Publish your content on The Drum
Agencies for Growth Festival Banner

Everything you need to know about programmatic optimisation: 'Test, learn, feed back – and fail fast'

It can be hard to keep pace in this fast moving industry, which is where we come in. The Drum has teamed with the Trade Desk for a series of short video primers and accompanying features in print to tell agencies and brands everything they need to know about complex issues. In this instalment, it’s the turn of programmatic optimisation, as we examine how to understand your audience through programmatic.

Big data alone is not enough today – marketers must be able to harvest the right data in real, or near-real time; to analyse it and optimise it in order to better reach their customers.

We have, says research firm Forrester, moved from the age of information to the age of the customer. And those who know and serve those customers the best stand to gain the most.

One of the key advantages of programmatically purchasing ad inventory is that the results are data-driven and measurable.

A data management platform, or DMP, collects user data from both first party (an advertiser’s own) and third party sources. These can then be used to generate insights that help power a demand-side platform (DSP), a system that allows advertisers to purchase display ad inventory via exchanges.

However, in order to optimise effectively people-power is needed – a DMP can process and sort billions of datum, but only a person can analyse what that data means.

For instance, the Trade Desk recently worked with a car manufacturer that was advertising its flagship motor – a very expensive, high performance sports car. The campaign aimed to encourage people to book a test drive, yet few were doing so despite having watched the ad.

Says UK general manager James Patterson: “Crunching the data we found a large proportion of viewers was made up of teenagers and young adults – not the target market at all – they were only ever window shoppers.”

The insights allowed the manufacturer to return to the brief and optimise for the right audience at the right time – those with both the mind and the means to buy such a car.

The example illustrates both the importance of robust campaign planning at the start of the process and the ability programmatic affords to change course part way through.

Martin Beauchamp, head of programmatic at MEC, says: “Despite the fact that we live in a cool, interesting and rich place for media and can quickly access huge amounts of data, the fundamentals still stand. We need to look at the function of data and what we can do with that. What is a client’s end goal? What do they want to achieve?”

He says it starts with optimising to something that is concrete, whether that be a set of KPIs, a certain audience or increasing brand reach and setting out the data to measure.

Plan, consider what tactics have either worked before in combination with gut feeling. Put all of that together, manipulate the data sets and test, he advises. Break the campaign out into different sets of data that you want to test instead of bundling up into one massive campaign. “You can then work out the influences of each element,” he adds. “Test, learn, feed back – and fail fast.” If it works, increase spend and optimise again.

“It is very much about the feedback loop,” continues Beauchamp, a former DMP solutions engineer.

“You might plan around an audience that you believe is perfect for you, but as the data gets fed back it might show your audience either isn’t right or that if you change something ever so slightly to one particular segment you get a better result.”

With programmatic, if something is over or under-indexing on a report a brand can feed back that data in real-time and weight the campaign accordingly. Insights should also inform future campaigns – and even parts of the marketing plan not bought programmatically, such as social or print.

Building a bigger understanding and digital footprint about a consumer or potential customer through optimisation has advantages beyond marketing, too, such as informing a company’s commercial team what products to stock and when.

Such possibilities are why the man versus machine debate that surrounds programmatic and the future of advertising and media is flawed, says Patterson. Technology gives agency and marketing teams the tools to understand and target audiences, but automation alone is not enough. “Machines generate data, people generate insights – and clients want insights. You can’t commoditise insights.”

Everything You Need to Know About Programmatic Optimisation is the fourth in the EYNTK series designed to help viewers quickly get up to speed with some of the most important issues in today’s marketing industry.

The first four episodes of EYNTK are available to view here.

This article first appeared in The Drum's 13 November issue which is available to buy now from The Drum Store.

By continuing to use The Drum, I accept the use of cookies as per The Drum's privacy policy