THE SHORT ANSWER
Incrementality asks about additional impact compared with a credible situation without the advertising being studied.
For the broader picture of how target customers, channels and sales handoffs work together, read the pillar guide B2B demand generation: from attention to qualified pipeline.
Separate impact from allocation
Incrementality asks about additional impact compared with a credible situation without the advertising being studied. Attribution allocates observed interactions. A correctly attributed contract therefore does not establish that advertising caused the contract or that the customer would have disappeared without those impressions or clicks.
A brand advert may receive the last click even though the buyer already intended to visit the supplier. Category advertising may create interest without receiving the final recorded touch. Both examples explain why source allocation and additional impact remain separate analytical questions. Explore this further in the guide Brand search campaigns: evaluate existing demand honestly.
Define treatment and control
Define treatment, comparison group, outcome unit and observation window. In B2B, accounts may be a more suitable unit than individual contacts because colleagues participate in the same purchase. Google distinguishes campaign experiments from lift studies that investigate outcomes against groups not exposed to the advertising.
When assigning accounts, keep known colleagues in the same test condition. Otherwise internally shared resources may reach the control. Specify how subsidiaries and unidentified contacts are handled before results arrive rather than classifying difficult cases afterwards to fit the preferred conclusion.
Choose a workable comparison design
First assess whether random allocation or a credible geographic comparison is practical. Keep other sales activity comparable and record deviations. Specify the business outcome before starting. Allow for reporting delay and uncertainty rather than ending the experiment as soon as an encouraging interim result appears.
For geographic comparisons, examine historical demand, market size and sales coverage. Regions following different trends are not automatically credible controls. When random allocation is impossible, disclose the additional assumptions and test them against prior periods before presenting the observed difference as causal evidence.
Interpret three additional conversations
Suppose two randomly assigned groups of one hundred comparable accounts produce twelve and nine qualified conversations. The observed difference is three conversations. That is a starting point for evaluating uncertainty, not proof that every additional budget allocation will have the same effect. One large contract could also dominate the revenue comparison.
If only the treatment group also receives personal sales outreach, the three extra conversations cannot be isolated to advertising. Log other account activity. The design must separate the investigated change sufficiently for its outcome difference to retain an understandable and useful interpretation.
Identify spillovers and other interventions
Salespeople may treat the control group differently, and buyers may encounter adverts through colleagues. Such spillovers weaken separation. A before-and-after comparison without a control also mixes seasonality and market changes with advertising impact. Describe these limits before a report encourages a stronger causal claim than its design supports.
Check exposure too: an assigned person may never see an advert. Do not replace original allocation with a post-hoc comparison of reached people without explaining the selection effect. Otherwise self-selected reachability replaces the comparability the experiment was designed to create.
Integrity before impact
First check whether group separation held and outcomes were captured completely. Only then interpret the difference and its uncertainty. An inconclusive result may justify more observation or a better design; it does not automatically establish that advertising had no effect or that a positive platform metric proves lift.
Report absolute differences and assess uncertainty using an appropriate analysis. With few events, a large relative movement may remain highly uncertain. A smaller next experiment may be appropriate; lack of clarity should not be converted into a confident claim of zero effect.
Limit conclusions to their scope
Document the design and decision rule before launch. Marketing owns delivery, sales owns comparable handling and analytics owns outcome checks. Record the audience and spending level to which the result applies so a bounded experiment cannot become an unsupported claim about universal scalability or all future campaigns. Explore this further in the guide B2B paid media experiments with clear decision rules.
Record offer, market, period and advertising intensity. Positive results among known target accounts do not automatically transfer to broad unfamiliar audiences. The next scaling step is another decision whose changed conditions require a fresh check rather than an unsupported extrapolation.
INTERACTIVE / FROM READING TO ACTION
Three checkpoints for your implementation
Choose a checkpoint and track your reading progress. Checks are neither saved nor sent and do not score your business.
CHECKPOINT 01
Choose a workable comparison design
First assess whether random allocation or a credible geographic comparison is practical.
Explore in the article ↗CHECKPOINT 02
Interpret three additional conversations
Suppose two randomly assigned groups of one hundred comparable accounts produce twelve and nine qualified conversations.
Explore in the article ↗CHECKPOINT 03
Identify spillovers and other interventions
Salespeople may treat the control group differently, and buyers may encounter adverts through colleagues.
Explore in the article ↗Frequently asked questions
Is attribution the same thing?
No. Allocating observed interactions does not measure additional impact.
Is before-and-after enough?
It offers clues but cannot isolate other changes occurring at the same time.
What does an inconclusive test mean?
The evidence does not support a sufficiently clear decision; inspect design and observation time.
How can I put “Separate impact from allocation” into practice?
Incrementality asks about additional impact compared with a credible situation without the advertising being studied. Attribution allocates observed interactions. A correctly attributed contract therefore does not establish that advertising caused the contract or that the customer would have disappeared without those impressions or clicks.
What matters when working on “Define treatment and control”?
Define treatment, comparison group, outcome unit and observation window. In B2B, accounts may be a more suitable unit than individual contacts because colleagues participate in the same purchase. Google distinguishes campaign experiments from lift studies that investigate outcomes against groups not exposed to the advertising.
