Self-service capabilities crucial to analytics success

Self-service analytics spurs growth.

That was a key finding of a 2020 report from the Harvard Business Review, sponsored by analytics vendor ThoughtSpot, in which 72% of respondents from organizations that gave employees self-service analytics tools and entrusted them to make data-driven decisions said they saw an increase in productivity.

On Jan. 12, HBR and ThoughtSpot released a follow-up to their previous study on self-service analytics.

Due to the COVID-19 pandemic, data-driven decision-making has perhaps never been more important. The ability to act and react quickly amid fast-changing conditions has been critical to the healthcare organizations battling the virus, governments attempting to keep their citizens safe and enterprises trying to survive and thrive amid unprecedented economic fluctuations.

Modern self-service analytics capabilities, meanwhile, equip frontline employees to make informed decisions based on trusted data and insights that give healthcare organizations, government agencies and enterprises an edge over those relying on instinct and slow-moving centralized analytics programs.

In their follow-up report, titled "Empowering the New Decision Makers to Act with Modern Self-Service Analytics," HBR and ThoughtSpot delved into why a modern self-service analytics platform is needed to enable data-driven decision-making in the environment created by the pandemic, and why the combination of certain capabilities with the right organizational culture is a critical element of successful self-service analytics.

According to the report, providing more users with modern self-service capabilities ranks second only to data quality in determining the success of an analytics program. Meanwhile, lack of effective people change management -- not the technology itself -- ranks as the biggest hindrance.

Recently, Cindi Howson, chief data strategy officer at ThoughtSpot and host of The Data Chief Podcast, discussed some of the findings of the report.

In addition, she spoke about the evolution of self-service analytics, what successful self-service analytics can net organizations, why data literacy is key and how to teach it, and what characteristics are common to organizations that understand the importance of data-driven decision-making.

Take me back five to 10 years. What was the target audience for self-service analytics capabilities?

Cindi Howson

Cindi Howson: That's a big range. Ten years ago, self-service was really self-service for the IT professional, whereas five to seven years ago it was self-service for the data analyst who may have been an IT professional or who may have been part of the business. It was still very much for the data professional.

Fast-forward to January 2022. What is the target audience for self-service analytics capabilities? Howson: When I talk to our customers, they'll use the term 'true self-service.' I think the industry has been using the term self-service for a while, but true self-service is enabling the non-data professionals and the non-analysts -- the true business users -- to ask their own questions.

What are the essential capabilities a self-service analytics platform has to have to enable those true business users to query and analyze data without involving data professionals?

Howson: Key ingredients are search, national language processing (NLP) and AI-driven insights. The other big trend is the rise of cloud data platforms where the data is live and can be accessed without moving the data. Key things have changed [over the last few years] about where the data is coming from and how fresh it is, and [query and analysis] are powered by NLP.

Assuming an end user is properly trained, what can a modern self-service analytics platform enable end users to do that they previously couldn't?

A self-service platform lets users ask a new question or find existing content. And if you wonder why this is so important, it's because the pace of everything changed.

Howson: It lets them ask their own questions in their language. You referenced training, and I want to be careful. The industry has over-indexed on training on the technology and not enough on the language of the business -- this is what we call data fluency and what some will call data literacy. A self-service platform lets users ask a new question or find existing content. And if you wonder why this is so important, it's because the pace of everything changed.

As painful as the COVID-19 pandemic has been, it's forced people to modernize their data platforms and make data-driven decisions. Intuition-based supply chains are tight. If one supplier is blocked, or one container has been sitting off the coast, a company needs to be able to source from a different supplier. And one of the trends I've written about is the rise of people analytics. We never had to previously know which employees are vaccinated, or who has been issued healthcare equipment. These are questions that weren't being asked two years ago, let alone five to 10 years ago.

What's the result when an organization enables end users with self-service analytics tools and enables them to make data-driven decisions?

Howson: It's higher revenues, higher revenue growth and higher profitability. And then you can go across every function and see increased operating efficiencies and increased employee engagement and joy, which is super important across the Great Resignation. If you are a procurement person or a call-center operator and you're getting shouted at [by a customer] and able to say, 'Don't worry, I can get you that product faster from a different provider,' or, 'We have a stock out on Item A, but is Item B a reasonable substitute?' -- if that front-line decision maker or call-center operator can answer like that -- they're more likely to stay. So it's higher employee engagement is as well.

