Big measurement programs are often conducted every year or two to look back holistically across activity. But what are these retrospective analyses telling us? That what we did two years ago was a good idea? Or what we’re doing now is the way forward? It’s often the former and rarely the latter. So, what should we measure, and how should we measure it?
Thibault Labarre, Partner & Marketing Measurement & Optimization (MMO) Lead, Ekimetrics
Measuring yes! But, what?
Optimizing the whole budget means using a holistic measurement solution, where the complete view underpins the more granular view. Very granular measurement facilitates very granular decisions, but can take away from the broader strategic decisions. For example, figuring out where to place a particular creative execution in a digital campaign can support optimization within that campaign, but if the budget allocation as a whole isn’t optimal, campaign or ad level optimizations will have limited impact on effectiveness overall.
What’s more, using a top-down approach ensures you have a credible starting point when it comes to individual campaign efficiency, making the measured impacts of specialist channels and campaign elements both more usable and more stable, especially when new data sources are introduced. Then you can use granularity to make sure that every last penny is efficient, having made sure you’re not wasting large pots of budget that can be used to better effect elsewhere.
The more you can measure, the better. However…
When it comes to interpretation of the data, it’s imperative to understand that:
Further, it’s important to look beyond advertising activity at what is driving business. Marketing is likely to account for driving business 5-40% of the time, which means 60-95% of the time, sales aren’t driven by advertising. For example, in FMCG we see distribution increases for smaller brands being much more impactful than increases in advertising, especially from a smaller budget; this can lead to stronger investment behind customer insights, NPD development, and the materials to engage buyers, rather than increasing advertising spend. Being able to quantify and act on this is critical.
Finally, it’s essential that measurement can be translated to decision-making. Having robust and stable predictive models is critical to turning that into prescriptive and quantified recommendations. Understanding the structure of your business’ marketplace, its place in that market, and your customer, allows you to be more reactive as you can see the fundamental shifts in the things that impact your business and react to them as they change and evolve. Marketers are not just expected to use multiple measurement techniques and metric tracking to understand the value they are adding to the budget spent, competition and market forces mean it’s essential. And it’s a specialist job!
The key for marketers: using the right measures of success ROI
It is often imbued with a singularity of meaning, primarily because reducing marketing performance to a single measure makes it easy for others to judge. When in fact, ROI can mean many things and it’s complex.
As advertisers embraced digital marketing to drive more immediate measurability that delivered immediate results, longer term measures of sustainable business were forgotten. From cost per click (CPC) or cost per thousand impressions (CPM) to cost per acquisition (CPA) and more recently Share of Search, digital marketing spawned myriad levers and measures. ROI and ROAS has often overtaken measures of lifetime value (LTV) and margin. Understanding how ROI is being calculated – what “return” is being measured – is critically important in its interpretation. It can be used with any metric, meaning ROI is not necessarily a monetary value.
The return could be numbers of leads, sales, website traffic or more, and that can also mean there is no break-even point. ROIs that use different metrics are not comparable, nor are they anything more than a measure of that specific activity or lever. Loose use of the same language to mean different things means ROI measures are often conflated. The context of ROI is everything. Big decisions may rest on calculated ROI and this reductive approach to a single measure may mean those decisions don’t have the expected effect. Plus ROI looks backwards, rather than forwards. This forced change in how we measure effectiveness provides businesses with the opportunity to take a breath and reconsider what they are measuring, what they hope to learn and why it’s important.
So how do you set yourself up for success?
Triangulation of methods and unification of measurement!
Each method has its own merits, and each gives a different view on performance across different timelines and different levers. This can often mean mixed messaging from measurement in answer to the same ROI question. Measurement of one type of activity without reference to others can never deliver an optimized budget and is more likely to result in competition for budget between specialist areas, without a view on how each contributes to ROI, profit or LTV.
A combination of measurement methods can be used to excellent effect in an effectiveness eco-system, with each suited to specific circumstances.
There is no ‘one-size fits all’ approach to measurement. Each business has its own peculiarities that need to be addressed. However, there is a common approach that can sensibly be applied anywhere; corralling different methods into a single framework can help to draw out the inconsistencies while delivering the benefits of each. Econometrics is a time-series based approach that has been around for decades. In its simplest form it relies on correlations to isolate the impact of different levers on business KPIs. Holistic media measurement, such as Marketing Mix Modelling (MMM), can be time-intensive, requires data from many sources and only delivers a certain level of depth of understanding. It had, for many, fallen out of favour due to the time it takes to realise results compared with the apparent immediacy of digital analytics.
Fortunately, these branches of marketing science have moved on considerably with modern approaches to MMM providing an holistic, and increasingly, granular and timely approach to measurement. This brings considerable value in the face of the focus on privacy, helping marketers to use data in a compliant manner to plan and direct budget for best effect. Multi-touch attribution relies on digital signals to track the direct path a consumer takes to purchase. And while it can never encompass all marketing, a version of attribution has its place in the use of those digital signals and the ongoing optimization during digital campaigns. However, it cannot exist alone or in a silo if it’s to deliver true value.
In addition, test and control experimental methods can be used, often available through large publisher platforms, where they can easily control who sees which ads in order to isolate the impact of specific activity. These experiments help to understand the drivers of incrementality. For example, you could look at users and user characteristics to understand uplift. Or you could use Google’s geo-experimentation, which splits geographical areas into discrete locations, or General Marketing Areas, with boundaries drawn to avoid highly populated areas and reduce geographical contamination while increasing confidence in where ads have been served. For many firms, there may be a skills gap in delivering all three, with Google reporting that while 70% of advertisers use attribution models to influence budget decisions, less than 20% run incrementality experiments. However, large publishers like Google often fund studies via partners, such as Ekimetrics!
3 key pillars to successful measurement in a cookieless world: MMM, Attribution, and Test & Control
MMM, using econometrics, captures the full range of business drivers as an integration core for the measurement approach. The integration of walled garden attribution solutions allows for the opportunity to dig deep into the granularity and make granular decisions about campaigns, while bridging the gap between the buying options and econometrics-led measurement. Using Test & Control means you can identify causal impacts through experiments and really understand true incrementality. This combination of methods within the common framework allows brands to support decisions from strategy to tactics, budget planning and execution.