When you start out learning about data quality management, you invariably have to get your head around the cost impact of bad data.
One of the most common scenarios is the mail order catalogue business case. If you have a 5% conversion rate on your catalogue orders and the average order price is £20 - and if you have 100,000 customer contacts - then you know that with perfect-quality data you should be netting about £100,000 per mail campaign.
However, we all know that data is never perfect. So if 20% of your data is inaccurate or incomplete and the catalogue cannot be delivered, then you’ll only make £80,000.
I always see the mail order scenario as the entry-level data quality business case as it’s common throughout textbooks, but there is another case I prefer: that of customer churn, which I think is even more compelling.
In Information Quality Applied by Larry English, there was a reported incident of one European bank analysing their customer complaints and discovering that 100% of records related to data quality issues of one form or another.
I can certainly believe that figure. When I’ve investigated complaints within utilities and telecom organisations, I’ve found incredibly high volumes of complaints linked to bad data.
Customer complaints are a great place to start with data quality improvements because there is a direct link to customer pain and grievance. By applying some “pareto profiling” of the most common causes, you can create significant benefits in a very short timeframe. Quite often the issues are related to billing anomalies that require minor fixes or procedural changes to get right.
But how can customer complaints be transformed into a business case?
It’s actually quite simple. First, you need to work out the lifetime value of a typical customer. Let’s say that on average, a typical customer nets you £10,000 over the lifetime of their engagement with your brand.
What you will find is that customers who register complaints will typically churn at a higher rate than those who don’t enter the complaint process. Of course there are many customers who churn naturally, perhaps because of competitive influence, change in financial situation or because they experienced bad service but couldn’t be bothered to complain. However, these customers form your standard customer group and will churn at an average rate like everyone else.
Interestingly, studies show that the longer it takes to resolve an issue the more likely customers are to churn. This gives further demand for much more effective data quality management during the complaints procedure. Root causes need to be resolved so that additional complaints are not created and the customer support teams don’t become even more over-stretched.
So let’s imagine that a typical customer who doesn’t complain at any point in his lifecycle with your business spends on average £5,000.
Customers who experience complaints churn more readily and therefore have a reduced lifecycle with your company that equates to a 25% reduction in average sales leading to an average sale of £3,750.
If you average two complaints a day, your total loss through complaint-related churn is £912,500 per annum.
By assessing your complaints data historically, you can ascertain what percentage of customer complaints were caused by poor data quality management. For argument's sake, let’s say that in the past five years, bad data has resulted in an average 50% of customer complaints.
So you can quickly see a business case forming here - but more important, you can track the trends. Is customer complaint volume increasing? Is the data quality defects per 100 customer complaints ratio increasing? Is the time-to-resolution metric increasing (more customers churn the longer the resolution time)?
The key is to demonstrate not just the financial impact but the trend. If senior management wants to grow their customer base (who doesn’t?), then how will that impact the customer support costs and performance metrics?
At some point for management it becomes a no-brainer. Without better quality data your performance metrics are only going to go down, your support costs are going to increase and your churn rates will follow suit.
Have you calculated the data quality impact from customer churn? Does this sound like a plausible approach for your organisation? Welcome your views on how you’ve tackled the issue of data quality value.
7 Comments
Dylan: You raise some great points in your article. Identifying the ROI for data quality is a key component to ensuring that your program is supported (and funded) by senior management.
While it's sometimes difficult to calculate the cost of poor quality data - in some scenarios these costs are called "soft costs" - it is definitely worth it to do so.
Thanks for the encouraging comment Karen, appreciated.
You raise an important point that you should never stop monetizing your DQ efforts.
I recently met with a utilities DQ team who had their funding pulled and they lamented the fact that over the years they had stopped promoting their "big wins" with DQ.
It also helps to provide focus on where you should be investing your efforts for the maximum returns.
Thanks again, always appreciate great comments like this, Dylan
Dylan, great points. It is very important to put a price to poor quality data.
The challenge is to find the elements that allow for a cost estimation that does not involve too many "assumptions". In the example you used, in many companies, the estimation of some of the variables may involve a lot of financial "creativity" (like the lifetime value of a customer) or most of the time unavailable (churn rate difference between customer that complaint and those that do not and the impact on sales). We can always say that "studies show" that the percentages or amounts are such or such, but management tends to be a bit skeptical of figures that are not coming from their businesses.
Another element we can add to the cost of customer complaints is the time staff is spending handling complaints linked to bad data. Most customer service departments keep records of staff time spent handling customer complaints. If a customer service team is spending an average of 20h/day and average salary is $30/hour, annually the company is spending around 216K /year on complains related to bad data.
Hi Dylan
A great approach. DQ projects (and they should always be projects) should always start with the questions, "Where is bad DQ costing us money? How can we fix it? How much would it cost? Is it worth it?"
Customer complaints and churn is a key area to address.
I would also add the cost of acquiring a new customer to the bad DQ costs here.
Great post.
Thanks
John
Thanks John, great tip.
You're absolutely right, it costs far more to attain a customer than retain an existing one, sometimes the payback is months or even years into the future, all the more reason to ensure they don't churn unnecessarily.
Dylan - don't suppose there are any real cases about how much incremental revenue a data clean-up has contributed to a company? Dare I say, any data on the "price of poor data", not just hypothetical numbers?
Worth a thought
Cheers
Brad M
Hi Brad
It really varies from sector to sector and for individual problems.
To give you one example though, I once visited a utilities company for a few days and within 3 hours found a simple billing error that had been costing them £60,000-£70,000 a month.
I've been on utilities premises and found up to 40% of equipment and connectivity data being wrongly recorded. The labour and service impact costs for something like that nationally would probably run into the tens of millions.
The SAS guys might have some more concrete examples from their improvement projects, I'll reach out and ask them.