Our system allows you to put all your data in the context of your overal results. It enables you to work with data across platforms (due to our customised attribution models), identify the best performing channels and campaigns (e.g. based on profitability or customer lifetime value), and to forecast your results up to 3 months in advance.
The rule of thumb is the more data the system works with, the higher precision of forecast. In terms of channels that record about 15 000 sessions per month, we typically work with data from the last 2 years, but if the the volume of available data is greater, we can work even with lower volumes.
The volume of conversions can be simulated using our system based on other interactions with your website, service or mobile app. The conversions do not necessarily need to be online orders or leads generated per se.
For marketing channels that have recorded 15k sessions per month in the last year and half, we can generate forecasts whose margin of error is minimal. For example in terms of customer lifetime value predictions, the precision of our system typically reaches the precision of 95%. And we make all our predictions even more precise every day.
We take security extremely seriously. The most complex calculations (e.g. comparison of precisions of various models) are run locally. The results are then saved on the safest platforms in the world such as Amazon AWS, Google Cloud Platform or Snowflake. We use Keboola to maximize the security during the transfer of data between all of the parts of the system. And we delete all data that are no longer needed for calculations.
Once the system is set up, we provide consultancy and training. In the first months, we typically focus on the following areas:
You are free to stop our cooperating anytime. We do not even require any notice period. This is how confident we are about the long-term value we offer.
The range of what our system can understand as a channel is quite broad. Most commonly, we group channels based on three dimensions (source, medium, campaigns) but other dimensions can be considered by the system too (e.g. special landing pages).
There are two ways of addressing this issue. We can either deduplicate users using Google algorithms, or we can use our own customisable and more precise deduplication method. However, the latter requires that the client has a system in place which pairs specific customers to a unique ID in at least 70% cases.