Advanced Mathematical Analytics

Using bespoke statistical analysis designed for PPC.

Why Use Statistical Analysis

S

tatistical analysis is a powerful tool for proving the accuracy of any test results. It is used by a wide range of industries – from proving that a new medicine works to setting insurance premiums for different risk groups.

At Web Atomic, we bring this level of statistical analysis to PPC experiments. We run experiments to a 95% level of confidence so that we know the changes we are making to the account are having a positive impact on the account performance.

Bespoke Statistical Tests

Standard statistical tests can work well for some test analysis. However, we have worked closely with Dr Paul White, Head of Applied Statistics at The University of the West of England,  to design a series of tests specifically for PPC tests that we run. These tests take into consideration all of the variables and the unique testing environment of a PPC campaign to give the most statistically valid results possible.

Using these bespoke tests gives us an extra degree of certainty that the test results are making a real difference to the bottom line for our clients.

Statistics and Mathematical Modelling

Through a deep analysis of our client’s business and user behaviour, we are able to develop mathematical models that represent the underlying relationships that determine profitability. From here we create algorithmic bidding tools that automatically adjust campaigns according to the mathematical model to maximise profit.

Analysing PPC Performance

We use statistical analysis to answer questions about our clients PPC campaigns.

Everything from ad testing, optimium ad positions and keyword bids through to differences in ROI based on location or time of day – we know with a 95% level of confidence which combination of settings make the most profit. Statistical analysis is at the core of our scientific approach to PPC management.