Attribution tests with paused campaigns: switching on and off with method

Abstract orange illustration on a dark background: a lab flask. Article topic: “Attribution tests with paused campaigns: switching on and off with method”
In brief

One of the most honest ways of knowing what a campaign is worth is to switch it off for a period and see what happens to sales. It is a simple attribution test: if sales fall a little, the campaign brought little; if they fall a lot, it was essential. It calls for method, so as not to confuse the effect with other causes.

The essentials in a definition

Attribution tests with paused campaigns consist of switching a channel or campaign off and on in a planned way to observe the effect on total sales. It is a simple way to estimate incrementality without complex tools, but requires method: comparable periods, one change, defined duration and a record. It complements conversion measurement, models such as Meridian and Think with Google studies.

The logic of the test

Instead of relying only on platform reports, you compare the business’s total result with and without the campaign. If the campaign is capturing sales that would exist anyway (brand, remarketing), pausing it shows this. If it is creating demand, the drop will be visible.

It is the low-cost version of incrementality tests.

Choosing what to pause

The campaigns most at risk of overstatement are brand search, remarketing and some product campaigns very close to purchase. Choose one at a time. Avoid pausing what sustains the business all at once. Define the duration (a few weeks, covering a cycle) and the metric (total sales, leads, margin).

If possible, do it in regions or segments, to keep the rest active.

Illustrative diagram with 4 numbered steps joined by an orange line, the last one marked as done. It represents: An attribution test by pausing

An attribution test by pausing

  1. Choose a campaign at risk of overstatement
  2. Define duration and total metric
  3. Pause, recording simultaneous changes
  4. Compare total sales with what was reported

Design cautions

Avoid periods with promotions, holidays or price changes. Record everything that changes at the same time. Compare with an equivalent earlier period or a control group. Also track effects on other channels (organic, direct). Have a plan to resume.

Accept that the result is approximate.

Large orange quotation marks on a dark background with abstract lines of text, illustrating the quote: “When an ad is switched off, you see how much of what it reported was really its own.”
When an ad is switched off, you see how much of what it reported was really its own.

Interpret and decide

Calculate the change in total sales and compare with the sales attributed to the campaign. If the drop is far smaller than the campaign reported, its real value is lower. Review budgets accordingly. If the drop is large, keep or reinforce. Repeat the test when conditions change.

Keep the record for future decisions.

Decision table

StepActionCaution
HypothesisChannel X generates incremental salesDefine the minimum relevant effect
DesignPause for 2 to 4 weeks in a stable periodAvoid special seasons
MeasurementCompare total sales with the previous period and a controlConsider seasonality
DecisionReduce, keep or increaseRecord results and uncertainty

Common mistakes to avoid

  • Pausing in peak season. The cost of lost sales is high.
  • Changing other things at the same time. You will not know the cause.
  • Pause too short. The effect may take time.

How to measure the result

Compare total sales, margin and brand searches between the pause period and comparable periods, and record the uncertainty.

Limits and exceptions

External factors (promotions, competition, news) can distort results, and a single test is rarely conclusive. In channels with deferred effect, the pause may underestimate the effect. Repeat the test when the decision is important, use control regions whenever possible and combine with other evidence.

Illustrative example

Illustrative example: a shop pauses remarketing for three weeks in a stable period and compares total sales, repeat-customer sales and brand searches with the previous three weeks and with a control region. Sales fall little, and the shop reduces the remarketing budget, reinvesting in new-customer acquisition.

Frequently asked questions

Which campaign do I pause first?

Usually brand or remarketing, where the risk of overstatement is greatest.

How long do I pause?

Long enough to cover a buying cycle, generally a few weeks.

What if sales fall?

Resume and record. The test showed the campaign had value.

How long should the pause last?

Long enough to observe the effect, generally two to four weeks, adjusted to the purchase cycle.

Is it risky?

It can cost sales; limit the test to one channel, a stable period and, if possible, one region.

Summary to quote

Planned pauses compared with equivalent periods estimate a channel’s incremental effect, provided there is one change, a defined duration and a record.

Sources and recommended reading

Last editorial review: October 2026. Informational content; always confirm current conditions in the official sources before deciding.

Keep reading

Contacto rápido

Quer falar sobre as suas campanhas?

Deixe uma nota curta. Respondo em 1 dia útil.