Everybody is looking for a quick fix when embarking on a new Marketing ROI Program. They think the latest trendy platform or the most recent popular tool will cure their less-than-stellar ROI Marketing results. The fact remains that strategic, effective ROI Marketing Programs are complex and require solid foundational pillars for successful implementation. Through benchmarking and numerous worldwide, industry-specific case studies, Marketscience has identified the five critical pillars that drive an effective ROI Program. They are:
- Problem-solving consensus on desired outcome(s)
- An organized data process
- A cross-functional team and management buy-in
- Appropriate, transparent, and vetted methods and tools
- Outside expertise
In this blog post, we will embark on a deep dive into the five pillars. We will examine them one by one with a detailed description and an example, as outlined below.
1. Problem-solving consensus on desired outcome(s)
Conventional wisdom argues that revenue, CAC/LTV, awareness, web traffic, brand equity, etc. are the outcome metrics influenced by marketing. While sometimes affected by such, the reality is that marketing is frequently tasked with influencing multiple outcomes which are expected to result in an increased economic value. There is often internal debate about how marketing can or should influence ROI and where sales, distribution, product and pricing take over.
Example– Outcome metrics in consumer lending
A good example is in the area of consumer lending. Here marketing can drive interest or applications. However, there can be long lag times, as the activity that drives the economic value, like funded loans, is a lengthy process. Furthermore, sales processes and credit decisions might influence the extent to which applications will convert. On the flip side, sales or credit committees might argue that it is marketing’s job to bring high value applications and/or quality applications that are likely to convert.
The reality is most marketing organizations are tasked with several overall value-driving concepts like:
- Drive traffic to sales channels like their website, store, call center, affiliate etc
- Acquire new customers
- Enhance the value of existing customers by deepening, cross-selling or retention
- Enhance the overall customer value profile by acquiring high-value customers
- Build brand equity that will facilitate long-term growth, preference and premium price
Before embarking on a marketing ROI initiative, it is imperative that we have an understanding of what we are measuring. In other words, which of these marketing influences, and their subsequent outcomes, are we evaluating? By beginning with the end in mind, i.e. clearly understanding the desired outcome, Marketscience is able to build a solid ROI Marketing Program that has quantifiable and measured results.
2. An organized data process.
A good marketing ROI program is dependent upon the acquisition of comprehensive quality data; however, there are a number of challenges in compiling such a database. We find the following three to be the most complex:
1. Building a comprehensive view
Robust marketing ROI models require a historical, multi-year collection of data, an analysis of outcomes, and an evaluation of other control variables. This can pose a challenge due to spending and marketing-pressure variables, differing metrics and granularity, a plethora of marketing vendors and platforms (each with unique data encoding), and potential agency changes over the data period, coupled with data breaks or missing data.
Exceptional ROI models also require the aggregation of non-marketing data, like our outcome variable paired with company-controlled data streams, which can include but are not limited to margins and LTV, pricing, promo, distribution, and product information. These data streams are potentially captured by competing systems or they are poorly measured, requiring the engagement of other organizational stakeholders like sales, finance and IT.
Finally, the compilation of data on non-controllable drivers such as weather, macro, competitive events, COVID, etc. is also required. This information often comes from a range of sources, both public and commercial, and may differ by industry and geography.
2. Creating the appropriate level of granularity and disaggregation
The right level of data aggregation/disaggregation is paramount when collecting data and building effective marketing ROI capabilities. Underlying data should be captured at a high level of disaggregation to facilitate aggregation across relevant dimensions. The extent to which models can be disaggregated has profound implications on the types of insights that can be provided. The following questions, and subsequent answers, should be included in any analysis:
- Which marketing programs drive higher LTV customer acquisition?
Programs that require customer data to be segmented by LTV tier and potentially marketing data by target.
- How did marketing performance differ across geography?
Performance should be measured by marketing and sales data disaggregated by geography e.g. DMA.
- What is the right mix of promo, product feature and brand messaging?
Marketing data across channels should be encoded by common message buckets to determine such.
Observation frequency is an additional and important aspect of data granularity. Historically, companies have relied on weekly data models. Of late, there is an increased focus on unique daily or within-a-day patterns, that along with tech-enabled marketing innovations, allow for more granular observations and learning. In addition, high-frequency data is useful for marketers with unique seasonalities or compressed selling seasons e.g. sports betting, tax preparers, retailers and certain insurance products.
