The intention was great: create an environment where all users are able to design, build and run reports of their own in order to gain insights. Quick turnaround. Greater flexibility. More informed decisions. Being proactive and less reactive. Corporate Utopia. Wonderful.
Unfortunately, self-serve has failed. Or has it? ….
What was the real intention?
For many organisations the action was to provide users with tools and/or data sources that could be used to construct reports in order to generate useful insight. For many organisation this could have meant a reduction or even total elimination of data analysts (usually expensive resources). By providing users with some training they would be producing complex (and accurate) reports in order to gain valuable insights in no time at all. No more waiting for the analyst or the IT department.
In some cases the user would be provided with a few desktop tools and a plethora of data sources and …. off they go! In some cases a beautiful data warehouse would be built with amazingly accurate and trusted data, users were trained and once again expected to produce their own insights.
What went wrong?
Well, we all know you shouldn’t expect your accountant to be a marketing expert, and your marketing folk to be HR experts. But for some strange reason we expect everyone to be expert data analysts.
Things went very wrong when we provided users with desktop tools (such as Tableau) together with a few data sets and asked them to provide insights. This was OK-ish when the user was a data analyst but for the marketing staff, sales org, HR and finance this was generally a disaster for two reasons:
- Lack of data governance
- Forcing people to do a job they’re not good at
To put it simple we set our people up to fail. Although we list a few more reasons for failure below these two are the main killers of self-service reporting and BI.
The following are some of the reasons why BI self-service adoption is very low in many organisations:
- Lack of data governance leading to
- Quality issues with the data leading to
- Lack of trust. Lack of trust is really really big! Nothing kills a BI stack faster than lack of trust in the data and therefore the resultant insights
- The tools provided are not always that easy to use.
- We expect non-BI personnel to be BI experts when in fact we employed them for a completely different skill set.
- Valuable staff resources are expected to perform their day job plus the incredibly demanding data analytics. In many situations one is done at the expense of the other.
- Lack of IT or Data Analyst support.
Where does self-serve work?
Are there situations where self-serve reporting works or is it truly dead? The simple answer is: yes.
Self-serve reporting works when:
- The user (who is NOT a data analyst) does not need to create complex reports (these should be created by an expert)
- The user is able to interact with a report that has already been defined. This can be achieved if your BI tool allows some form of interaction such as drill-down
- Reports are designed with interaction in mind
- Data governance is in place
- Data quality is assured due to effective data governance
- Output (information) is trusted due to effective data quality and governance
and an optional. 7. The BI tool allows some form of collaboration (this is not absolutely essential for self-serve but why not expect this as the new normal?)
Criteria 1-6 above should all be met.
If all that is required is a very simple report of, say, total sales by product then anyone could generate that report with virtually any BI stack. Not terrible useful (unless you go back in time to around 1980).
How to implement Self-serve?
This section assumes that the required reports should be advanced as to provide insights or at the very least useful information. This excludes simple reports such as “sales by country”. Different data sets would most likely be combined. Data should be trusted!
In order to successfully implement self-service reporting with data governance and collaboration one must correctly identify and take care of:
- The technologies to be leveraged (suggest discount desktop solutions and those that don’t facilitate real collaboration and consumer-interaction). Products such as Yellowfin are good options (for longevity suggest choose a product in the Gartner Magic Quadrant). Ensure that the product does actually do everything it says on the tin: in particular look out for misleading features (collaboration is a great one because the definition is so broad).
- The way in which governance will be effectively implemented. Where possible data should be governed centrally in order to ensure quality which in turn ensures trust. Data, if left to itself, tends to a state of chaos. Data should be owned and there should be mechanisms in place to ensure it is fit for purpose. Choose technologies that facilitate governance.
- Data analysts. This skill could be provided in a variety of ways:
- Full time in-house
- Externally sourced
- Contractor – short term
- BI implementation experts – short term or long term strategic relationship
- BI implementation experts (in-house or externally sourced)
So has self-service BI and reporting really failed?
What it really boils down to is the definition of self-service. If the accepted definition is that all users are able to construct their own complex reports and obtain valuable insights from the then the low adoption rates indicate failure. If the definition is that users engage, interact and collaborate using effective tools then success is more likely to be positive.
Yellowfin performed a small survey at one of their events, see The Truth of Self-service BI Adoption (It’s Not Pretty) and Warning: Mind the Data Governance Gap in Self-service BI
Please feel free to contact us if you’d like more information about this paper.