Data has never been more integral to more aspects of marketers’ businesses, but the space has never been more complex and dynamic. If you are reading this article, then I assume you recognise the importance of a well-constructed, holistic marketing data strategy. But what goes into such a data strategy?
While each business must titrate the elements of their approach to fit their objectives, there are foundational pillars that I believe should always be present in some form. Ensuring coverage and intent behind these provides a solid bedrock from which to form more nuanced tactics for specific marketing objectives and organisational needs. These pillars are complementary, ranging from foundation to end-point application. Let’s dive into each in more detail:
Technology stack strategy
First and foremost, it is imperative that marketers take an organised approach towards building their marketing technology stack. Technology selection, whether purchased or built, should be informed by one’s objectives, desired use cases in service of those objectives, and the talent one has or will have to operate the technology in-house or via partners. That said, there are guiding principles I believe to be consistent for all brands. They are as follows:
Completeness: This refers to the completeness of the stack relative to marketing goals. Specifically, the question marketers should ask is whether the technology allows them to achieve said goals with greater effectiveness in terms of campaign performance and operational efficiency. Those goals will often vary by brand. For example, the needs of a medium-sized local marketer operating primarily out of their website, whose goal is to improve online revenue, would be very different from that of a multinational brand, with a multitude of paid and owned communication channels, looking to build a single customer view for analysis, insight generation and activation.
Modular, flexible, built-to-specifications: As an organisation evolves, so too does its marketing needs. Rather than a single one-size-fits-all architecture, which rarely fits anything for very long, I recommend a modular approach that begins with a foundation on top of which one is able to build and add elements that are not only additive, but synergistic with the existing infrastructure. This should be done in a way that drives value from day one while simultaneously building towards longer-term objectives.
Synergistic: Each component should work seamlessly with others to form a whole greater than the sum of its parts. This synergy can exist either as a function of technology selection or stack design. For example, one can select multiple elements from a single marketing cloud, combine technology from different platforms, or even build components in-house assuming adequate engineering and development resources.
Ways of working: A well-planned stack architecture will be largely for naught without multiple parties working behind a shared vision of how the technology and data will be put to use. As such, it is imperative that organisational processes between key stakeholders for different stack components be aligned to facilitate shared goals, synergy and results.
Data collection and tagging health
With the foundations of the technology stack in place, the next step is to ensure data is being captured in a consistent and robust manner. The degree and sophistication of one’s data collection and plumbing setup will vary in accordance with the number of touchpoints and the volume of data flowing through.
Beyond technical setup and plumbing, this also encompasses operational and best practices to ensure data is collected in a format that lends itself to use. This includes everything from processes to enable hygiene such as tag and audience naming convention, to efficient and consistent platform and campaign setup structures, such that collected data can be easily mapped back to the activity that is generating it.
For marketers collecting and centralisting data in data lakes or data warehouses, the pipelines feeding the data in for centralisation, and out for deployment, must also be engineered and maintained in an effective, scalable fashion. Most importantly, marketers must ensure that everything above is done in a manner consistent with current regional and local legislation. Capturing data shared and generated from customers for operational purposes is one thing, but consent for marketing applications is another. It is up to each marketer and their agencies to do their due diligence.
Audience management and engineering
After data is collected and centralised, it should be organised and segmented in a way that facilitates use cases across marketing, creative and media teams.
This can be broken down into several parts. The first is managing the collected data, with the exact mechanics being different depending on the technology. This can include a cloud-based data warehouse, a customer data platform (see explainer), site or app analytics software, or even media buying platforms, as most programmatic platforms today have audience creation or organisation functionalities. Regardless of the underpinning technology, marketers should maintain a centralised view with clear processes to ensure data veracity and freshness. Similar to data collection, governance and hygiene are also paramount at this stage. Marketers should combine technology and automation with codified processes to ensure data accuracy, quality and adherence aligning with established taxonomies and conventions.
Once the collected data has been cataloged and set up for work, one can use the data to create audience segments in accordance with marketing use cases. There are a myriad of methods to do this, code or no-code, heuristic or data-driven, but the key is to have a framework through which deployable audiences are created. For example, a marketer with performance objectives may want to align their baseline segmentation taxonomy with different customer states of their sales journey, leveraging a combination of rules-based and model-based clustering to derive segments for each funnel stage.
Beyond first-party data, marketers can incorporate value-adding external data (second or third-party data) to supplement and fill gaps depending on use case. The extent to which this is needed depends on the marketer. Larger brands with more customer data will likely require less, whereas smaller brands still building their customer data pool will likely need more. That said, external data, whether for direct deployment, or to model or amplify existing customer data, can almost always add value when used judiciously. As such, just as brands should have a clear taxonomy for organising customer data, they should have a framework for evaluating and organising external data sources such that deployment is at the ready and well-informed.
One of the unintended consequences of the programmatic media era is that it has trained an entire generation of marketers to equate data use with media targeting, but that is just one of many angles in the prism of data use. Using data to address users across media channels is but one of many ways to deploy data. Marketers and their agencies should construct codified processes for leveraging data not just to augment media campaigns, but also to generate insights to inform upstream business and strategic planning. These processes should include not only first-party data sources, but also external data such as those from larger, more representative but less granular sources, such as syndicated panels and population-level data, to calibrate and reduce the risk of bias that may come with the more granular but narrow event-level data that marketers collect.
Marketers and their agencies should have an organised approach that applies the previously established data taxonomy to their media planning process. For example, there should be processes for selecting data segments depending on the execution scenario across the customer journey. Tag-based audiences can be applied to media platform line items in a straightforward manner, whereas data from a customer relationship management system or data warehouse will require a more involved pipelining process. The key is that there is a holistic framework that covers all viable sources.
Aside from media communications, brands should place equal emphasis on the ways data can be used to augment creative messaging. There are many angles from which to approach this, whether it be leveraging audience insights to inform creative strategy and territories, or merging audience and contextual signals via creative technology for dynamic, personalised experiences across paid and owned touchpoints. The key point is that a well-planned data-driven media campaign loses much of its impact if the messaging does not match the targeting.
It is probably not an overstatement to say that this is the most pivotal period in the history of the industry in terms of shifts in favour of user privacy and consent. A future-resilient data strategy refers to operational and technological preparedness in response to current and impending changes to the ability for brands to collect data against, attribute, measure, analyse and target users and audiences.
While much remains uncertain, we can begin mapping a number of trends with a fair degree of directional certainty. As such, it is paramount that brands and their agencies take these trends into account across each of the previous pillars to maximise future-resiliency. Several such trends include an increased focus and need for more persistent forms of first-party data, greater reliance on probabilistic methods, technology with a privacy by design approach such as data clean rooms and federated learning, the growing integrations between advertising technology and managed cloud systems, and many more. While it is impossible to predict the future, marketers who weave these considerations into their data strategy roadmaps while maintaining the agility to align with future developments will be well-positioned to thrive both today and tomorrow.
Vincent Niou is VP of data and technology, APAC at Essence.