Why ‘great visualisations’ are not always great help
The importance of providing context with visualisations
Share this blog
Recently several high profile companies have heavily sold the idea that a great looking, fast data visualisation the greatest business intelligence.
It is certainly great to have beautiful visualisations of data, but that is less than half the picture if you need to manage the business better.
Great visualisations don't always give you the whole picture
Consider a great looking visualisation like a chart/graph in a nice dashboard showing sales up 15% year on year in all territories except Alaska where it is up 5%.
Is that good?
15% year on year increase may look great, and sound great, but:
- What if that includes the launch of a new generation of product that was budgeted to recover a recent 50% decline in sales and was expected to increase year on year sales by 30%?
- What if breakeven profitability requires 20% year on year sales increase for the new product?
- What if the sales increase was solely due to a promotional item that is a loss leader, and actually it was targeted to be only 5% of sales, and every % over 5% is costing the business money it can’t afford?
All of these scenarios are real examples.
Managers need context to make an evaluation on the data inside the visualisation, and often then discussion, before being able to make the best decision on what actions to take.
Data must be considered in the context of the business plan. Initially, the goals for that specific metric, to see how it is performing against target.
Comparing against targets is not simply ‘colour coding’ the data against the current target. The target itself may be changing over time, as growth, or decline in a metric is targeted.
What is the target for the coming months, or through to the financial year end, compared to the trajectory of the actual data?
Are we targeting an increasing pace of growth, but the growth in the actual data is already slowing?
What level of variance from target is important?
For instance 5% variance from target may be fine for one measure, but business threatening for others.
A 5% variance in how many days it takes us to pay invoices may be well within our cashflow and supplier terms, whilst a 5% increase in costs on a low margin business could cripple the business irrevocably.
A 5% increase in waiting times for a hip replacement may cause some patients to suffer pain longer than had been hoped, a 5% increase in demand for emergency heart transplants may be beyond our supply capacity and result in a devastatingly high fatality rate.
Speeding up a process step may look good on the scatter chart, and sound sensible. But what is the context of the process. Let’s say the business strategy is to minimise the time in the whole end to end customer process, so speeding up each step may seem a good thing.
But speeding up one particular step, may have negative impacts on the next process step and slow the overall end to end process down.
Consider an application and approval process.
A simple 4 step process of:
- Customer Application,
- Assessment of application, and
- Advise applicant and
A ‘beautiful visualisation’ showing how the customer application processing team have reduced time to gather the customer application data and pass it to the assessment team looks great!
But performance management in context shows that the step ‘2) Assessment of application’ is now slowing down, and the overall process is now taking more, not less time.
Cause and effect links in performance management software show that process step ‘2) Assessment of application’ processing time is dependent on 2 factors:
- a) how long it takes to carry out the assessment function on the data and
- b) completeness and accuracy of the customer data received from the Customer Application step.
Speeding up process step 1 has been to the detriment of data accuracy and completeness, meaning the 5% gain in step 1 is being more than offset by a 15% worsening of duration in step 2.
This is a real world example, where changing the focus from elapsed time in step 1 to accuracy and completeness improved the overall process elapsed time by 20% within 2 months.
What to do about it?
Yes, we want the great visualisations that InPhase like many big-name tools can provide, but we also need the context, of targets, of variance tolerances, of business plans and cause and effect interactions that only a performance management system like that built into InPhase can provide.