Are organizations doing enough to enable employees to make data-driven decisions?

Howson: No. The report highlights that for technology, data and analytics professionals, it's so much easier to change the technology and it's harder to change the culture and the people, but it shows how critical it is to change the culture and the people. According to the report, 44% said lack of attention to people change management has been the top barrier to self-service analytics. This tells you that you can't just throw a shiny new technology out there and assume it's going to work. You have to pay attention to the people change management and the culture.

Beyond people change management, what else stands in the way of organizations adopting self-service analytics?

Howson: Historically, they might have been hard to adopt in an on-premises world. There should be no hesitation in a cloud world. They're easier to deploy now.

But lack of awareness and confusion in the marketplace [is a barrier]. I think what's happened is some vendors say they can do NLP, but there's a difference between search and NLP, and there's a difference between doing search and NLP on a live cloud data warehouse like Snowflake or Google BigQuery. If you're only doing search on a small data set, that's not helpful, and that's what [some] vendors do, so I think there's confusion there. This is the difference. With search, [organizations] are really getting down to SKU-level analysis, and that's where the money is.

Earlier you referenced data fluency/data literacy -- is lack of data fluency a barrier to self-service analytics?

Howson: It absolutely is. But I don't want to say it's a barrier. It's a chicken-and-egg scenario. It's like learning how to read. To teach someone to read, you have to give them books to start with, but don't give them Homer's Iliad or Odyssey. Give them something easier to start with. This is where self-service analytics with guardrails and governance is key. You want to be teaching them the language of the business and not hard-to-use tools -- that's a profound difference. The industry has over-indexed on teaching hard-to-use tools and not enough on what a term means in the context of the business. Some may say they don't want to enable self-service analytics because their people aren't data fluent, but if they're not given a way to read, practice and interrogate the data, they're never going to grow that muscle.

A way to start, a best practice to start, is to give them live analytics on a liveboard -- not a static dashboard. Maybe they're looking at sales this month versus last month by a particular brand. Give them the key metrics, and then let them drill down and explore. Give them something familiar but then let them drill, filter, sort and then ask the next question.

When business users are given self-service analytics tools to make data-driven decisions, do organizations still need data professionals?

Howson: A federated organizational model is the best of both worlds. You centralize for economies of scale, and you decentralize for domain expertise. You have the domain people leverage the common data and focus on the why. If you think about the tiers of prescriptive analytics -- descriptive, diagnostic, predictive -- this lets the domain people focus on the why, on the diagnostic and predictive. What you want the centralized team doing, which actually gives them the joy of their job back because they're not working on the [repetitive tasks], is provision data, enable new data sources, focus on the hard problem of data quality and work on higher-level analytics.

Some of our clients, when they measure the tickets coming in to the centralized team, see that the routine questions have gone down while the higher-level questions have gone up.

Can you give an example of an enterprise that benefited from adopting a self-service analytics strategy?

Howson: Medtronic is an interesting organization. One of the impetuses for them is they saw how much they were spending on IT rather than on actually delivering medical devices. Self-service analytics allows them to do more with less, in terms of technology. I think that's a core part of their technology. Medtronic typifies what you want in a culture in that they're innovative, that there's trust, and that it's really about empowering all decision-makers, but especially frontline decision-makers. The central team does not have us-versus-them mindset that IT is in an ivory tower.

You said Medtronic typifies what you want in a culture -- do organizations that enable self-service analytics have a typical profile that differs from those that don't?

Howson: I often say that technology and culture are two sides of the same coin. Show me a legacy technology portfolio and I'll show you a culture that's afraid of failure, settling for complacency and risk averse. If you look at someone who has a modern technology stack, then they also probably have a culture of innovation, pursue excellence, and they have a highly competitive business environment that can be a forcing function.

It's really modernizing for the cloud, and the integration across the ecosystem is now more possible than ever before, [but] the gap between the leaders and the laggards has widened. You can't keep settling for good enough, otherwise you're going to end up being legacy. If you go back 10 years, data was optional, but in the digital world, data is the only thing that gives you the visibility into what's going on. In a way, it's survival of the smartest, which is the analytics leaders. Everyone has gotten the message, but [the laggards] have an inability to execute.

Editor's note: This Q&A has been edited for clarity and conciseness.