The aberrant stats at the NHS are the result of the code system used for data entry. So if an administrator types ‘460’ instead of ‘560’ on a patient’s record, the figures will show he visited a midwife instead of an eye clinic.
It’s not the only instance of huge data sets being distorted by a slip of the finger. The price of American pecan nuts has plummeted by about 40% since the autumn, after the US government revealed a clerical error that wildly overestimated the volumes it had delivered between August 2010 and September 2011.
And financial markets around the world have seen numerous ‘fat finger’ trades send stocks into a tailspin. Apple lost 10% of its value for five minutes on 23 March, while in May 2010 a mistake in a selling order on a US exchange caused a domino effect of automatic trades that has since become known as the Flash Crash. Procter & Gamble lost a third of its value, while Accenture’s share price dropped from $30 to 1 cent.
At least a fat finger trade can be discovered and corrected quickly. The 90% of companies that lose money because of inaccurate records, according to Experian, generally won’t know where that bad data is. And they won’t find out until they’ve sent out enough unnecessary mailings to irk someone into complaining. By then the money has already been wasted.
So instead of spurning that 15% of your marketing budget, why not invest it in ensuring that data is collected accurately and cleaned regularly?
We often hear how data will be the driving force of modern marketing – the killer app that keeps you more relevant to your customers than your competitors can manage. But acting on incorrect information can be equally damaging. A wrong address, a misspelled name or an assumption that Joe Bloggs wants antenatal care as opposed to an eye test will lead to money being frittered away on sending irrelevant messages.
Worse, it could turn off a prospective customer that otherwise might be interested. And all it takes is a typo.
A chain is only as strong as its weakest link, the cliché goes. Likewise, your database is only as good as its errors.