every month, walk from what the platforms claim to what finance booked, and label every gap. The bar is that the bridge closes. A bridge with a large “unexplained” line isn’t finished, and the unexplained line should shrink every month.

The numbers differ for ordinary reasons. Each platform counts credit under its own rules and windows, so claims can overlap. Gross sales differ from net sales after returns and discounts. And for retail brands, a lot of revenue happens where the platforms can’t see it, on marketplaces, at retailers and in stores.

Line	Amount
Revenue the ad platforms claim in total	$1.30M
Less overlap: orders claimed by more than one platform	-$0.38M
Orders attributable to marketing in the agreed view	$0.92M
Plus orders with no tracked marketing touch (organic, direct)	$0.33M
Gross direct-to-consumer (DTC) sales in the store system	$1.25M
Less returns, discounts, tax and shipping	-$0.20M
DTC net revenue booked by finance	$1.05M
Plus marketplace revenue outside the platforms’ view	$0.55M
Total revenue in finance’s ledger	$1.60M

Illustrative numbers, not client data.

In this example the platforms claimed $1.30M, more than the $1.05M of DTC net revenue finance booked. Meanwhile $0.55M of marketplace revenue that marketing may have influenced sits outside the platforms’ view.

When the bridge doesn’t close, settle the tie-break rule before the argument starts. Finance’s booked revenue sets the total, and marketing’s view explains how it splits across channels. Then investigate in a fixed order:

Definitions. Are we counting the same thing?
Timing. Same dates, same time zone?
Data. Any missing or duplicate orders?
Settings. Did anything change in the log?

‍

5. Give finance a seat and the team a rhythm

Bring finance into planning before budgets and targets are set, and agree how often each number refreshes. In the Bain and Google survey, leaders lock their metrics before campaigns begin, and a strong chief marketing officer (CMO) and chief financial officer (CFO) relationship correlated with almost two times higher revenue growth. Again, that’s correlation from self-reported data.

In practice, hold a standing monthly session with named attendees from both teams, and put a sign-off line in the agreement. Name one owner on each side. Marketing owns the measurement view, and finance owns the revenue total.

Then match the rhythm to the decisions. If the agreed view only refreshes monthly, the team will keep running the week on whatever updates daily. Decide how often each metric needs to refresh, and which view is the default in the Monday trading meeting.

Treat the whole thing as a change program, not a reporting project. Expect the first quarter to feel slower, and train the people who’ll live in it. For how to order a plan from the profit and loss (P&L) statement, see our lesson on starting every marketing plan from the P&L.

‍

The one-page measurement agreement (copy and adapt)

Dictionary. Name, definition, data source, owner (ten metrics at most).
Settings log and calls log. Where they live, who maintains them, date last updated.
Evidence ladder. Thresholds for small, medium and large decisions.
Revenue bridge. Monthly, on [date], owners [marketing name] and [finance name], and a target for the unexplained line.
Tie-break. Finance’s booked revenue sets the total. Investigate in order: definitions, timing, data, settings.
Planning and rhythm. Finance joins planning in [month], sign-off by [name] on [date]. Refresh schedule for each metric, and the default view for weekly trading.

‍

Score your team

Tick each statement that’s true today.

1. Marketing and finance use the same written definition of revenue.

2. Changes to measurement settings are logged with a date and a reason.

3. Larger decisions need stronger evidence than smaller ones, and everyone agrees how much.

4. Marketing’s numbers are bridged to finance’s revenue every month.

5. Finance helps set targets before the budget is approved.

0 to 2: start with the dictionary. 3 to 4: add the missing clause. 5: review the agreement every quarter.

A good measurement system gives you the numbers. A great measurement system gives every team numbers they all agree on and run the week on. The agreement is how you get there. Build one dictionary, keep the settings open, size the proof to the stakes, bridge to finance’s revenue, and give finance a seat.

‍

FAQ

What do we do when marketing’s numbers and finance’s revenue disagree?

Follow the tie-break rule. Finance’s booked revenue sets the total, and you investigate definitions, timing, data and settings in that order. Until it’s resolved, apply the stronger evidence tier to any decision it touches.

Who should own the source of truth?

Both teams, with clear halves. Marketing owns the measurement view and the settings log. Finance owns the revenue total and signs off on definitions.

Does a source of truth mean using one measurement method?

No. The agreement sets how much evidence each decision needs. Any method, or combination of methods, that meets the bar can supply it.

Which revenue should a retail brand reconcile to?

The revenue finance books, defined in writing. Decide whether it includes retailer and marketplace sales, and when each counts. Then bridge the part the ad platforms can see, and list the rest separately.