#DataDump 2: Self-Service
Welcome to #DataDump my weekly thoughts on a subject I’ve been pondering this week.
Today I wanted to follow up on some of the discussion points raised in the #datadump twitter chat last Monday on Self Service. “Self-serve” is something we all talk about with regards BI now but I wonder how many organisations are on the path to true self-service and what it means to them.
The Self-Service Dream
The software companies I work with, Tableau and Alteryx, both talk about helping their customers get away from “Traditional BI”. But what do we mean by traditional BI?
Generally what we mean is the “report factory”; normally this will be an IT-led function that looks to fulfil the organisations reporting needs. The few data-wizards servicing the needs of many across the organisation with reports and dashboards that might take weeks or months to produce. When the reports arrive the data may lead to further questions but, faced with going to the back of the queue, users give up and make do. Reports lie unused and decisions are made without data, all those happy faces soon turn sour and IT gets the blame.
So what happens? Users turn to Excel - they turn the handle on their own data and can get quick answers to the basic questions they have. IT gets huffy that data isn’t secure but the business can move forward. The more technical users develop macros and scripts that produce half-decent reports but then they leave the business. So people are left turning the handle on Excel “databases” that no-one understands but which are critical to running the business. Is it any wonder then that Excel gets a bad rap?
Tableau and Alteryx offer an alternative to this “report factory” / Excel nightmare. Pitching a solution for the many, rather than the few with drag and drop data preparation and reporting, secure data access and reports that take hours and days to produce rather weeks and months. Subject matter experts can work directly with raw data sources in Alteryx, performing quick data preparation and advanced analytics before piping it into a Tableau Data Source for the wider business to consume.
Self-Serve Nirvana
So, in this modern stack we have three “flavours” of data consumers in the business, which actually marry up closely with Tableau’s licencing model and so for the sake of simplicity we’ll use the same names to describe them:
Creators These are people who know data well enough to connect to the sources, prepare it for analysis and then produce data sources and reports to help others get the answers they need.
Explorers Here we have the data curious, those Excel tinkers who need answers but don’t have the time, or inclination, to source data themselves. They rely on trusted published sources to get their data and then craft insights and share dashboards they create with others.
Viewers These are the bread and butter of self-serve, the users who interact with data in dashboards via filters and drill-downs and apps in Alteryx Server to get the answers they need.
A typical organisation will have many more Viewers then Explorers, and more Explorers than Creators, but importantly all can self-serve in their own way to quickly get the answers they need.
The Self-Serve Reality
The reality of this modern world though can be very different. Without careful management one of two situations will develop:
Report Factory Mark II: A handful of creators support the wider business, with many viewers consuming their reports. Reports are quicker to produce and the burden is removed from IT, and interactive reports mean data decisions can be made faster and consumers are happier. A status quo is achieved and the organisation is happy with their investment but the full potential of the tools aren’t achieved. The tools aren’t sticky with the organisation and the toolset could easily be disrupted if something faster and quicker came along.
Excel Hell Mark II: Business users, frustrated with Excel, find Tableau and Alteryx and it’s their dream come true. They are able to quickly build reports and get answers. They develop solutions in Alteryx to crate datasets that drive Tableau Reports. There is a wide base of creator users across all facets of the organisation sharing reports, but no real governance. People leave and reports get unloved, Alteryx workflows grow complex and difficult to manage. Essentially the organisation have implemented another version of Excel, yes it’s better and faster but over time the lack of controls will cause problems.
Understanding why these two situations develop is key to understanding how to prevent them.
Strategy Tableau and Alteryx both use a “land and expand” sales tactic, both organisations know that their tools are going to solve problems in the right hands and so use short trial periods to encourage adoption. Once the tool has “landed” in an organisation it will spread, through word of mouth and will be championed by users keen to see growth and investment across the business.
The downside to this approach is that often organisations will adopt tools through the backdoor, sleepwalking into a large implementation. In the early stages Excel Hell Mark II will look successful and senior leaders will be only too happy to invest in a tool that delivers results quickly. A lack of strategic implementation though means training investment is sporadic, and no vision is applied to how the organisation will approach self-serve.
