Good Data Quality Is The Basis For Good Analytics
Your marketing data should be the jewel in the crown of your marketing assets: if complete it will allow you to understand your customers, communicate with them in personally relevant ways and measure the impact of your efforts.
But that can be quite a big if.
Your data decays from the moment you collect it.
Maintaining high quality data should be a critical enabler of your marketing objectives and the foundation of your marketing data strategy.
Knowing if you have a problem and therefore what to do about it can be difficult. Follow these 7 tips to work out if you have a data problem and how you should fix it before you consider any data analysis.
Your customer data decays the moment that you collect it.
People move house, change job role and get new mobile numbers on a continuous basis. This is called data decay.
And in addition, there’s always a likelihood that customers have given inaccurate information on your data capture forms - they might have quickly selected an option at the top of a list or deliberately inflated their job role.
So how can you tell if you have a problem? And what should you do if you discover problems? Here are our 7 tips.
1. Quality check
On a periodic basis you should screen your marketing data for:
- Duplicates - is the customer on your database multiple times with slightly different details?
- Field completion - what proportion of the fields that you collect are completed?
- Validity - is the data that you capture valid? For example, do all of the email address you have collected have an '@' symbol?
2. Use your campaign responses
Assuming that you are actively communicating with your customers then your marketing campaign results will be a great indicator of whether your marketing data is fit for purpose.
- Hard Bounces - these are where, most likely, the email address that you are using doesn’t exist
- Goneaways - when direct mail or operational communications aren’t delivered
- Unsubscribes - your unsubscription rate could relate to your proposition and offer but it could also be due to the fact that you aren’t using the right customer details
You should look at the absolute rates of these metrics but as importantly look for changes in the rates.
And if you do notice changes, analyse where these changes are occurring - for example, is it for customers who have come via a call centre or via your eCommerce platform?
3. Don’t Rely On Claimed Data
This really means the powerful combination of observed (claimed) and inferred data.
Cross-reference your customer data with your transactional data and create derived fields.
A customer may tell you that they are a high frequency user of your business but your transaction data may tell you otherwise. At a more extreme level, a grocery retailer may believe that a customer doesn’t have children but notices that they are in fact purchasing child products.
Also create derived fields for things like age.
4. Compare Cohorts
You may have different asset classes of customer, for example based on when you acquired them.
Compare, say, the profile, data quality and campaign responses of customers acquired in 2014 vs those acquired in 2013 and 2012.
You would expect some profiling variables to be consistent and if they're not then you know that you have a data issue.
5. Validate Against Third Party Data
There are a wealth of third party data providers who will screen your data (often for free or at low cost) and who can then enhance your data (for a fee).
Using the screening option is a good starting point. Have a look at Experian Intact for one such solution that offers a free data audit service.
Ask your customers It sounds obvious but there's no reason why you shouldn't check the data test you hold with your customers direct and incentivise them to keep it up to date.
Your value proposition is key.
6. Broken Loops
Your aim should be to close the loop on your marketing performance - having an holistic view of your customers across all touchpoints and understanding the value that each marketing activity drives.
Setting up closed-loop analytics is not easy, and there are plenty of areas where the loop can break. A broken loop can directly hurt your business results.
For more details on the specific loops that could be broken and the actions that you can take, download our free eBook.
What to do if you have got an issue
Here are 3 tips:
- Focus on the critical information. What’s the data that will enable you to market to your customers and improve their experience of your brand? This will be very different to the data that you collect to give colour to the critical information that you are collecting. Prioritise your response accordingly.
- Build a plan. You need a concerted effort to tackle the issues rather than a scattergun approach.
- Fix at source. If you have a problem, first fix the route cause so that you aren’t increasing the scale of the problem by continuing to capture poor quality data. Then consider how you fix the data that you have already collected.