Strategy is a tricky word.
What constitutes a strategy for analytics? Ask a dozen people, get a wide swath of answers. In some contexts, that ambiguity might not matter—as long as you’ve defined a strategy that actually moves your organization toward its goals, great.
However, Jennifer Belissent, Principal Data Strategist at Snowflake, noted that strategy can become a point of paralysis, in part because it’s easy to argue about.
“People get so caught up in having a data strategy that they lose time. They put off doing anything at all, under the guise of defining their strategy,” Belissent said. “I’ve always advocated for an iterative approach, with broad principles and interim deliverables.”
The not-secret secret to getting things done, as Belissent explained, is to define a clear set of terms and your top priorities, and then create a roadmap for achieving those priorities. (Belissent’s post Ensure the View Is Worth the Climb provides more insight into the prioritization process.)
Belissent offered a simple set of clear terms that can help get companies out of strategy paralysis and on to productive work:
- Goals: High-level outcomes you want to achieve, with the top priorities clearly noted
- Objectives: Specific actions taken to achieve those top priority goals
- Milestones: Incremental steps, projects, and priorities
- Metrics: How you measure if you’ve fulfilled your objectives, and ultimately the goal
The roadmap outlines those interim milestones to be delivered. “Don’t get hung up on strategy. Define what you want to do, and how you plan to do it,” Belissent said.
Here are six further qualities to guide the creation of your initial roadmap, reducing the time to value while still guiding the organization toward its overall goals.
A great analytics program roadmap…
1. …starts with a steering committee, and a listening tour
Belissent said many companies find the hardest part is knowing where to start.
A business-focused listening tour within the organization is a great first step.
“Don’t ask peers about their data needs, but rather what keeps them up at night,” Belissent said.
Rich Peters, founder of consultancy Tongere Partners, agreed that the business focus is key to this step. “If you don’t understand the business context, how will you understand the data?” he said. Questions might include how “customer” is defined across multiple business units.
Information gathered on this tour provides a foundation for creating a steering committee and guiding its initial work. The tour will help identify potential high-return projects, as well as the most enthusiastic line-of-business stakeholders who will help champion those projects and possibly serve as data stewards for their specific areas.
The steering committee can use all this information to prioritize the work and identify the next steps on the roadmap.
And the listening tour can even help with funding for analytics work. “By doing it in a coordinated manner, you can figure out everyone who will benefit, and sometimes there is a little tin-cupping: Business units might be willing to chip in” when they discover shared interests in projects initiated elsewhere in the company, Belissent said.
2. …is principle-driven
Belissent advocates for defining a concise set of guiding principles, part of or derived from the data strategy, to help guide roadmap decisions.
For example, principles for the roadmap might include:
– Prioritize revenue growth over operational efficiency. Some organizations might reverse this principle, Belissent noted. It just depends on the current business priorities.
– Utilize existing platforms and analysis tools where possible, and gradually reduce the number of other legacy systems. This principle might be broken where there’s a business case, but the overarching goal might be to limit integration and maintenance demands, while consolidating technical skill sets.
– Create self-service capabilities. A principle such as this suggests creating data catalogs and prioritizing easy-to-use analysis tools, reducing the demand on a limited set of data specialists—who are in scarce supply these days—to answer every question.
Building out the roadmap will inevitably require some tough decisions, choosing to allocate resources to one project over another. Solid guiding principles help make coherent choices based on the best interests of the business.
3. …addresses multiple audiences
A successful analytics program will ultimately need support and participation from all levels of the organization.
That means your roadmap isn’t just a guide for project managers or data stewards. It’s a communication tool as well. As such, a single document isn’t likely to cover all the bases—it would likely have too much detail for some audiences, and too little for others.
It’s more effective to build the roadmap in a few layers.
The top layer is a strategy-level document that connects the roadmap to the business strategy and plans. “Executives care about investment decisions: ‘Am I going to invest these dollars, people, and time in something valuable?’ They don’t care about the minutiae that project managers care about,” Peters said.
The middle layer identifies constraints and dependencies that your analytics projects will encounter. Which capabilities will the organization need to build before taking on more ambitious projects? What are the quality, security, and compliance requirements for different kinds of data? For example, Belissent said, in many marketing contexts 80% data accuracy may be enough to yield useful insights, whereas healthcare data use cases may require 99% or better.
Belissent noted data stewards are very helpful in building this part of the roadmap. Ideally, data stewards sit in between the line of business and IT or an analytics center of excellence (CoE).
The third layer of the roadmap drives down to the project management level of specificity.
4. …is action-oriented, hunting for quick wins
The perfect strategy, the perfect roadmap—experts say these alluring ideas are a mirage and shouldn’t be chased, especially in the early days of an analytics program. It’s more important to get going.
Most organizations of any size already have projects underway. According to participants in a CIO panel discussion on analytics, some have even chosen to delay creating a full-fledged data strategy until they have secured some “quick wins.” Delivering measurable business payoff quickly provides important ammunition for securing the next round of support and investment.
You also shouldn’t wait for perfect data. Peters said some companies have effectively spent a decade or more “getting your master data so screwed up that you can’t answer a basic question.”
“Sometimes, by getting a version of the data scratched together, we can answer the question at hand, but it’s not ready for production,” he said. It’s often better to build that quick win into the roadmap.
The following points will help make sure the work on these pilots and quick wins add up over time, moving you toward greater organizational maturity and bigger payoffs.
5. …is iterative
If there is no perfect plan, it stands to reason that your plan will require changes and updates. In fact, many times those changes will demonstrate that the overall roadmap is working, and the organization is learning and improving its approach to analytics.
“A roadmap is a living document,” Peters said. He did note that, if the roadmap is done well and the organization does the necessary fundamental steps, then the top layer should only change when something significant changes in the business plan or environment—for example, an economic recession or an acquisition. Frequent revisions to your top goals and priorities would suggest that you didn’t do enough up-front work, perhaps on the listening tour or in defining common terms.
The third layer of the roadmap would logically need more frequent updates, both to incorporate discoveries from completed projects, and to reflect timeline changes, such as a missed deadline affecting dependencies further down the roadmap.
Plan on revisiting and revising the third layer of the roadmap on a quarterly basis, particularly at first.
6. …includes never-ending internal marketing skill development
“Technology, as much as I love it, it’s not the hard part anymore,” Peters said. “The hard parts are the people and processes.”
The “people and process” buckets include things such as building data literacy across the company and increasing the actual use of data in decision-making.
To build an effective insight-driven organization, “you need an executive mandate, yes, but you do need grassroots support also,” Belissent said. Your foundational work, particularly the listening tour, will pay off in this regard by increasing stakeholder engagement, she said, but that’s just the beginning.
“Then you’ll also need ongoing ‘How can we help?’ outreach and internal marketing,” she said.
Marketing of any sort is a skill, or a set of skills, and it behooves an insights team to build these skills continually: data literacy, storytelling, effective presentations, and value measurement and communication expressed in the terms that are most meaningful to each different business unit or audience.
Rather than leaving it to chance or a one-off training course, smart organizations can include this skill-building in the roadmap itself.
Like the other elements preceding it on this list, focused improvement in the ability to market the program will help make your data strategy roadmap and programs more effective as time goes by.