Clicks and video completions describe responses to delivery. Brand Lift measures differences in awareness and attitudes between audiences with and without recorded advertising exposure. The strength of the conclusion depends on how comparable those groups are.
Start with the question you need to answer
Plan the study around a decision: assess awareness of a new medicine, message understanding or willingness to consider a brand. Choose the primary measure before the campaign. Additional measures help explain the finding.
| Measure | What is assessed | Report label |
|---|---|---|
| Advertising recall | Recall or recognition within a specified period | Ad Recall |
| Brand awareness | Unaided name recall or aided recognition | Brand Awareness |
| Consideration | Including the brand in the choice set | Consideration |
| Attitude | Rating the brand on an agreed scale | Favorability |
| Purchase intention | Planning a purchase under stated conditions and timing | Purchase Intent |
Considering a brand means including it among the options; planning to buy expresses a purchase intention. Sales records establish the purchase itself. Advertising impact on purchases is assessed through Sales Lift.
How the groups are compared
Our methodology compares responses from people with recorded advertising exposure and people without recorded exposure. Both groups receive the same questionnaire within comparable periods.
Test group
Exposure recordedControl group
No exposure recordedCheck comparability on characteristics that influence the answer. Active category buyers may know a medicine better before any advertising. If more of them enter the test, the full difference cannot be attributed to the campaign.
Confirm exposure through records
Agree tracking or identifier exports before launch and define exposure: an impression, a viewable impression or a view. Clickers are a more active subset; findings about them cannot be generalised to everyone reached.
Check the control group separately
No impression record does not mean someone never saw the advertising. Other devices and channels may remain untracked. Account for that limitation when interpreting findings.
Conditions that strengthen the comparison
Random allocation of one eligible audience before delivery, with controls excluded from advertising, provides a stronger basis for causal inference. This is a holdout design. Check availability in advance; group assignment and actual exposure are different events.
When groups are formed from recorded exposure, check their composition and use matching or weighting where appropriate. These methods address observed differences, but cannot remove every source of bias.
Each metric needs adequate response bases in both groups. A 90/10 split is not a quality standard. Sample planning depends on the baseline, smallest important effect and required precision. Publisher or regional cuts need adequate bases of their own.
What a survey banner can look like
Survey the research audience using an agreed format. These three illustrative layouts show how the question changes with the metric. Actual wording and response lists should fit the category.
Thank you for answering
Question order matters. Ask unaided recall questions before showing brand names or creative; otherwise the survey supplies a prompt. Afterwards, show materials to assess recognition, brand attribution and message understanding.
Specify the period and choice situation. In pharma, people may choose for themselves or a child, and remember advertising long before a need arises. Those distinctions change the meaning of the answer.
Two ways to describe the same lift
For a proportion-based metric, compare the underlying rates in both groups first. Absolute lift is expressed in percentage points. Relative lift expresses the difference between groups as a percentage of the control rate.
percentage points
24% minus 20%
relative lift
4 divided by 20, expressed as a percentage
In this hypothetical example, 200 of 1,000 controls and 240 of 1,000 test respondents selected the brand. The difference is 4 percentage points, or 20% relative to control. A “20% increase” without underlying rates says little about the scale.
A difference is visible. How convincing is it?
The report needs both response bases and a confidence interval, describing uncertainty under the chosen calculation method. If the difference interval includes zero, a standard two-sided comparison does not confidently distinguish the result from random variation. This does not prove no effect.
A statistically significant difference may be too small to affect a budget decision. Define practical importance before launch. For rating scales, agree whether to compare means or proportions; the calculation above specifically concerns proportions.
From a finding to the next campaign
Brand Lift is useful when the team understands what to test next. These are interpretation scenarios rather than findings from particular projects.
Advertising recalled, brand poorly recognised
Check whether the name and packaging are noticeable and the message is clearly attributed to the brand. Test revised creative.
Awareness up, consideration unchanged
Check the reason to choose and relevance of the situation. Recognition alone does not persuade someone to buy.
No convincing difference detected
Check precision, exposure tracking and the questionnaire. Then consider new creative, another design or more responses.
Publisher, creative and frequency cuts help investigate causes. A group with high lift is not automatically the best buying option: check size, composition and exposure overlap. Many comparisons increase the risk of a chance favourable finding. Confirm a working hypothesis with a separate test.
For pharmaceutical campaigns, account for seasonality, geography and other advertising. Parents or travellers in thematic segments may respond differently to a message; membership does not establish interest in a particular medicine.
What to agree before launch
Plan the study alongside the media plan. A completed campaign cannot be given a random control retrospectively, and untracked contacts cannot simply be recovered.
- Objective
- Primary measure and decision to inform
- Audience
- Category, geography and buying roles
- Comparison
- Test/control and exposure recording rules
- Questionnaire
- Exact questions, order and answer options
- Data
- Sample plan, timing and reporting cuts
- Report
- Calculation, precision, limits and recommendations
At Pharmatic, Brand Lift availability, methodology and cost are agreed for the specific placement before launch. The output should include survey questions, response bases, underlying values, lift, precision and recommendations tied to the data.
Include Brand Lift in the measurement plan
Combine brand measures with delivery quality and search-interest changes. These sources answer different questions and provide a fuller view of the campaign.
Building an analytics plan