Analysis: The four key methods for marketing measurement to maximise impact

Joseph Enever is a Research Director covering Marketing Data and Analytics. Enever is focused on how marketing teams can evolve their marketing data and analytics capability, and navigate the key issues and technologies to achieve the best possible results. He has a deep understanding of digital analytics and optimisation, from both digital transformation and practitioner perspectives.

When looking at organisations today, market research and marketing analytics now absorb around 13-16% of marketing operations budgets. However, evidence from Gartner’s Marketing Data and Analytics Survey (client access required) indicates that these investments are mismatched with their output and this presents a significant risk to the ongoing investment.

So, what does one do to find what works and what does not – which marketing activities should be doubled down on and which should be axed?

Standard analytics tools and methods would suffice to count conversions, report on what happened and calculate performance metrics on channels in isolation. The result of this simplicity is a myopic view of marketing’s impact.

Frustration can emerge in teams due to an inability to articulate the full picture – what was the cross-channel impact of our marketing at the customer level? Where organisations maintain this level of analysis because of constrained budget, skills or technology, a ceiling of insight limits them.

Only by leveraging advanced marketing analytics methods and technologies can organisations break through this ceiling of insight and develop more accurate quantification and understanding of the impact of their marketing investments.

To achieve this, marketing leaders should consider the following four marketing measurement methods:

Marketing mix modeling (MMM)

MMM is a top-down methodology that uses aggregate data, such as historical sales, media spend by channel and comparative competitor insights to generate models.

Using MMM reveals the relationships between the marketing goal, like sales, and a range of controllable factors, such as marketing spend and frequency. It can be used to generate a broad range of insights, including the most prominent and sometimes elusive insight, sales and revenue incrementality – a sibling of ROI.

A model, once created, allows marketers to identify the relationships between naturally disjointed channels (like radio broadcast and online search) and outside factors to then holistically quantify the return of diverse marketing investments. With this data, more accurate predictions for the results of media investments in different scenarios and future market conditions can be made.

Multitouch attribution (MTA)

MTA is a bottom-up approach that requires user-level data to identify the relative contributions of consumer touchpoints along the path to a goal.

The method takes the traceable customer journey into account, and attributes fractional contribution – sometimes called data-driven attribution – to the different touchpoints, allowing higher granularity when identifying optimisation opportunities (than is possible with rules-based attribution).

It’s a direct evolution of last-click, rule-based attribution, where value is placed with the channels that are involved immediately before the conversion, such as display retargeting or paid search branded terms.

As such, MTA more accurately allocates value (and importantly highlights where no value is provided) to individual touchpoints in the customer’s pathway to purchase. This makes it a favourable option when looking for insights into the customer pathway to purchase and performing in-flight budget allocation optimisation.

Holdout testing

Holdout testing, often referred to as test and control, is a crucial method for testing hypotheses. Though commonplace in the digital marketing arena, it can provide a statistically significant measure of the effect of a marketing tactic in a relatively simple and cost-effective way.

Holdout testing can be done in both the long-term (with ‘always on’ global holdout segments) and in the short term for tactical insights. Here are some common scenarios:

  • Proving the impact of new marketing channel investment. For example splitting the new marketing channel audience in to test and control groups, then measuring speculative investment campaign outcomes for any significant effects
  • Validating a hypothesis generated by attribution model. For example an attribution model which implies that one channel is better for generating prospective customers relative to another, but is unable to quantify the incremental impact of the channel had no advertising been done
  • Quantifying the impact of online investment across retail stores with a geo-distributed test. For example using a paid search experiment in matched regions, randomly split into two groups – test and control. The test group regions are subjected to an increase in paid search investment, with all the subsequent online and offline data analysed to identify a statistically significant uplift on retail stores sales vs. the control

A test-and-learn culture in marketing measurement facilitates innovation by providing a framework for taking calculated risks, such as new channel investment, and increases the knowledge and intellectual property of marketing analytics teams.

Unified measurement approaches (UMA)

UMA answer questions that span both the tactical and strategic impacts of marketing. These approaches attempt to resolve the challenges of using and relying on sets of insights based on unlinked methodologies. If you are looking to understand the combined impact of online and offline marketing, or connect the context between MMM and MTA, you should look to UMA.

UMA requires a significant commitment, both financially and operationally, since the complexity involved in achieving unified models across MMM and MTA is high.

The core of the challenge lies in acquiring and leveraging the user-level data that can provide continuity between MMM and MTA. If achieved, the resulting insights improve the marketer’s ability to both measure impact holistically and respond comprehensively to scrutiny from other departments, such as finance, and senior stakeholders.

Using one, none or all

A marketer’s approach to measurement methods will be based on different circumstances like the availability and quality of their data, the existing marketing mix, available skills, and budget. Successful marketers use multiple methods because each provides unique insights and addresses different challenges.

However, whichever method a marketer chooses, it is best when used within a wider ecosystem of measurement — not in isolation.

Photo by Daniel Andrade on Unsplash

Interested in hearing leading global brands discuss subjects like this in person? Find out more about Digital Marketing World Forum (#DMWF) Europe, London, North America, and Singapore.  

Tags: , , , ,

View Comments
Leave a comment

One comment on “Analysis: The four key methods for marketing measurement to maximise impact

Leave a Reply

Your email address will not be published. Required fields are marked *