Why insight is essential and technology can’t provide it

[reprinted from the Institute of Business Consulting's Body of Knowledge]

Essential: vitally important, fundamental

Insight: penetrating and often sudden understanding, as of a complex situation or problem

The recession and its causes have reminded us to take nothing at face value - when things seem too good to be true, they usually are. The credit crunch is a salutary reminder that, in good times, as well as in bad, we need to know the real health of our businesses: where we are, where we’re going, and the risks we’re taking.

If, armed with such insight, managers can make informed and balanced decisions about cost-cutting, risk reduction and investment, businesses will emerge, not only unscathed, but stronger than ever.

So how can we best inform these tough decisions?

What’s the challenge?

I once challenged the inventor of a sophisticated search engine to build a business analysis engine - a software machine into which gigabytes of the latest business data could be poured. I wanted him to come up with something that, in response to a series of broad business questions, would produce valuable insights to inform executive decision-making.

He declined on the basis that artificial intelligence wasn’t yet up to it, unconvinced it could shed much light on an organism as complex as a business. But what about so-called Business Intelligence software? Couldn’t we just enhance that? Make it a bit more intelligent?

On reflection, he was right to say no. There are two main reasons why computer-based business ‘reporting solutions’ fall short.

Firstly, they rely on models that simulate the way a business reacts to inputs - with built-in priorities, structures and behaviours. Such models are rigid and crude. As the business and its environment change, rapidly flexing and evolving, the model can’t keep up.

So, what happens? Well, either there is no model - diverse measurements are reported and little attempt is made to connect them - or the model flags a Red when it finds two Ambers, and so on, until it becomes obvious that the model is wrong.

Ultimately this is a Catch-22 problem: to gain insight from the system we must first programme insight into it.

Secondly, even if we do make the most of these imperfect models, we’re often hit by poor implementation. Typically, making data and software the starting point rather than users and their needs produces a result that is out of line with the priorities of the organisation. As for the vital ‘user interface’, it more often displays a fondness for gimmicks rather than the sound principles of graphical communication.

And when the system is used? Just don’t ignore the Business Intelligence small print when it says, “this software won’t totally remove the need for people to think.” The business case may promise a “headcount reduction”, but what happens when the few people who remain are too busy to look at - let alone think about - the output?

The typical outcome is expensive failure: long implementation time frames and high expenditure, approved by senior executives who rarely see the benefits.

So, should we dismiss technology altogether? Of course not. It can play a highly valuable role - in enabling managers to discover what is going on.

What we do need to do is distinguish between management information - accurate, pertinent business data - and a report, which should add insight by analysing this management information and presenting conclusions.

So, the ideal ‘reporting solution’ is a balance of two things: the technology, to provide consistent, efficient access to information, and the people who have the knowledge to interpret this information.

How do we put a reporting solution in place? What do we need to know?

The right approach forgets about technology and first thinks hard about what the organisation wants to achieve, how it does this and how it knows it is succeeding.

Again, events have reminded us that apparently sensible performance measures can lead to unintended (and unwelcome) behaviours. “Maximise shareholder value and share the benefits” can be interpreted as “maximise immediate return and get rich”. And whilst investors are warned they can lose as well as gain, remuneration schemes have been less symmetrical.

To incentivise the right behaviours, we need to focus on the right measures. If growth is the key to success, then both business development and growth indicators will be important, i.e. contract wins or revenue growth, not revenue itself. If the customer experience is key, looking at the efficiency of internal departments misses the point. But a single view of performance from the customer’s perspective will hit the spot.

Next, how is the organisation structured to achieve these objectives? This is the organisation as a single body, not independent organs. For example, customer feedback is connected to research, the resulting new products require production and sale, after which support will lead to further customer feedback.

Managers already receive objectives specific to their functions so the task here is to define measures that encourage the connections between functions to work well. Do we have a match between production resources and what the sales team has sold? This can often expose weak connections - where things are thrown over the fence from one function to the next and it’s the final, customer-facing department that takes the flak.

If the relationship between functions is mature, success will relate to efficiency. If it is brand new, then to effectiveness - simply making things happen.

All organisations are subject to external influences, so reports must also include relevant information on suppliers, customers, regulation, the economy and so on.

And finally a completeness check: do we have a balanced suite of measures that will encourage the organisation to achieve? A mix of current and forecast performance? Of financial, operational and project measures? Across products, people and geographies? Internal and external? Highlighting opportunities as well as issues and risks?

If so, we can progress to the next step. No, not technology yet.

How should this content be presented?

Here we need clear, concise reports. Information that jumps out. Reports that communicate problems so clearly that decisions become obvious.

Unfortunately, this is not most people’s strength. The most common presentation medium is out-of-the-box Excel, which leads to randomly-selected 3-D graphs, garish colours and dark grid lines.

