The source of everything: mobile SDK data

GDPR compliance & Deloitte as DPO

Always vs in-app data: what's the difference?

Accuracy of location data

Granularity of the data

SDK mobile data vs other sources of 'mobile' data

Myths about mobile SDK data

Geofencing vs building matching

Frequent locations (home, work...)

Representativity as a cornerstone

Transport mode definition

Billboard passages explained

Drive-to-store & attribution algorithms

Moving transport measurements

Audience profiling

Getting Started

All ins & outs of the insights we provide

Using Accurat Products

Reports & Dashboards

All there is to know about the data behind

Data & Integrations

Support & Services

How is data processed? How are insights calculated?

About our models

Learning Resources

Every coordinate has an&nbsp;accuracy, a number between 0 and infinite (theoretically) indicating the region where the coordinate is recorded and the mobile device actually is.The horizontal accuracy radius is expressed in meters of this location at the 68th percentile confidence level. This means that there is a&nbsp;68% chance that the true location of the device is within a distance of this uncertainty of the reported location. Another way of putting this is that if a circle with a radius equal to this accuracy is drawn around the reported location, there is a 68% chance that the true location falls within this circle.Most coordinates are between 15 and 25 meters with coordinates ranging from 0 to 50m being kept for further modelling. Other coordinates are removed from the dataset.

Geofencing (on the left) is a commonly used technique to estimate foot traffic. For a building, billboard or other location, a circle is drawn with a radius of typically 50-100m. Every enter or exit event - whenever a smartphone enters or exists the geofence - is counted as an interaction. People passing by the crossroad nearby are counted equally for people having actually visited the location.&nbsp;<figure class="image image-style-side"><img src="https://lh7-rt.googleusercontent.com/docsz/AD_4nXe9REPy7vSzIybDHLiAh-62WxujJmbxu6J63m5AuKeMg77wyRwlmHYy3KRFe5KPRegRbA4iOk1MskguBaO1wvxxLazN6h591j7zpihOhocWJx4a0aJQYwrdTOgyr73H6V019zysGA?key=NlIFlyvP2goJg2m9PXDJb1TM"></figure><figure class="image image-style-side"><img src="https://lh7-rt.googleusercontent.com/docsz/AD_4nXeYwWJUCN0i7Cj8OOgf6mtJ61hGYPGLx3iITulgXfhJgf8HNMJ7J6TQR4ho7NOcr4jA7eLRWiYd8TePNRWIYdDorPeGoZnG4U2H6ECyn0UM_sorfg4buBOVYf6Hdj9jkNxHbVW3aw?key=NlIFlyvP2goJg2m9PXDJb1TM"></figure>Building matching (on the right) or&nbsp;route mapping of the taken route based on a combination of coordinates provides an alternative and more accurate approach to define visits or “interactions”. In this case, smart algorithms define the&nbsp;visit probability&nbsp;based on all coordinates, the building shape or the orientation of a billboard, the travel direction, the dwell time of the visit (time between coordinates), the speed of travel, opening hours, the time of day, the&nbsp;transport mode etc.&nbsp;The picture below illustrates how this works in practice.<h2><img src="https://lh7-rt.googleusercontent.com/docsz/AD_4nXe0yeJWmkR7QY7Yj8lZP8nN-8A3HVFIpPZzFPo_P0YfIa6X7d16AVD-0Ewvqepn36NKQPwiG3pLKXEEL0gRjlDz6TxRO_8k067_4RdH7sTOT_7fryOMiU1lZbN-F-sHNbdjrbtX-w?key=NlIFlyvP2goJg2m9PXDJb1TM"></h2>

On one hand, it is important to collect data everywhere. The map illustrates the distribution of the Accurat panel vs the distribution of the actual inhabitants in Belgium. As can be seen, both graphs look very similar.However, results are not a 100% fit with the Belgian population. Therefore, similar to traditional market research, techniques are used to re-weigh users.For each selected user in the subset, the&nbsp;frequent location algorithm derived the home location. For each user, we hence know its NIS6 zone to which he belongs (2.600+ zones in Belgium) that can be compared with Statbel data resulting in 4 parameters:<ul><li>#sublocality_users: # of panel members living in the NIS-6 sublocality</li><li>#sublocality_inhabitants: # of Belgians living in the NIS-6 sublocality according to Statbel</li><li>#users: total users in our panel</li><li>#inhabitants: total # of inhabitants in Belgium according to Statbel</li></ul>All subset users are weighted for sublocality using the following formula:<img src="https://lh7-rt.googleusercontent.com/docsz/AD_4nXdiXLrVEkUdkiEbvkEApOnamdeDwCvKgk6PccEQ-rVeICkJTuxwDOHEoTTZuzC7TvlSknGvtvRUiA03qqMFzIL8XHE7CDzOFVoa0LrZtzhJKW4fvNhWMRplA47KsKVXCvtNdgoCMg?key=NlIFlyvP2goJg2m9PXDJb1TM">This reweighting is more strict than traditional methodologies in which 10 provinces are typically used. Similar to this reweighting, users are also reweighted for additional information regarding their&nbsp;socio demographic profile using Dividuals data (provided by Sirius Insights). More info is shared in the&nbsp;<a href="https://docs.google.com/document/d/1OHC_7dv6SfiUCwAJYmvtxTCAXR5jbE4kqfvLNJ_40eA/edit?tab=t.0#heading=h.ry3p3b2avw5a">‘Audience profiling’ section</a> below.

Go back to application

Was this article helpful?

You recently requested to reset your password for your account. Use the button below to reset it. This password reset is only valid for the next 24 hours.

If you’re having trouble with the button above, copy and paste the URL below into your web browser.

Help Center

Data & Integrations

About our models

Featured Articles

Accuracy of location data

Geofencing vs building matching

Representativity as a cornerstone