Handing over MMM data like a professional

Welcome to the Wheelhouse, a popular series of blogs from Ebiquity’s Marketing Effectiveness team.

In this latest edition of The Wheelhouse blog series, group Director Nic Pietersma considers what should happen when, as invariably happens, relationships come to a natural end. He focuses on improving handover between one MMM provider and the next, in the best interests of the advertisers.

Even the best teams lose clients. It’s normal, unavoidable, and arguably healthy for advertisers. Always having the same perspective can stifle growth and innovation, and fresh blood is helpful from time to time. We in the Marketing Mix Modelling (MMM) industry need to do a much better job of handing over the reins to our clients’ new econometrics partners, when and if the time comes.?

For context, MMM measures the return on investment of advertising investment, while controlling for price, market forces, and other internal and external factors. To do this reliably, we need to collect at least three-to-four years’ worth of data for the first round, and the data sets accumulate over time from there.?

All the data needs to be cleaned, validated, classified and transformed into a usable format. Emails cross between the MMM partner, the advertiser client, and their agency partners to confirm such details as the buying audience, whether the spend includes taxes or not, and to double-check that one funny thing – that curious, unexplained outlier in the data – and so on.?

As a by-product of this process, good MMM partners can quickly become a kind of data archaeologist. We support and reinforce institutional memory when internal promotions, churn, and agency hopping wipes the slate clean every few years. Advertisers may find that their MMM partners is the only place they can find out about campaigns that happened 12 years ago. Truly a lifetime in marketing … when the Blackberry was king and 80% reach on linear TV was feasible.??

Building MMM datasets can be painful and time-consuming. This task accounts for most of the hours for which clients pay their analytics partners. But this schlepwork also genuinely adds value. Not only in the MMM itself but also in ad-hoc analysis. Your MMM partner is the first place you would go to answer a question like: “What happened to share of voice between 2010 and 2023?”.?

Today, when an MMM account changes hands there is a good chance that the client will be given back the raw datasets the vendor originally received. Years of work unprocessed, unvalidated, with gaps, in various formats, without email chains for context or adequate notes. It’s a handover strategy that is value destroying for the client, and even though it’s the norm, we can all do better than that. Norms can and do change. When we talk about ‘data handover’, that should mean clean datasets that can be used to build models, not an avalanche of file formats from the last ten years.?

Summing up: good practice in handover?

Advertisers: don’t leave it to chance or good will. The incumbent shouldn’t be invited to pitch unless they commit to data portability. At a minimum, this means clean input datasets, model results, a list of variables used, and transformations. Be explicit about what you want in the pitch process.?

Practitioners: this is the better way. Making a commitment to data portability is the professional move and is certain to get you a call back next time the pitch comes around.?