The standard financial reports are tables of figures - variance to budget, forecast, last period and last year, for one figure after another, on and on.

Then there are vague project essays, listing what has been done but - crucially - not what hasn’t, the implications and what is needed to bring things back on track.

And finally, there is the technical approach. Business Intelligence dashboard gimmickry - simulated fuel gauges where the least prominent feature is the ‘needle’ showing performance. As Stephen Few¹ suggests, “insisting on sexy displays similar to those found in a car when other mechanisms would work better is counterproductive”.

Like the weather forecast, we lose interest in current reports long before any relevant fact emerges - sometimes the writer’s intended outcome.

So how do we find precisely the right form of presentation?

In most cases, the financial report is not for Finance, nor is the project report for the project team. Their purpose instead is to enlighten others. So it is important to employ the most suitable form of presentation to communicate each type of information. Too much has been written on this subject (and generally ignored) to do it justice here, but a few words might be useful.

When we just want this and last month’s numbers, a table of figures is efficient (if not very enlightening); if it must be a table, we keep borders pale so the key facts stand out.

But if we want to spot trends or present forecasts, a graph will reveal what even the most skilled financier can’t find in a table. Stephen Few again, “graphs ... give shape to numbers and ... bring to light patterns that would otherwise remain undetected.”

In many situations it is of course these patterns that are of interest - are sales or costs rising or falling? Programme delays random or systematic? Staff or customers staying longer or deserting sooner?

We have already mentioned the (mis)use of garish colours. Particular care must be taken to use colour appropriately in graphs. As Tufte² says in the context of map design, “aggressive colours ... render the map incoherent”, whereas the standard “grey lines are a miracle of information design”.

With the traditional red, amber, green traffic light system for performance measures, green (“OK”) is often superfluous and distracting. Instead the use of just two colours, say red and pink, is very effective, with red strictly limited to where there is a problem.

The discipline of project management is well established, but project progress reporting is a typically neglected element – hence the essays mentioned above, or the endless and impenetrable gantt charts. Effective one-page alternatives include commented milestone charts and strategy charts (the latter hinted at in the picture below).

So, it is possible to communicate complex messages effectively. With subtle emphasis - bold text or some colour highlighting areas for attention - the right mix of text and graphics will communicate performance and trends at a glance.

Here it is also worth noting that much of what is monitored should not be reported - otherwise, key messages can too easily be drowned out. This requires a systematic approach, not simple exception reporting. First, an analyse-prioritise-distil sequence and second, careful attention to context - presenting related information together, so connections are clear and key points are more easily absorbed.

Finally, and most importantly, what brings life to a report is commentary - something software can’t yet do - so we leave space for that.

The outcome is a report structure: concise headlines of performance and key messages, clearly presented supporting information and analysis, and a final pointer for readers to sources of detail.

How do we produce it? And make use of it?

Now we know what we want, how do we put in place a process and ensure it continues to provide regular value for decision-makers?

This is where we take advantage of technology - its ability to collate data from diverse sources, perform rapid calculations, and present the results in a consistent format. To tell us how things are.

So we automate, lightly.

We need a reporting tool that will sit on top of powerful systems and large data sources. But we don’t need the tool itself to be powerful – just flexible.

As we said earlier, priorities are unpredictable - we don’t know in advance where issues will emerge - and the business will in any case evolve. This must be reflected in the report, so its producers need a tool that enables them to flex content at speed and control how it is presented.

The degree of presentation flexibility this requires is absent from much of today’s off-the-shelf business software. But it is absolutely feasible to build a robust, user-friendly tool - and a small ‘library’ of reusable graphical elements - from familiar desktop software.

Most managers are familiar with the MicroSoft PowerPoint report format, so an MS Access-based tool works well. Data from disparate sources across the organisation is collated each month and quickly imported into an Access database; the tool then populates template pages with the latest data and exports these pages to a PowerPoint outline with gaps for commentary.

With careful design, this tool can be very user-friendly - used by non-technical people not only to produce reports but also to analyse the data (often data that is held in a single location for the first time) and to make rapid changes to report content and layout.

This approach has proven suitable for FTSE100 businesses - don’t let the IS team tell you otherwise!

The speed brought by automation frees time for the report producer to interpret the information, identify emerging issues, drill into them and form conclusions. And finally to prioritise and present this insight for decision-makers in a clearly-accessible format.

Leaving no excuses for poor decisions.

So, my feelings about technology? Think first about what you need, not how it will be produced, and you won’t be constrained by technology. Human/machine teamwork can bring real business benefits, as long as we don’t expect too much of the machine.

Until, that is, we really do have artificial intelligence.

¹Few, Stephen, 2006. Information Dashboard Design. Sebastopol (CA): O’Reilly Media Inc.

²Tufte, Edward R., 1997. Visual Explanations. Cheshire, Connecticut: Graphics Press