What Does Self-Service Analytics Actually Look Like?
You’ve undoubtedly heard the term “Self-Service Analytics” thrown around, but what does self-service analytics actually look like in practice? What does a self-service user look like? And what prep work is needed to enable these people to serve themselves?
I spoke with Crystal Meyers, our resident Tableau guru and self-service analytics advocate to learn more. The following is a conversation with Crystal, where she explained some of the nuances of self-service analytics.
The Challenge
Joe: Hi Crystal. So, first of all, why would people do the analytics themselves, and what challenges do you see them run into when they attempt this?
Crystal: One of the most important factors in successful analytics is rapid time to insight. With today’s intuitive, user-friendly tools for both analytics and data preparation, it has become common for the people closest to the business problems to leverage emerging technologies to try to answer their business questions more quickly and easily than working through a centralized team.
However, because the variety and volume of data sources found within most organizations now exceeds the capacity of even the most mature data teams to maintain in a fully modeled Data Warehouse, we are seeing many of the challenges which existed prior to the Data Warehousing approach to analytics again today.
Challenges like awareness, access, quality, and integration of various data sources result in those using end-user focused technologies, like PowerBI, Tableau, and Alteryx, spending a large percentage of their time understanding, cleaning, and preparing new data sources before they can do any valuable analysis.
In addition to that, once insight is found in user-created analytics and there is a desire to scale that solution to a larger audience, people often have difficulties automating, distributing, and gaining adoption on those solutions without a central team intervening to do extensive re-work to ensure that the content is accurate, performant, and aligns to any organizational standards.
Implementing a successful self-service analytics strategy, which minimizes both initial time to insight and the effort to scale content of value, starts with understanding who may want to use data for decision-making (we call these personas), how they would use it (what capabilities do they need), and why (what is the potential business value). This initial understanding is followed by enabling them to do so by defining a set of processes for identifying and managing new and existing sources of data based on value.
Joe: This sounds a lot like the Experience Design approach to research and requirements gathering. Would you say that a team starting up a Self-Service process needs someone with this skill set on the team?
Crystal: Definitely, and I think we’ve seen evidence of that with the enablement of that skill set within analytics teams at many of our clients. Ultimately, the first step to success is understanding who is doing what and why.
Joe: What do we do next, once we understand the business context in depth? Does this help us narrow down the sources we’ll have to work with?
Crystal: I wouldn’t say it helps us narrow them down, as we don’t necessarily want to limit the availability of data. It helps focus the effort devoted to integrating, automating, and modeling each source appropriately.
The Different Types of Self-Service Users
Joe: Tell me about the type of people who are doing Self-Service Analytics. Do they tend to have similar roles and skill sets, or does it vary?
Crystal: When it comes to self-service, not everybody wants to exert the same level of effort to get their questions answered. For example, you might have executives that prefer content delivered to them already summarized, consolidated, and annotated because they don’t have time to interact with it, while others would like to be able to change parameters on pre-built dashboards. Then there are users who want to create their own queries and content to answer questions that arise during meetings or conversations, which requires a more robust set of capabilities.
Joe: So it sounds like there are two camps of users here: those who will be getting their data through dashboards, and those who like to roll up their sleeves and interact with the data directly.
Crystal: Absolutely. At the highest level, you can classify users into two broad categories – Consumers and Analysts.
Analysts are creating content, either to answer questions for themselves or to share with others. This category includes everybody from Business Analysts to Data Scientists. The key to success for these users is facilitating their insight and reducing the time it takes them to achieve it.
Consumers will be using content created by others rather than creating their own, and again there is a spectrum of capabilities that they will need. Their success in using analytics will depend on the ability of those creating the content to design solutions that will be useful and usable.
A successful design requires a high level of understanding of the consumers’ business needs and how they will use the solution, the insight it offers, and what actions they may take based on what they learn.
Joe: So, if we’re talking about someone in a pure consumer role, can we really consider what they’re doing as Self-Service Analytics? If the content is being designed for them, isn’t their experience still the same as it would have been if a central IT team was designing it?
Crystal: Yes, you are absolutely correct, for a pure consumer, their experience will be the same. The benefit of employing a decentralized, self-service approach, for them, is that they have more people enabled to create that content for them, and ideally should be able to obtain it more quickly (reducing time to insight).
There is a risk in this more distributed analytic approach, however. With more people creating content – people with different backgrounds and skill levels designing the experience – the potential for inconsistency and inaccuracy increases. It’s critically important to ensure that the right standards and skills are propagated throughout the team.
Joe: Tell me about the Analysts. How do they engage in Self Service Analytics, and what resources do they need to do it?
Crystal: Those who are doing more interactive analysis may be using various kinds of data. To simplify, there will be data that is refined and has demonstrated value over time, which has likely been cleansed, modeled, and stored in a central data repository of some kind, using an automated process. This data is relatively easy to access and use. However they are also likely to have a need to pull in other, newer sources of data that have not gone through those cleansing and integration processes, and they will often need to combine several sources of data together.
Managing data is the key for this type of persona. People need to be able to quickly find out what data is available to them, how to access new data sources, any nuances or definitions relevant to their use case, and finally, what standards they should be following when creating initial content or prototypes to avoid extensive re-work should they find the value they are looking to prove.
Getting Up and Running
Joe: What are the biggest pitfalls people run into when trying to spin up Self-Service Analytics in their organization?
Crystal: It’s tempting for those who are trying to enable a self-service strategy to think the right tool will be the best answer. But in reality, different groups or users will have varying needs, and a single tool or component of a suite may not be suited to meet all of those needs.
A key part is having the processes and standards in place to be able to take analytic content that demonstrates value, regardless of who creates it or what tool they use, and be able to operationalize and distribute it more widely as it becomes relevant and useful to new teams or groups within the organization.
Joe: So it’s more so the process and mindset of team, than tools?
Crystal: Absolutely. I believe that defining a process for providing users with the right capabilities and managing data is more important than the selection of a tool itself.
Joe: If you had to boil it down, what would you want people to remember when they are trying to roll out Self-Service Analytics?
Crystal: I think that we can summarize a successful self-service analytics strategy into 3 main categories:
- Identifying the analytics personas within your organization and the capabilities they will need
- Effectively gathering requirements with a focus on the user experience to allow those creating content to understand and design what consumers truly need
- Implementing a flexible yet governed data architecture and management process, including standards around content development
Joe: Thanks for explaining this Crystal!
Crystal: Thanks for listening Joe, I really enjoy talking about this topic and feel that organizations of any size can optimize their business insights with the right self-service approach.
So, there you have it. If you’re thinking about enabling Self-Service Analytics in your organization, give Ironside a call. Our BI team will walk you through the process to get you up and running in no time.
This article was co-authored by Crystal Meyers, Engagement Manager at Ironside.
About Ironside
Ironside was founded in 1999 as an enterprise data and analytics solution provider and system integrator. Our clients hire us to acquire, enrich and measure their data so they can make smarter, better decisions about their business. No matter your industry or specific business challenges, Ironside has the experience, perspective and agility to help transform your analytic environment.
