Houston, we have a measurement problem

If we stop thinking about digital as an amorphous mass then we can demonstrate value to brands and make their communications more effective, writes Ben Dudley

Measuring return on investment is important. This is hardly a revelation: brands invest in advertising in order to sell more of their products and services and those that don’t do it effectively or profitably won’t last long.

It should go without saying that understanding which elements of the plan are working better than others – and how they work together – is crucial to improving performance over time. And as digital advertising takes an ever greater share of budget, so does the imperative to measure the effect it has.

So how to do it? In theory, the increase of digital advertising has made it easier to demonstrate a link between investment in marketing communications and results, but – as has been pointed out many times on these pages and beyond – demonstrating a link isn’t the same as proving effectiveness (and let’s not start on whether the ad has been seen by a human or not).

Econometric models are still the predominant industry choice for quantifying the incremental impact of communications and, to a greater or lesser degree, they do incorporate digital.

Isn’t that the solution then? Well, sort of, but here’s where the problem appears. Digital comprises lots of different content, in lots of different places, consumed on many different devices, all of which give a different experience to the person consuming it.

It’s not an amorphous mass, but it’s often treated as such when we’re measuring it – partly because that’s the easiest thing to do, but also because the ubiquity of programmatic trading means the content the ad is placed next to is typically deemed much less important than “who” it’s being served to.

Some of that online investment goes to digital newsbrands but, in Newsworks’ ROI study last year, there was no way of proving their effectiveness as separate from the amorphous mass. So Newsworks set us at Havas Media Group a challenge: how can we show how digital newsbrands are performing when they are not tracked separately on plans? This needed some creative thinking and a very deliberate approach to measurement.

Firstly, we needed to find and be able to isolate sufficient investment in digital newsbrands and secondly, we need to ensure we could robustly measure its effect and compare it to other activity.

We decided to tackle this question in two different ways – 1) focusing on a content partnership which worked across print, digital and social, and 2) very deliberately picking out digital newsbrands in a digital attribution model.

Content partnerships provide an ideal opportunity to measure the impact of digital newsbrands. They typically have a higher proportion of media value delivered digitally, levels of investment are high enough for robust measurement, and the phasing often provides natural variation from other campaign activity.

We used our econometric models for Westfield to allow us to isolate and quantify the business impact of both the print and digital elements of the campaign, and the combined benefit of a newsbrand partnership.

In this case study the whole was greater than the sum of the parts – the combined effects of print, digital and social were 50% more effective than using each of these channels in isolation – showing the importance of having digital newsbrands in the mix.

On to step two. Digital attribution is a perfect way to understand the role of individual newsbrands within a digital consumer journey and the effectiveness of newsbrands at generating a digital conversion, relative to other types of display activity.

While it’s possible to debate the pros and cons of attribution models, some are definitely better than others. If they’re data driven, take all non-converting journeys into account and include direct site visits and SEO, as well as paid media, then they can clearly provide valuable insight.

We worked with Visual IQ – an independent market leading digital attribution provider – to measure and optimise O2’s digital marketing activity.

By thinking about the measurement and the data requirements in advance we were able to isolate and compare digital newsbrands vs. other digital display activity. We found that digital newsbrands are 10% more cost-efficient at delivering viewable ads and equally efficient as other elements of the campaign at generating online orders.

We work in an industry which is rightly obsessed with measurement, but how can you measure something if you don’t think about it in advance and put the time, effort and data in place to make it happen? This project served as a reminder of getting that right – and that difficult questions can be answered with a bit of creative thinking and rigorous analysis.

Both these case studies demonstrate that digital newsbrands can be effective and should be considered as part of a marketing plan. They also show that if we stop thinking about digital as that amorphous mass then we can learn new things, demonstrate value to brands and make their communications more effective.

After all, that’s the real benefit of solving the measurement problem.

Ben Dudley is head of data modelling at Havas Media Group