How large is our panel size?

Accurat collects data of hundreds of thousands of individuals each day. However, not all data can be considered qualitative. A pre-processing flow normalizes data from different sources and guarantees quality. 

The final number of respondents takes into account the use case at hand. Some examples clarify not all data can be used at all times:

• Traffic today: when users do not generate coordinates all day, how is it possible to understand where a user was when he did not share data? That is why for example in app data is not used in Accurat’s models. Yet for this use case, data for most individuals that are tracked during a given day can be used, provided they have been active multiple days in a row to make sure their home location is known, an important ingredient to preserve representativity. 
• Reach of a campaign this week: For this use cases, it is required users are active for the full week and have shown activity the days before in order to understand their home location. A subset of the total dataset can be used.
• Drive-to-store impact of a campaign: In this case, we want to understand how visits evolved for an exposed group (those who saw a billboard) vs a control group (those who did not see the campaign). The evolution compares the campaign week with the average traffic of the four weeks prior to the campaign. Given this definition, users have to be active for 5 weeks in a row. This stricter definition leads to lower numbers.

Given our positioning to offer “market research on steroids”, our goal is to allow significant metrics to be shared for a subset of users, based on a combination of variables (e.g. a specific period x a specific number of panels x a specific audience definition). Numbers vary between a few thousand and a hundred thousand users.