Media Management

The ROI of measuring ROI: considerations when contracting platforms

Project vs Subscription Model

Change the procurement from a project to a subscription and the incremental cost of another MMM project can be driven down to nearly zero. No pre-determined scope of work, no bureaucracy to initiate a new MMM project creates many more models produced.

Procurement of Market Mix Models (MMM) needs to view each proposal as either a Platform or a Project. MMM is now available through a Service as a Software model which dramatically changes pricing, function and scope. These changes make the procurement officer’s job more complex as they are radically different engagements which makes cost/benefit analysis much more challenging.

Measuring ROI has a cost. That investment either pays out or it doesn’t. To improve the ROI of measuring ROI, reduce the cost or improve the return. Simple, but it misses the point of what happens when the methods of producing ROI calculations change.

Chris Williams will be presenting a session on MMM at the Marketing Procurement & Financial Management Summit in Toronto on 22nd October. Click here to find out how to book a free place.

Typically, MMM was bought on a project basis. Write a scope of work, RFP it to multiple vendors, choose the best price for alignment to needs and draw up a contract. Now with MMM offered on a subscription SaaS basis, the marketer can dispense with writing up a scope of work for each MMM project.

It’s the cab vs car decision. Once you have a platform (car) you use it far more often than you would a project (cab). Cabs are easy to define from a cost, scope and benefit perspective. Of course, cars can be quantified in terms of cost, use and benefits however there is elasticity in the scope. You use a car more often and find new uses for it and then there are other members in the household who help to amplify benefits which amortize costs.

Think of it this way. If Marketing Mix Models are bought as a project the price tag could be upwards of $200,000 and the scope of work defines how many models are produced. Cost per model is known and so the Return on Investment of measuring Return on Investment can be calculated. Your $200,000 MMM project must return an incremental $2,000,000 in top line sales if the profit margin is 10%. But what if there were many more projects to include?

Change the procurement from a project to a subscription and the incremental cost of another MMM project can be driven down to nearly zero. No pre-determined scope of work, no bureaucracy to initiate a new MMM project creates many more models produced.

If the cost is nearly zero then the threshold which makes the Return valuable also drops, which dramatically expands the number of insights which can be applied to the marketing plans. Cost per model down, number of insights up yields a dramatic change in the ROI of measuring ROI.

Who does the work also changes the engagement. In the car/cab analogy, the cab comes with a driver which is part of the cost. With Market Mix Models as a subscription not only are you the driver but there are suddenly multiple cars which can be driven by staff. Each staff member can extract and apply the value from the MMM data science to their specific task or area.

The skills required to produce insights for marketing are reduced because the complicated math is handled in the cloud instead of custom built. Just as desktop publishing overtook typesetting, the technology enabled mere mortals to produce polished professional grade publications.

Again, the value of the platform is magnified through scale of users and usages.

Market mix models can be a standalone solution or they can be fully integrated into downstream tasks. The value of the MMM is magnified when its output is seamlessly pushed into the media planning tool. The media planning tool is the hub where the entire media budget is being allocated across channels and providers.

Any increases in effectiveness and efficiency suggested by the MMM are put into action against the largest spend in the marketer’s budget. Ideally the procurement officer is working closely with the marketer to calculate the value captured through reallocations in terms of incremental sales.

Whether the Market Mix Model is produced manually or in the cloud has huge impact on scope of engagement and how much benefit can be produced by measuring advertising return on investment.

So Cloud or Manual?

Who produces the MMM – an agency, the advertiser or a 3rd party? Let’s examine the options.

Third Party: Advertiser contracts MMM platform directly and Platform provides service. Agency SOW required to provide data. Upside: Inexpensive but does have additional cost, data succession, works with media plan – if media agency subscribes

This is why Arima moved from a project price to a subscription model for Marketing Mix Models. We wanted to create incentives for continually experimenting by pushing new hypotheses through MMM. Some cloud-based MMM providers have stuck to the established project-based business which expands their margin in the short term but doesn’t allow them to participate in new developments that integrate MMM into the media planning workflow.

Procurement must also pay specific attention to data succession terms and conditions. Once an MMM is established by the marketer, it becomes a tool to compare periods over the long term. Continuity is key, so it is up to the procurement officer to ensure the advertiser is not locked into one specific arrangement. Legal clauses must be written into the contract to ensure raw data and the model can transfer from the current provider to the next. Key to that is avoiding any data the agency determines is proprietary to them.

The advertiser may believe that all campaign data belongs to them however agencies may take the position that certain media information is theirs. Often, CPM or GRP rates are negotiated specifically for each agency, and therefore any models that include media costs and impressions or GRPs may be withheld. Before including this granular data in the MMMs, question whether it is necessary for the purposes of calculating ROI (it’s not) and whether its inclusion would prevent data and model succession.

However, some marketers stick with a low-cost digital attribution approach to measuring ROI. The assumption is any insight generated yields a high ROI on measuring ROI because the data and reporting from the digital platforms is free. What’s missing is whether the measurement of Return is correct or misleading.

Many brands using this approach find themselves over-invested in digital media, all claiming excellent (and proprietary) outcome numbers while their business suffers. In other words, the cost of calculating ROI is extremely low, but the insights are leading to enormous waste in the media budget, essentially a negative ROI on measuring ROI. Here it is the role of the procurement officer to propose the “pound smart – penny foolish” approach to acquiring technology, tools and processes that provide actual sales gains instead of vanity metrics.

Marketers aiming to improve what they get from measurement schemes; the ROI of ROI, need to focus on the business model in which the technologies are offered. Too often the focus is on the math, the methodology, price or monstrous size of the analytics firm. Marketers who focus on the number of insights generated and the ease of application will yield greater marketing results.

About the author

Chris Williams is CMO of Arima Data

Chris served as VP Digital at Association of Canadian Advertisers (ACA) for over 4 years. During that time, he was one of the authors of ACA’s Marketing Communications Service Agreement in addition to working with the Media Ratings Council (MRC) on audits and accreditations of major digital platforms. During his 30 years in digital media, he was President of IAB Canada, a Board Member of the Destination Ontario and participant in numerous trade associations.

He is currently the CMO at Arimadata, an advanced data science platform which offers self-directed Market Mix Models integrated with Canada’s first synthetic data population for marketers