Joining the data dots


At SMRS, we are a data rich agency with multiple custom-built planning tools, each utilising various datasets to build market leading audience insights for our clients.

But like many organisations – we still had a fundamental challenge; how do we join the (data) dots.

We have incredible offline data sets and planning tools. Unique to us, and powerful.

And we know how to utilise the market’s best online data sets. So, whether it’s for planning, understanding and finding audiences, or designing strategy – we had all the component parts.

But we were still reliant on individuals digesting all the separate variables. And that is a challenge.

We had to find a way to consistently take deep and advanced strategic planning outputs and link them to activation on a live campaign.

We wanted to mash together our historical media performance, market insight and planning data and live data from the platforms in which we advertise. And we needed it to be automatic. That way it would be scalable, accessible, have synergistic benefits – and everyone would have the same insight.

We explored black box solutions such as DMPs but in the end, we decided to build it ourselves. Brave, and potentially challenging, but the right solution to the need.

Our goal; a piece of ‘middleware’ that joined the separate data systems.

Specifically, we had to take our offline data sets and insights, and enrich these by harnessing the power of online, real-time data.

The eventual benefit; a tool that will ultimately better support SMRS’ planning decisions across all channels and media.

The first stage we undertook was EDA and data engineering. Unsurprisingly we had barriers to overcome, but all were eventually resolved with a little head scratching, exploration and teamwork.

Once that was complete, we moved on to the Modelling & Analysis stage. Here we created algorithms that dynamically set scoring models (based on the aggregated sources). These allow us to create a further insight layer – but one that digests and adjusts based on all of the separate source variables.

It does things that our brains simply can’t do. And doesn’t make the errors that we would inevitably make.

The next and final stage will be Visualisation. Making this open, accessible and digestible to all who need and want it.

Oh, and this is only phase one. Once this project is done, we won’t stop there. We will continue to add other data into the mix, and that’s where it gets even more exciting. Watch this space.