What does self-service mean in BI?

What immediately comes to mind when you think of “self-service”? My mind immediately takes me to these two scenarios:

• Self-service check-out at the grocery store. You’ll consider using these express machines to ring up your own items if you’re just buying a few things — if you have a credit card handy, a tub of yogurt and a box of cereal doesn’t justify waiting in line for the cashier.
• Self-service gas station. Different to the grocery store example, this isn’t a “time-saver” option. If you had the option of someone pumping gas for you, you’d probably prefer it since you could stay in your car, likely saving some time, and you could drive away without oily hands. In this gas station scenario, it’s self-service because there’s simply no one there to do the job for you. The station itself is fully self-service.

There are many other scenarios that imply that the customer can take care of the end-to-end request on their own, without requiring human assistance. Across all examples, there are varying degrees of support that’s available if you need it.

Let’s now transfer “self-service” to the data world, where it’s most commonly referred to in association with business intelligence. In self-service business intelligence at an organization, the “customers” are usually the business users, who consume data in order to derive insights for decision-making. These business users — think Sales, Marketing, Operations, Finance teams — tend to not be technical in their roles, and their responsibilities don’t require them to know SQL, python, or any kind of software engineering.

The “human assistance” tends to be the technically-skilled data teams, who are generally responsible for owning the company’s data, centralizing data in the warehouse, ensuring high governance and cleanliness, and transforming the data for analytics. The end goal is to serve up the analytics for the “customers” that need it.

While the data teams are generally deemed responsible for “all things data,” the disconnect is that they’re not the end users who need to consume the data for mission-critical decision-making at a company level.

The profiles between the data owners (data team) and the data consumers (business team) are highly different, with the major barrier being the SQL skillset. While there are data-savvy business users out there, there’s only so far they can go with data modeling and transforming without learning SQL or code.

So, to close the gap and encourage efficient collaboration between these two teams, for the purpose of quickly going from raw data to insights, what can be done?

From an organizational standpoint, the difficulty is that there’s not really a “middle ground” on the spectrum of who puts in more effort to meet the other.

• We could ask of more effort from the business users, and require that they all learn SQL. That way, they’ll be as equipped as data teams to access and analyze data on their own.
  - But is that a good use of time and resources? If it’s not an essential part of their core job, that sounds like a steep learning curve that won’t be worth the returns. As a result, this is rarely the option companies go for.
• We could ask of more effort from the data teams. Their job is “all things data” anyway.
  - They can be the ones responsible for delivering data in all kinds of formats (charts, dashboards). Business teams can submit all data requests to data teams, who will act and respond accordingly.

The second scenario is far more common, for obvious reasons, and is what happens in most companies. As a result, data teams are painfully swamped. They’re expected to act like a support team (quite literally, the “human assistance”), fielding ticketed requests all day from business teams. Oftentimes, they’re expected to drop everything they’re doing to work on last-minute business requests, which prevents them from progressing on the more strategic or long-term work they have on their plate, to improve data quality, or to drive data process efficiencies. What’s worse, these requests pile up, resulting in a bottleneck and a longer wait for business team members. The consequence: this impacts the speed at which they can leverage the right data for their decisions.

Since this disconnect is so tough from an organizational perspective, we have higher hopes for technology to step in and enable a world where business users can self-serve themselves analytics, whenever and however they need it.

Usher in the era of the self-service business intelligence platforms. Metabase, Domo, Looker, Whaly, Tableau, Qlik, and many more — these platforms are a godsend in finally freeing up some of the data teams’ time and bridging the gap.

Before blindly implementing a self-service BI platform, however, it’s important to define what “self-service” means and what it should look like, taking certain questions, cautions, and nuances into consideration. For example:

• What’s the right balance of “self-service”? Should it be more like the grocery store example, or the gas station example in terms of available assistance? Should business users be left on their own entirely to fend for themselves? If they don’t know how to use the gas pump, it could get pretty inefficient for them to figure out how to use it in time for that important business decision that needs to be made in 15 minutes.
• From a technology standpoint, what does collaboration look like? What if the business teams, who don’t know anything about data or the tool, go in and break the data? What if it leaves a messy pile for the data teams to clean up? Data teams would likely rather keep fielding the requests from business teams than have to clean up their mess.

