By Dimple Dinesh, marketing science partner for MENA at Meta
Developments in knowledge and processing power give modern marketers an array of tools to optimize for effectiveness and minimize excessive spending of their budgets. As marketers are continuously challenged within their organizations to prove that activity is driving revenue, they can select from many available tools to do so – Marketing Mix Modeling, Attribution, Causal Analysis, Segmentation Analysis, Exploratory Analysis to name a few.
But the measurement of activities within the ecosystem gets even more difficult every day as we live in a complex, multi-screen, multi-device world, constantly evolving. In addition, privacy regulations and restrictions further compound these challenges. As a result, it is tricky to craft a good measurement strategy that shows how a certain marketing activity has performed against specific KPIs. Therefore, a company’s advertising budget might end up being cut simply because its broader impact on marketing is not fully understood.
Making Decisions in Full Knowledge
Henry Ford once said, “A man who stops advertising to save money is like a man who stops a clock to save time.” I’ve been immersed in the digital world for many years and I have seen firsthand how quickly marketers prefer to put a hold on their advertising spend via certain channels when they believe that it does not deliver on a set of metrics. This could be under pressure from a finance team who will rightly make cuts where there is a lack of evidence of return of investment.Marketers should be constantly assessing if they are allocating their advertising budgets efficiently and this will only work if they are using the right metrics to begin with.
Which brings us to the most important question: How do we identify and implement the right effective advertising decisions in light of the ever-changing marketing ecosystem?
The precise answer should always be making decisions driven by accurate data. There are quite a few data-driven methodologies in the toolbox. The most prominent things we have heard advertisers wanting to understand more about in the last 12 months are Marketing Mix Modeling, Attribution, and Incrementality.
Marketers often use the terms Marketing Mix Modeling (MMM) and Multi-touch attribution (MTA) interchangeably when talking about understanding digital value. It is true that both are techniques used for understanding value. Both these methodologies have different outcomes and business cases. MMM is used prominently to measure the omnichannel performance of the overall business activities. In contrast, MTA is designed for digital-only channels and assigns credit to individual consumer paths to purchase touchpoints in their online journey. MMM gives you a top-down, macro-level view of your marketing efforts across all channels, while MTA provides a bottom-up, granular view of your approach for digital-only media.
The Advantages of Marketing Mix Modeling
Highly resilient and using first-party data, MMM is not impacted by privacy regulations and data restriction changes happening today. In addition, it does not rely on individual people, IDs, or consumer journeys but aggregates all the data in time series.
MMMs are used to measure the aggregated incremental contribution of various marketing, non-marketing across channels, and external activities over your base sales (base sales – if you didn’t spend anything at all, the organic sales you would get). The benefits of a MMM study are that it can incorporate many media channels and other influential factors that impact one’s business. For example, seasonality of sales or even the impact of weather where relevant to an advertiser such as a sun cream manufacturer. However, the viability of the study depends on the quantity & quality of data available.
Nevertheless, MMM is a powerful tool to optimize advertising dollars towards efficient channels and forecast future-facing scenarios. MMMs used to be the domain of consumer packaged and autos industries earlier. Today, it is being adopted by many more industries such as Retail, eCommerce, and Mobile App to plan and forecast future revenue. In a nutshell, MMMs are an eagle-eye view of your business.
Moving to Understand Multi-Touch Attribution
MTA is designed for digital-only channels and assigns credit to individual consumer paths to purchase touchpoints in their online journey. These models are developed using a flexible set of ‘rules’ built to measure each digital channel’s contribution along the consumer journey online. Businesses often select the rules for their MTA based on their best understanding of a customer’s path to buying a product.
For instance, a rule might credit the value-driven by your ad only to the last platform on which they saw it – or it could be set to give credit to each platform evenly across everyone who saw the ad. If these rules are set arbitrarily and without validation, these models can be a by-product of correlation and do not reflect the actual causation of activity (and as everyone who has ever listened to a statistician, correlation does not equal causation!). However, rule-based methods can fail to capture the value of many other factors that may have contributed to the desired behavior or to understand if the behavior might have happened because of an advertisement.
Today, MTA models are provided by different publishers (Meta, Google, Adobe) and can also be customized and built by third-party measurement providers to fit the needs of a business. As a result, there is no one-size-fits-all MTA available in the market, and marketers pick models close to reflecting their business model. Validating these models using a causal experiment, set up scientifically is the technique used to know whether assumptions in MTA rules have been correctly set.
Attribution models can be as simple as an in-house system for assigning value to each activity, recorded on an excel spreadsheet, or as complex as a sophisticated model built, tested, and validated by a specialist company with much experience. These models provide the single and micro view of each digital channel and campaign in real-time. However, they do not consider base sales (the organic sales you would get if you didn’t spend anything).
The MTA model will provide valuable insights, but it is also essential to understand its limitations: an MTA model might do an excellent job of evaluating the contributions of branded v.s. non-branded campaigns for advertisers with short purchase cycles. MTA models do not, on the other hand, do an excellent job of evaluating channels designed to produce long-term effects such as awareness. The depreciation of cookies in the digital world increases the challenge, reducing the ability to understand a customer journey across all touchpoints.
Underpin Everything with Incrementality
Incrementality is the term used to describe additional business value that is generated as a direct result of marketing activity. Ever since Wannamaker uttered his notorious words about 50% of a marketing budget being wasted many marketers still struggle to determine if their activity drove impact. Incrementality measurement helps marketers determine just that. Using experimental design – a method borrowed from the medical industry – it compared the actions of people who see the ads versus those who do not. We refer to these as ‘lift studies,’ and they allow advertisers to measure the actual incremental business impact of the activity as they look at groups scientifically designed to be truly representative with sufficient statistical power to be confident in the result. A study of this nature is the gold standard for measuring true incremental business value.
One challenge is that these studies are not multi-media and are done in silos for every media channel. We often see that outputs from MMM and MTA can contradict each other, leaving marketers wondering what to trust and eventually giving up on one of them. That’s precisely when testing for Incrementality enters. As Incrementality shows us the actual causation of our marketing activities based on lift studies, calibrating MMM or MTA by using a lift study alongside these methods is essential to validate the results they show and choose the right one to meet one’s business needs.
There is no bulletproof measurement strategy for navigating changes brought by the ever-expanding marketing landscape and all the changes brought by access to the data needed to do it.The work in getting better and better at understanding value works just like science – new things are learned, tested and implemented. The journey of getting better at understanding value is constant and we will likely not see the end of it. As new marketing opportunities appear, each needs to be understood.
These are some steps that can be taken to help prepare for the upcoming future:
- Invest in building teams with the data skills necessary to run methods that show you where money is being most effectively spent. It will save many multiples in return.
- Do not depend on just one measurement methodology at all times for your business. Different methods have different outputs that will help enhance your business.
- Embrace changes with open arms. Change opens the door to innovation and innovation leads to greater accuracy.
- Incrementality is based on causation. Therefore, trust Incrementality as a fundamental source of truth of impact, and use lift studies to calibrate other results such as MMM or MTA.
These methods are all available to marketers today. Understanding their impact will help deliver a more significant business impact for your marketing spend.Investing time and resources in understanding how to apply these different tools correctly can minimize budget wastage and put your data effectively to work.
