Normative Weighting
Overview
Zappi weights survey data to make sure results are driven by what's actually being tested, not by who happened to answer. This type of weighting is used in Activate It and Amplify Systems, and we do this in two ways:
- Target group weighting to keep demographic profiles consistent across cells
- Normative weighting to control for brand and category usage by weighting to historical category or brand usage norms.
Target group weighting
This is where collected data is weighted to match the quotas that were set in sample, typically done for two reasons:
- To weight to an interlock where one was not set in sample. For example, flat quotas may have been set for age and gender, but we would like each cell within a solution to have the exact same profile of each age and gender combination for reasons of comparability.
- To weight results to their intended stratification where either quota minimums or interventions in fieldwork have meant that a cell does not hit quota targets exactly.
Normative weighting
Normative weighting applies to the Amplify and ActivateIt systems. For each solution in these systems, normative weighting keeps our data consistent by ensuring the group of people we survey always reflects the same brand and category usage profile. Rather than trying to source external benchmarks for every brand and category on the platform, we weight to what we've observed historically. This is simpler, and has the added benefit of targets that naturally shift as brands and categories evolve over time.
The result is that score differences reflect the stimuli being tested, not variation in the sample.
Calculating weighting targets example
The below gives an example of how normative weighting targets are calculated for a given cell in Amplify TV:
- Category usage norms are calculated at the child category level within a country. A category usage norm is created for each usage frequency option in the ‘category usage question’ for all cells that match the criteria for creating a norm.
- Brand usage norms are scoped to the specific brand and child category within a country. A brand usage norm is created for each usage frequency response option in the ‘brand usage question’ using all cells in the database that meet the criteria for creating a brand usage norm.
Since normative weights are based on historical data, the first test of a category or brand can't be weighted, since there's nothing to compare to yet. From the second order onwards, weighting kicks in, with targets calculated from the cumulative average of all previous cells up to that point.
Criteria for a category or brand usage norms
Count of cells
A minimum of 2 cells is required before a category or brand usage norm is created, ensuring targets are based on a stable baseline. Category usage norms tend to be available sooner, so it's common for cells to be weighted for category usage before a brand usage norm is ready.
Criteria for a cell to contribute towards the norm
| Response Options | Category or brand usage norms are only calculated using cells that share the exact same response options, both in wording and number of options. Different response scales produce different answer patterns, making them incompatible for weighting. For example, if one survey uses '2 or 3 times a week' as an option and another skips straight from 'every day' to 'once a week', those two surveys can't contribute to the same norm. |
|---|---|
| Time of completion | Category or brand usage norms are built from all cells that completed up to the date the cell being weighted finished, meaning targets can differ between earlier and later cells. This can even cause slight variation within the same order if cells complete on different dates. |
| Audience | In modular solutions, category or brand usage norms are also scoped by audience. This is necessary because flexible quotas and screening conditions mean different audiences can have meaningfully different brand and category usage profiles. |
Onboarding
Normative weighting keeps your results consistent over time by ensuring the group of people surveyed always reflects your brand or category accurately. To work properly, the system needs at least two previous tests run with the same questions and audience settings. This is what allows meaningful comparisons across tests.
This approach ensures that a high score reflects a genuinely strong ad or concept, not just a skew in who happened to be surveyed that day. It also removes the need for manual data updates because the system learns continuously from new data, so as your brand evolves, the benchmarks adjust automatically.
This applies independently per solution. If a you have already run a cell for a given category on Amplify TV and later want to run that same category on Amplify Digital or Activate It, a new cell is required for each additional solution. Weighting is tied to the combination of brand, category and solution.
What to expect
For any new brand or category, the first test needs to be run on its own as a single submission before additional tests are launched. This creates a stable baseline that all future results are measured against, and is what makes the platform's speed and automation reliable going forward.
The first test will not yet have brand or category-specific weighting applied. This is expected, as there is no prior data to reference yet. Think of it as a calibration step. Once complete, every subsequent test benefits from consistent, accurate benchmarking.
For high-priority projects where immediate stability is needed, it may be possible to preload relevant historical data so weighting is active from day one. Speak with your implementation team for more information.