Accurate Vs Useful – is more accurate data a good thing?

If you’re an accountant (like I am) then you live for accurate data. We love it.

We put it into our spreadsheets, model it, graph it, present it and it makes us feel warm inside.

But sometimes we go a bit over the top.

Accuracy is really important. But it’s not that important.

You see there’s a trade off between accuracy and speed of delivery. Very often it takes forever to get really accurate data and in the meantime the world has moved on.

The opportunity has passed us by

So let’s look at a graph.

The problem with accuracy

There’s an ‘S curve’ of accuracy. The more effort we put in then the more accurate our data becomes.

At the start, if we just get a data set and use it without checking or cleansing then it is potentially suspect. (we could always substitute ‘reliability’ for accuracy here).

The problem with accuracy is that it takes time. If we want to spend forever getting accurate data then we often have to spend time making sure it’s right especially if our systems and processes aren’t great.

The longer we spend cleansing and checking our data then the more accurate it gets, although I’d argue that very few data sets are 100% accurate.

We can also suffer from decision paralysis through data collection.

Drawing conclusions based on crappy data is a bad thing to do.

So we spend time making sure it is correct. We cleanse it, we check it by eye, we cross match elements against other sources, we put it through statistical analysis and we sacrifice a chicken and read its entrail to make sure the gods approve ( we don’t really do the last one, much).

But this all takes time. And truthfully by the time we’re choosing Henrietta from the run the amount of accuracy we’re adding is starting to reduce rapidly, hence the S curve.

Meanwhile the business moves on apace.

While we’ve been doing all this our competitors have launched a new product, we’ve employed a load more people, the Bank of England have put up interest rates and Leicester City have won the Premiership title (they really have).

And suddenly our data set has stopped being as useful.

What we need here is another graph

Data can be useless

The longer we take over getting our data accurate then the less useful it can become.

And when we are presenting our data then we have to add in all sorts of caveats about data age and our credibility starts to wane. Oh dear.

Let’s put this into a business setting.

Imagine you are head of an accounts department and you normally close the month on the 24th.

You’re really proud of your work because you’ve taken almost a month to make sure that your books are as right as they possibly can be. You’re king (or queen) of the world!

The problem is that it takes a day or so to get the management pack out and then it’s a weekend and then people are doing stuff and before you know it your managers and directors are making decisions on information that is a month old!

It’s lost a lot of its usefulness and instead of being a hyper useful decision making document it has become merely an historical record.

It’s interesting but still…

So what to do?

We need to assess individually the requirement for speed and the requirement for accuracy.

So do we need something ‘quick and dirty’ to make a relatively unimportant decision on photocopiers?

Or are we making a (business) life or death decision on the pricing mix?

In other words are we launching The Space Shuttle or putting fuel in the car?

In the first case making a good decision based on bad data could be catastrophic, in the second it could mean we have to fill up again before the end of our journey.

We also need to add in a bit of top quality management here.

We’re being paid to choose the point at which data collection stops and the information we have is reliable enough to make our decisions.

It’s not an exact science it’s more an art, but don’t tell people that.

The best tip I can give is that once you have assessed your speed and accuracy quotients that you make a regular review of whether putting more effort into the data collection process is actually worthwhile.

Sooner or later you’ll get to the point where you say ‘no’ and that’s the point you write up your powerpoint.

Leave a Reply