How to be smarter with customer data audits
Each year, the data that marketing teams hold on their customers will degrade by around 10 – 20%. This is simply because approximately 1% of the population will die, 10% will move house, and email addresses and phone numbers will inevitably change.
Not all data will take the same amount of time to degrade. For example, details on customer segments, products or customer type will all deteriorate at different rates. The age of the data plays a part too. If it is over 3 years old then 30% of customers will have moved, so the entire database essentially changes address every 10 years.
It’s also not entirely equal. Data provided by existing customers may be more valuable to a business than that held against prospective customers, and this may vary between marketing campaigns.
With the General Data Protection Regulation (GDPR) coming into effect this May, it has never been more critical to keep data clean. Article 5 of the regulation states “every reasonable step…” must be taken to fix inaccurate data, but how can you know what to fix if you haven’t properly audited?
Here is how you can be smarter with customer data audits:
Narrowing a data audit across either different products or customer types (such as active, lapsed, prospects etc.) gives a basis for comparing results. This means you also have grounds to justify investment from appropriate teams if needed.
Imagine letting a product manager know the address quality of the total customer base is 95%. They would probably consider that a good score, so no action needed. However, if you can use tangible proof to show that the specific product’s customer address score is just 80%, despite the 95% overall quality, you are in a stronger position to get the support and resources needed to look into why the product’s score is lower.
Focusing the audits enables marketing teams to benchmark internally across products and business areas, as well as spotting problems. If one set of individuals has purchased two different products, but there is a difference in the data quality, you are well placed to dig deeper into the issue.
Analysing other variables, such as contact data validation, opt-ins, field populations and distributions will provide a full perspective of the data. If the product manager sees there are email addresses against 50% of customers, for example, they may not realise if, say, only 60% are verified. In this instance, only 30 in every 100 customers might be valid, not 50 in 100 as originally assumed.
When comparing datasets of products, if some have more opt-in rates and a higher percentage of valid email addresses, it could be that something is wrong. There could be issues with the system that passes the data across, or perhaps the on-boarding or data capture processes aren’t working properly.
2.) Regularity is key
As a rule, the longer the time left between data cleanses the higher the cost will be, as more data will have degraded and will need to be fixed. It’s much better to be proactive than reactive, to avoid paying out higher costs.
You can use regular audits in the same way you would an exception report. If they are set up to automatically run they can fix and improve data quality when it drops below a certain level. For example, if the amount of invalid email addresses gets to 10% or telephone numbers drop below a 90% match.
This approach means the investment is frequent and regulated, so that a high level of data quality is self-maintained. When data quality is taken care of, it allows teams to focus on pending marketing campaigns, or elsewhere in the business that needs attention.
Frequent data audits are a vital way to monitor and repair any inaccurate data, and from May you could receive a fine under the GDPR of €20m or 4% of a business’s global revenue, whichever is greater. Regular data audits mean you are in control of your data.
3.) Maximise the value of the data
Auditing your data for its quality is a vital step, but it is just the start of understanding it fully. Other work needs to be done to get a holistic view, and the audit lays the groundwork for this.
A data quality audit will tell you that, say, 10% of telephone numbers are invalid, but a next step would be to conduct further investigation into this variable, to understand the distribution of values, and the formats used (e.g. +44, (01423), 1423 etc).
It may be that the field is inconsistent in how the telephone number is captured and stored and this can impact how much of the data can be actioned. Correcting the data at source to prevent more dirty data getting into your systems is extremely valuable. Equally, correcting data so it can be efficiently transferred across systems reduces the manipulation time analysts spend on correcting data.
Ultimately, a data audit is a great place to start. Coupled with a more comprehensive view of data variables, you can move towards a position of maximising the usage of your data, and maximising the returns on it, as well as ensuring you meet best practice and comply with GDPR.
- » Netflix and machine learning is one thing - but for the rest of TV, AI is just hype
- » A guide to campaign tracking for simple and stress-free analytics
- » Grad Conn, CXMO, Sprinklr: On changing CMO strategy and the Copernicus effect for customer experience
- » Walk before you can run: Start small and smart to get data-driven marketing right
- » How privacy has become the new competitive advantage in a world of personalisation