Data strategists at SSW are responsible for ensuring the “first mile” of a project lays the groundwork for a successful “last mile.”
Over the last five to ten years, data science has begun to enter the corporate mainstream in earnest. Executive decision-makers are increasingly recognizing the value of data-driven operations, gravitating away from the longstanding — and fatally flawed — HiPPO (highest paid person’s opinion) approach to strategy. But despite this high-level progress, few organizations have managed to build data science architectures that successfully bridge the deep-seated divides that have traditionally delineated departmental fault lines.
As Harvard Business Review Senior Editor Scott Berinato recently pointed out, “The presentation of data science to lay audiences — the last mile — hasn’t evolved as rapidly or as fully as the science’s technical part…Until companies can successfully traverse that last mile, data science teams will under deliver.”
To wit, no matter how sophisticated a data science team’s analyses are, no matter how great its effect size or p-values may be, its number crunching is only valuable if it is (1) performed in service of a key business question (KBQ) and (2) communicated in a manner that is intelligible to nontechnical stakeholders. While there is no shortage of discourse on the latter, it is the former that is a precondition not only of clear cross-expertise communication, but of successful data science at large.
In short, while Berinato (and many others) is correct that traversing the last mile of data science is essential to driving meaningful results, in nearly every scenario, a successful last mile is almost entirely contingent upon a successful first mile.
Keeping the Narrative Chain Intact
In many ways, last mile success can be reduced to a question of expectations. Reaching across data literacy gaps is important, to be sure, but clarity and relevance are coequal goals. It’s one thing for a group of nontechnical stakeholders to understand an analytics presentation; it’s an entirely different thing for them to take interest in (let alone appreciate) its core insights.
As such, to ensure its first mile is charted along the appropriate trajectory, every data science project should be designed in such a way that its architects are able to answer a straightforward but absolutely critical question: will this analysis lead to actionable results that address our key business questions? If the answer is, “No,” the project is fated for a chilly reception from business-minded stakeholders who have little time for analytics for analytics’ sake.
As I’ve written about before, at Saatchi & Saatchi Wellness (SSW), we begin each and every one of our analytics projects by conducting an in-depth discussion with our client to tease out the specific strategic objectives they want to accomplish. These objectives serve as a starting point from which we can build out a series of data analyses that, while highly technical, speak to the client’s actual business needs.
At this point, the challenge becomes conducting these analyses without losing sight of the real-world concerns that inspired them. Ideally, the grammar of the first and last miles of data science will be nearly identical — the latter consists of presenting answers to questions posed in the former, after all — but achieving this alignment requires deftly executing a series of “translations” across the intermediary miles.
Overseeing the translations from KBQ to database query to algorithmic analysis to data visualization to boardroom presentation — all while keeping the narrative chain intact — is a job unto itself, one which, at SSW, falls under the banner of data strategy.
End-to-End Visibility: The Key to Effective Data Strategy
If data analytics is a team sport, then an organization’s data strategist is its quarterback (or middle linebacker or point guard or coxswain — take your pick of sporting analogies). At its core, data strategy is the art of translating plain language business problems into the language of analytics and back again, which is why data strategists are optimally positioned to ensure the last mile of a data science project actually reflects the KBQs posed during the first mile of the project.
In order to orchestrate this alignment, a data strategist must be afforded the leeway — and must possess the requisite skill-set — to “call plays” during every stage of a project. Data strategists should take an active role in formulating a project’s KBQs and crafting its tactical contours (integrating input from both business and analytics stakeholders), but they should also be tasked with carrying these questions and tactics along throughout the entire process.
Oftentimes, an analytics team will include a data strategist in the KBQ creation process only to then hand off the project to data engineers and data scientists, who eventually hand off their insights to data visualization experts. The data strategist might get one last crack at the final product before it’s presented to a client or internal decision-making team, but their exclusion from the intermediary steps of the process severely compromises their ability to ensure this product speaks to the project’s KBQs — an alignment which is the crux of the project’s reception by its ultimate audience.
Facilitating Others’ Successes
And while it’s important to include data strategists in conversations taking place at a project’s beginning, middle, and end, it’s equally important to acknowledge that mature data strategy is an iterative process — a journey, not a destination. Especially when operating within the context of an ongoing relationship — i.e. not a one-off project — a data strategist should be constantly experimenting with their approach, collaborating with various data science stakeholders to generate continuous improvements to the analytics process.
New questions will arise as results are delivered to key stakeholders, priorities will shift in response to competitors’ market activity, and technological developments will necessitate organizational evolution, and data strategists must account for all these changes in as close to real time as possible. As such, viewed through the prism of data strategy, there is arguably no distinct “first mile” or “last mile,” but rather a dynamic feedback loop that drives results through an iterative, self-informing process.
At the end of the day, it falls to data strategists to maintain the integrity of this loop in a way that enables other stakeholders to hop on board for a cycle or two without becoming debilitatingly disoriented by a rush of unfamiliar — or worse yet, irrelevant — information. Ideally, a data strategist will have enough breathing room to fine-tune this loop over time, but in the context of a finite project, it’s incumbent upon data strategists to do everything in their power to facilitate the alignment of the project’s first and last miles. To paraphrase Berinato, only then will data science cease under-delivering.