Self-service BI remains a priority for many companies – both to make analytics more accessible and to reduce the burden on IT – BARC Research BI Survey

For IT teams, there can be a mistaken assumption that this is simply a case of giving business users a BI tool like Microsoft PowerBI, Tableau or Qlikview, and then sitting back.

Want to stay ahead of the competition? Our guide on Business Intelligence Tools can help you choose the best tools to achieve your business goals.

Self-service does not generate adoption

BARC research shows that adoption rates for BI tools remain stuck around the 20% rate. Adoption of self-service BI tools remains extremely low. An informal poll by Yellowfin indicates that while around 20% of business users have access to self-service BI tools, only around 20% of this pool actually use them. So, around 5%.

BARC identifies 5 factors that will reliably kill BI/analytics adoption and usage:

1. Lack of access to data. Users who cannot find (or access) the data sets needed for their analytics requirement will stop trying
2. Poor data integrity. If users cannot trust the data then they will stop trying.
3. BI tools are hard to use. There is a reason that Excel remains a preferred analytics tool for many business users. More complex BI tools are not always simple to understand or use, particularly when combined with the lack of access and poor data integrity.
4. Poor performance – Users will not wait for slow-running queries and reports
5. Lack of support – users are more likely to adopt if they have access to coaching/help to solve problems.

The bottom line?

The “build it and they will come” approach that is often applied to self-service BI simply isn’t good enough.

In a post on LinkedIn, Aaron Wilkerson wrote, “We need to admit defeat on self-service. It’s not going to work in the way we originally thought it would. It’s time to abandon it and move on.”

How can we increase the adoption of self-service BI?

Self-service BI means different things to different users.

When it comes to finding the most suitable self-service BI tool for your company, it is crucial that your solution is exactly tailored to the specific skills, interests, and information requirements of your business users.

Wilkerson suggests that self-service requires business users to have skills only a small percentage desire to have. So-called “power users” describe the very small set of users that have both the analytical skills and the tool skills to understand and leverage their own data.

Some users are happy to explore and prepare their own data, particularly as more focus is being given to self-service and low-code data preparation tools. For most, however, data preparation is a bridge too far.

More users may be comfortable with the idea of applying their own filters or aggregations onto existing datasets – in effect creating personal variations of existing reports, or adding new analyses.

Depending on the user type (business analyst, power, or casual user) and their needs, very different BI tools may be preferable.

In some cases, this can be an interactive dashboard or an ad hoc reporting tool. For others, a visual discovery capability, guided advanced analytics, or self-service data preparation capability might be needed. Self-service BI may better suit the data team, accelerating their work, than genuine business users.

Deliver data with integrity

Users that do not trust the underlying data will not adopt BI. For many of us, a lack of trust is linked to poor data quality.

“Garbage in, garbage out“, or GIGO is an expression that dates back to the earliest days of computing. Poor quality data affects every BI developer, but where more technically astute BI developers may be able to work around poor quality data, most self-service business users expect that the data sets provided to them can be trusted.

But good quality data is not enough. A 2021 survey of Chief Data Officers’ perspectives indicated that users also need to understand the business context to assess whether the data they are working with is likely to be useful.

Wilkerson writes, “Most data requests have context so trust in giving data to people is generally low and you end up writing custom queries. How many times have you had someone ask a question and your response is, ‘It depends’, or, ‘It contains this but you’ll have to consider…’

I’ve seen most use cases end up with a data team member having to write a custom query to answer the question that the user has.”

Wilkerson adds that existing data architectures are not suitable for self-service. “For self-service to work, we need a true business/metrics/semantic layer that exposes the business to data that is easy for them to understand and consume.”

For self-service to work, we need a true business/metrics/semantic layer that exposes the business to data that is easy for them to understand and consume.

Aaron Wilkerson

Users are more likely to adopt if they are able to easily answer questions like:

• What data sets exist that contain data that might be relevant to my analysis?
• What is the source of the data set?
• Who else is using this data set, and for what purpose?
• Does a similar report already exist?
• Can I trust the data?
• Who can I talk to about it?
• How do I get access?

Historically, many companies have not prioritised the ability to answer these kinds of questions. Where this kind of context is available it is often difficult to find, in many cases sitting in a subject matter expert’s head.

A data catalogue, like Data360, makes it easy to get answers to these kinds of questions – by combining metadata, business context, data integrity scores, lineage, and responsibilities in a single web-based dashboard. Where gaps are identified, the platform’s workflows can be used to coordinate between business stakeholders and data engineers to ensure that required data sets are prioritised and added to reporting data stores. Automated lineage tools, like MANTA, help users and data engineers to understand how data flows between systems, and how it has been manipulated.

A cultural shift must take place to increase adoption.

Wilkerson suggests that self-service approaches can create a disconnect between data and business goals and objectives, and, almost as a throw-away, “it cuts off the data team’s ability to be at the table as it relates to data value creation.”

Successful self-service initiatives prioritise data governance – involving the right people in decision-making and ensuring that the business context is documented and accessible. The right people should include stakeholders from business, IT and the data team, to make decisions that advance data in line with the enterprise data and business strategy.

The research also shows the value of investing in data quality. When users know that they can trust the data, they are more likely to spend time understanding the insights that can be generated.

More organisations are investing in basic data literacy programs. Knowledge workers need to understand how to interpret and, more and more, manipulate data to get answers to the questions they need.

From self-service to DataOps

While self-service BI can be a valuable tool, it should not be the only source of insights, and in either case, success remains highly dependent on IT support.

Wilkerson points out that most organisations “don’t have the org structure to support self-service. Data teams are usually built to create and enhance data pipelines. We don’t have teams dedicated to supporting the education of users and supporting a self-service model. You’re usually pulling team members away from other projects or tasks to field business questions.”

The DataOps approach to analytics is one approach to bridge the traditional gap between business and IT stakeholders – by improving collaboration and driving a more flexible and agile approach to delivering useful and trusted data sets, reliable data pipelined, and active, meaningful metadata.

DataOps brings together diverse stakeholders – from business, the data and analytics team, and data and cloud engineers who may be sitting in IT – to ensure that all of the complexities of delivering trusted data for each use case are considered and planned for. This reduces frustration on the part of both business and IT and makes business users less dependent on IT for answers.

The data and analytics team should budget a percentage of their time to provide coaching and support to self-service users. As these users’ skills and confidence grow, more time can be focused on advanced analytics and data science. Even these advanced users will be more productive when enabled with a foundation of trusted, high-integrity data as a starting point.

Read more about how self-service BI can accelerate analytics.

For another perspective on modern BI, read how Artificial Intelligence and BI are combining to accelerate insights.