3. Geographic granularity
Creating geographic granularity in the data improves analyses usefulness. Nationally-distributed marketers, with a national consumer footprint in both competitive and macro environments, differ across geographies and consequently marketing effectiveness is likely to change based on locality. Furthermore, high geographic granularity also provides additional data points, which create unique opportunities to experiment and to create laser-focused test-and-learn pilots at a geographical level. These can then be analyzed in the modeling framework and acted upon appropriately.
Building a repeatable process
During a first cycle, the initial data mart buildout can be difficult and time-consuming for some of the reasons outlined above. The goal of any data process is not just to get it right one time but to improve the timeliness and efficiency of the process by:
- Building guidelines, templates and encoding standards
- Automating feeds and processing efficiently
- Ensuring one source of truth for data points and metrics
As a repeatable data process is established, the process will become faster and easier to run, which will facilitate higher quality and more timely insights. In turn, replicable systems will promote faster course corrections and/or facilitate moving from annual to quarterly or monthly updates.
3. A cross functional team and management buy in.
Marketing ROI capabilities are not meant as marketing’s weapon to attack finance, or finance’s tool to beat up on marketing. The goal of these capabilities is to provide an analytically-driven foundation to encourage rational planning discussions. To do this, the organization’s different stakeholder groups must engage and buy into the process, method, and outputs. This is achieved through:
Credible approach
Use a modeling approach that is not only well-tested with clients, but also extensively reviewed by academia. It is helpful to have necessary regulators or via model governance reviews. For example, in industries like financial services, it is best to have experts “chime-in.”
Early engagement
It is important to involve the key stakeholder groups early in the process and operate in an agile model that allows stakeholders to influence the project’s direction. In other words, it is paramount that the decision-makers have an under-the-hood understanding of the methodology to alleviate concerns and to drive buy-through at each stage.
Transparency
Trust is crucial. There should be no black boxes. Methods, data, and technical details are freely shared and discussed. And when appropriate, client analytics teams are trained and informed of all aspects of the work under a “Do-It-With-Me” model.
Ultimately, success is driven by how the project is managed and how different stakeholder groups collaborate in the process. Having good data, a good model, or a cool tool is not the true goal, but rather having a solution that enables improved decision-making and greater marketing productivity is the objective. The graphic below illustrates the typical collaboration between major stakeholder groups.
4. Appropriate, transparent and vetted methods and tools.
As the discipline of market science matures, and the panacea of overhyped black boxes fail to deliver, it’s important to have a rational conversation about what methods and tools are appropriate for the task at hand. Marketers are often tempted into believing the latest tool or method will solve all the problems, whether it’s MTA, AI, machine learning, MMM, etc.
Whereas, marketers should ask the following questions to ensure the methods employed are appropriate and transparent:
- Again, what elements and which tools and methods are appropriate for the task at hand?
- How has the methodology worked thus far? And what are the theoretical underpinnings of said methodology?
- What are appropriate uses and limitations of this method?
- Do these methods withstand non-biased scrutiny, like model governance or academic reviews?
In addition to business vetting, internal teams and/or independent outside parties should conduct a technical review of the method’s appropriateness. Technical or methodological weaknesses can result in biased or incorrect inferences, which would result in counter productive business decisions. Examples of topics that should be addressed are:
- Estimation methods
- Variable selection
- Causality and endogeneity
- Non-stationarity and errors
- Sampling and validation
- Unobserved components and missing data
5. Outside expertise
Different organizations have varying capabilities and expertise available to support marketing ROI initiatives. Some organizations have little technical resources available with no prior experience and require a full-service outside partner to steer them through the process from start to finish. In contrast, other experienced organizations have available teams of data scientists and analysts running the processes and they might only need minor consulting concerning best practices, technology licensing, or benchmarks.
Either way, it is important that any outside expertise engaged meet the client organization at its point of need and not force feed an off-the-shelf solution. However, it should not be overlooked that there is value in engaging outside assistance for programs as they ensure:
- Access to best practices
- Experience from other organizations
- Neutral perspectives of internal biases and politics
- Dedicated resources
- Benchmarks
- Specialist knowledge
Increasingly, clients evaluate their current state of data, organization, process, and method and then engage outside parties to help them advance. As marketing analytics mature and organizations build extensive internal analytical capabilities, the role of the outside consultant goes from Do-It-For-Me to more of a Do-It-With-Me and sometimes is even a supporter of the Do-It-Yourself model.
For more information about how we can help you implement a successful marketing ROI program, please contact us.