Without further ado, here’s my take on self-service in BI.

What self-service should achieve

• Non-technical business users should be empowered to autonomously answer their own questions about their company’s data, without knowing any SQL.
• They should be able to go a step further and drill down deeper into the insights on their own, without having to request this information from the data team.
• They should be able to do this easily in minutes or even seconds, to then quickly come up with insights that will inform their actions to excel in their day-to-day business tasks.
• Ultimately, a high-performing self-service model should lead to data adoption across the company, and trust in data at a company level. The more the business teams feel comfortable in their ability to autonomously work with data, the more they’ll trust it and lean on it.

How self-service should be achieved

Organizational

Self-service should be about collaboration and open communication between both teams, like in a team sport. In a team sport, there are specific positions and objectives that different players are responsible for. Each player has their own space to do what they do best, and shine in their own right. With this in mind, don’t expect business teams to “become analysts,” but expect them to become experts at fetching and interpreting actionable insights from data.

• Data teams still undeniably have a big role to play. The idea of self-service is that they shouldn’t be a bottleneck for getting analytics out to inform business insights. Data teams should still be in charge of owning the data and data experiences — building out a single source of truth with security and governance that scales. They are the ones who will model the data and assign the metrics and dimensions. Since data teams tend to be the ones working most closely with the BI platform and all the tools in the data stack, they should be responsible for training the business teams and providing support where necessary.
• Business teams are responsible for exploring data and running queries to answer their own questions — serving themselves what they need — and drilling deeper when necessary. They shouldn’t be left in the dark, or feel abandoned to fend for themselves. Self-service is not meant to add extra work on business teams. As such, they shouldn’t be spending hours manually exploring data on their own in tools that they don’t fully understand how to use. This defeats the purpose of quick insights for key decisions. They should be going to the data teams with any questions around the platform and how to run queries, but not so much for specific data requests.

For this to work, the self-service component of the BI platform must be easy enough for business users to use. Which takes us to the next section.

Technological

We’ve gone through how roles should be divided, like players in a team sport, and boundaries should be clear. This applies to the work environments within the BI platform as well. There should be clear lines around where the data team works, and where the business teams work, within the same platform. Of course, these two areas (or tools) must be highly effective at communicating with each other, leading back to the collaboration point. Collaboration is of high importance in both the organization aspect and the technology aspect.

Not only does the separation of having two tools/interfaces enable each team to do their jobs better, it also ensures high governance so that data isn’t at risk of breaking or becoming unreliable.

• Data teams should have their own interface or tool within the platform that’s bespoke and tailored specifically to them. This tool should equip them with everything they need to excel in their day-to-day tasks — data modeling and defining metrics and dimensions — and make them do their jobs better.
• Similarly, business teams should also have their own area or tool that empowers them to most efficiently get the analytics they need, when they need it. Their “workspace” should cater specifically to them as well. As non-technical users, this means a familiar and intuitive visual interface with drag & drop features. This is part of what empowers business teams to build confidence around data quickly, and to get insights right away. Once business users are familiar and comfortable with the space that they operate in, and what they get out of it, that’s when trust is fostered, and when adoption goes up.

What self-service shouldn’t be

• The gas station example, where business teams are left alone to fend for themselves, spending hours trying to figure out what to do with the gas pump (BI platform), or how to get what they need. Needless to say, they would give up out of frustration, and this would lead to an adoption issue.
• A playground where data teams and business teams play freely in the same area, which makes for a governance headache.

All in all, if we had to tie it back to one of those real-life, every day self-service examples, it would be a little closer to the grocery store example — but not because you’re “just buying a few things.” Regardless of the complexity and depth of the questions they have, business teams should always be able to leverage “self-checkout” (self-service) to save time, be autonomous, and avoid the bottleneck that “waiting in line” for the data teams could have. When it comes to making business decisions, there’s no time to spare.

This is what makes self-service BI so important, and something every company should invest into. You’ll be glad you did, because it comes with some pretty great benefits like boosting data adoption and data familiarity beyond the data teams, setting the foundation for implementing a data-driven culture at your company. Plus, you’ll be empowering data teams to do what they do best, as opposed to treating them like a support team whose main job is to react to business requests 24/7.

To learn more about Whaly's self-service BI platform and how it can help you, get in touch!