This lack of strategy means organisations, already used to the report creator -> report user paradigm, fail to capitalise on the explorer role and features such as Tableau’s Web Edit and Ask Data.
Trust Issues
Fundamentally, data owners don’t trust users with their data. When it comes to users making decisions using our data, we’d much rather package up the answers in a gift-wrapped report than trust them to find them for themselves.
This perpetuates a cycle where users aren’t given the freedom to explore data on their own terms, lowering their data literacy which in turn decreases trust. This cycle can be pervasive and pushes organisations back into Report Factory Mark II as they try to avoid Excel Hell.
Pricing
Clearly organisations are price sensitive and so, faced with a higher Explorer price and lower Viewer price the temptation will be to start with the lower-priced product — aiming to “grow” down the pyramid as the organisation matures in its usage.
With features like Tableau’s Ask Data lying in the Explorer licence this can leave that progression difficult to attain — those curious users are unable to do much more than use dashboards and soon are driven to pulling data out into Excel and building their own analysis.
Personally I feel Tableau’s Ask Data would fit better in the viewer licence— allowing users to find data and be curious (within the confines of the Natural Language interface) and perhaps giving them room to grow into the more developed Explorer role over time.
IT
In the modern BI world, as touched on during last weeks #datadump chat on Twitter, IT need to bear some responsibility for the lack of governance that can lead to Excel Hell Mark II.
Dory Owen on Twitter
@ChrisLuv @seamie_b I've been thinking about this lately. After PC boom of 80/90s IT depts had in-house Help Desks, programmers down the hall from business users, Excel & Access super users. Now IT has reverted back to HW, backend, security, & mega-projects behind the curtain.
IT need to provide thought-leadership for the business and actively help users adopt self-serve technologies. All too often IT are seen as blockers, restricting the usage of the software without offering practical and realistic middle-ground to help business users accomplish their goals.
While data was in Excel then IT could bury their head in the sand and ignore issues around data governance, the same can’t be allowed to happen again if we’re to avoid a similar outcome.
The Solution
I’m sad to say there is no golden bullet for implementing a self-serve stack and avoiding issues. However, there a few key areas organisations need to work on in order to get the most from modern BI tools:
Strategy, Culture and Leadership Implementing a self-serve solution is as much as a cultural change for an organisation as it is a software implementation. Senior Leaders need to get involved to drive that change through a strategy designed to make the most of the software utilising people across the business and IT.
Data Literacy Training in tools simply isn’t enough any more, organisations need to push training right down into the business to help people understand data, where it comes from and how to use it. Pushing people from the Viewer to the Explorer role is key if companies are going to use data properly without dependence on just a few individuals.
Acknowledge Mistakes will Happen Finding the right route through a software implementation is hard, balancing governance and flexibility is hard and it’s inevitable that at times mistakes will be made. Avoiding knee-jerk reactions is key — swinging wildly from Excel Hell to a Report Factory is jumping from the fire into the frying pan. Even implementations with a long pedigree can adopt a strategy to change culture and make more and better use of existing software.
In conclusion, there’s no right way to implement self-serve. Many organisations with Tableau and Alteryx will have implemented a faster, quicker report factory and be happy. It’s clear though that modern BI offers an alternative and we shouldn’t be blinded to the possibilities on offer.
In tonight's #datadump discussion (9th December 2019) I’ll be asking Twitter:
are you brave enough to admit you have a Report Factory or Excel Hell Mark II? Have you narrowly avoided creating these?
How do gain senior leadership buy-in in a “land and expand” world?
We talk about cultural change, how does that manifest itself? Can you really change the culture of an organisation when it comes to data?
Have you identified “explorers” in your organisation, how do we nurture them?
How can we implement data literacy awareness and training at a grassroots level?
See you there at 7.30pm GMT.
If you’ve not participated in the discussion before and aren’t sure what to expect then it’s very simple. Simple watch the #datadump hashtag or follow me (@ChrisLuv on Twitter) and I’ll post a series of the above questions. Dive in, answer them and follow the threads of replies to see what others are posting. It can be a bit of a free-for-all and hard to follow but I’ll regularly tweet out interesting responses and conversations on the hashtag. Pro Tip: Tweetdeck can make following the conversations easier.