Is Your Email Marketing Creating Revenue—or Taking Credit for It?

A useful place to begin

Your email platform says a campaign generated $12,000. Your store reports $12,000 in total sales for the same period. Your paid advertising dashboard claims credit for much of that revenue too.

Which number is right?

Several systems can accurately report the credit assigned by their own rules while leaving the business question unanswered: how much additional revenue did the email create?

Email attribution and incremental impact answer different questions. Attribution assigns credit to interactions. Incrementality asks what changed because of the marketing. You need to understand both before deciding which emails deserve more investment.

This guide uses invented numbers to show how to make that decision more carefully. The examples are not industry benchmarks or reported client results.

Start by defining what the dashboard measures

Write down the platform's attribution rules. Does it assign revenue after an open, a click, or either? How long can the gap be between the interaction and purchase? Can another channel also receive credit for the same order?

Google Analytics, for example, lets teams configure attribution settings and lookback windows that affect how credit is assigned. That makes settings part of the interpretation, not merely an administrative detail. Source: Google Analytics attribution settings.

Do not add revenue totals from multiple marketing dashboards and call the sum company revenue. Reconcile actual orders against a single commercial source of truth, such as the order system or CRM.

Keep the attributed numbers. They can help describe paths and compare patterns within a consistent system. Just label them accurately.

Separate a click from the useful action

A campaign might be intended to help a trial user create a first project. Its useful outcome is project creation within a defined window. Clicks are a step along the path.

A post-purchase message might help someone set up a product. Its useful outcome might be successful setup, fewer avoidable support requests, or a later repeat purchase. Immediate sales may be an incomplete measure.

Write the goal before choosing the dashboard. Specify who is eligible, what counts as completion, and how long the person has to complete it. Keep the same definition when comparing groups.

Opens require particular care. Mailchimp documents that Apple Mail Privacy Protection can distort open reporting. Link activity can also include automated behavior; use the filtering available in your system and investigate patterns that do not resemble real use. Sources: Mailchimp MPP guidance and bot activity guidance.

Use a holdout when the decision justifies it

For an optional promotional email, a holdout means randomly assigning some eligible customers to receive the message and others not to receive that message. Both groups otherwise continue through the normal experience.

Do not withhold essential service, safety, security, or transactional information. Define exactly which optional campaign you are testing.

Randomize at a level that limits crossover. If several contacts at one company influence the same purchase, assigning the whole account together may be more appropriate than splitting individual contacts. If a household shares purchasing decisions, individual assignments may also contaminate the comparison.

Decide eligibility, group size, duration, exclusions, primary outcome, and a commercially meaningful effect before looking at results. The sample needed depends on your baseline rate, the effect you need to detect, and acceptable uncertainty. There is no universal “send to 1,000 people” rule.

A worked example: the revenue looked better than the contribution

Suppose 10,000 eligible customers are randomly split into two groups of 5,000.

The observed difference is 0.5 percentage points. Relative to the holdout's 2.5% rate, that is a 20% increase. These are two ways of describing the same observed difference, not two separate benefits.

At the point estimate, the email group has 25 more purchasers than expected at the holdout's rate: 5,000 × (3.0% − 2.5%) = 25. Multiplying by the assumed $80 average order value gives $2,000 in estimated additional revenue for that group.

That is the point estimate, not a proven $2,000 gain. A simple approximate 95% interval around the purchase-rate difference is roughly −0.14 to +1.14 percentage points. It includes zero. This illustration would not establish a positive effect at that conventional threshold.

Now examine the economics. Suppose contribution per additional order is $32 after product cost, discounts, payment fees, variable fulfillment, and expected returns. The point estimate of additional contribution is 25 × $32 = $800. If the campaign cost $300 in incremental labor and tools, the estimated contribution after campaign cost is $500, before fixed overhead.

The commercial conclusion depends on both uncertainty and margin. A revenue number alone leaves out half the decision.

Watch for sales moving in time

A promotion may encourage a customer to buy today instead of next week. A short measurement window can make this look like additional demand.

Choose a follow-up period that makes sense for the buying cycle. Examine later purchases, returns, and margin, not only the immediate response. When discounts are involved, measure realized revenue and contribution in each group rather than assuming the same order value or economics.

You are asking whether the campaign created useful business value over a meaningful horizon. An early spike can be part of that story, but it is not the whole story.

What if your audience is too small?

You can still make better decisions without pretending to have a conclusive experiment.

Verify delivery, triggers, exclusions, and destinations. Review replies. Observe whether the message helps people complete its intended task. Use a before-and-after comparison as directional evidence while recording other changes in audience, pricing, seasonality, and product behavior.

Accumulate comparable cohorts when the campaign and conditions remain sufficiently stable. Do not silently combine unrelated campaigns simply to produce a bigger sample.

If evidence remains inconclusive, say so. You may keep a low-cost, useful message while collecting more information. You may stop a costly one that cannot justify the uncertainty. Document the reasoning.

Build a report someone can use to decide

Show assigned or eligible counts, outcome counts, rates, attributed revenue, estimated incremental impact where defensible, contribution, costs, and uncertainty. Include complaints, unsubscribes, support burden, and customer outcomes as guardrails.

End with a decision: keep, revise, investigate further, or stop. Name the owner and the next review date.

Put this to work: Choose one email journey and write down what success means beyond the open or click. Use the First-Win Field Kit to record your measurement plan, or discuss the journey with Ashley. If the problem is a broken trigger or destination, Growth Fix may fit; a broader measurement rebuild needs its own scope.