When discussing data quality, marketers have an interesting point of view. They don’t view information in terms of mass trends, even though that’s essentially what a market is.
Every marketer has five key concerns:
- How will I retain customers?
- How will I build trust with clients?
- How will I meet targets?
- How will I meet deadlines?
- How will I ensure profitable outcomes for the people that pay my salary?
The marketer has a single, personalised view of prospects and wants to deliver the right message at exactly the right time. In order to close the loop, the marketer needs to ensure they get meaningful data as a foundation of a marketing campaign.
They are utterly reliant on data quality to produce accurate reports and improve customer satisfaction.
Planning, Not Fixing
At DQ Global®, we often hear data being blamed for poor marketing campaigns once they’ve failed. Despite this, businesses frequently fail to address data quality challenges before the campaign begins.
For marketers, quality data is important. They need information that is:
- Fit for purpose
- Relevant
- Trusted
Absolute Accuracy in Data
Marketers need to know that data is reliable before the campaign begins. To achieve this, the business must implement an 8-step proactive approach:
- Identify the data quality problems
- Collect information about the problems
- Determine the cause
- Identify possible solutions
- Select the best solution
- Plan an efficient implementation
- Implement and test the solution
- Review and debrief
But a data quality project isn’t just about matching strings of letters and numbers. The context of the data helps us to understand what it means: it is only then that it becomes information, and that’s where the real challenge lies for the business.
Consider an acronym like NASA. If NASA is one of our clients, we need to have a convention for storing the first piece of data: the organisation name. Without this, we may have a number of versions: NASA, N.A.S.A., the National Aeronatics and Space Administration and… well, Nasa, to name but four.
This seems like a trivial problem, but it can introduce data quality challenges down the line. Most systems would interpret at least three of the four as different entities.
What about the date NASA became operational?
- 1 October ‘58
- 1st October, 1958
- 58/10/1
- 1/10/58
- 10/1/58
Look closely at the last two dates. These could be interpreted as 1st October or 10th January, depending on the date system we are using, and this is why data types are so important when tackling data quality for marketing purposes.
Address Verification at Source
Bad data enters a database in a number of ways. One of the most common is the point of entry. For marketers, this means names and addresses.
Consider a system where a new customer has to be added to the database. A customer service operative takes the call and asks for their surname: Smith. This is a surname that can be spelled multiple ways, so we have an immediate vulnerability.
The next customer is female, and their surname is Smith-Jones. Are they:
- Ms Smith Jones?
- Mrs Smith-Jones?
- Prof A Smyth-Jones?
- Dr A Smythe Jones?
House names can be used instead of numbers, or perhaps there is a flat name alongside the number? Road names can be misspelled or abbreviated, and many customers don’t know their postcodes. If they do, they Royal Mail may have changed the postcode when a new estate was built next door.
Also, the customer service representative can simply mistype a word – and that’s something that is difficult to catch in real time.
Minimising Errors
There are countless opportunities for error at the point of entry, and it’s easy to see how duplicates appear. Data quality software has been engineered to tackle this problem. It can:
- Check existing records in the database for a potential match, so our marketers always get a good set or records
- Check third party records, such as the Royal Mail postcode database, to establish a standard version of an address, so our marketers know the records have been verified
According to the Quality Assurance Institute, errors detected in testing can cost 100 times more to fix than errors detected during design. In a typical example, we found 4.1 per cent duplication in a B2C database containing 18 million records. The cost incurred was a staggering £3.96 billion – or £4.11 per erroneous communication.
In our experience, the amount of waste in a B2B database is even higher – between 10 and 30 per cent in a typical set of 50,000 records. A simple mailout to this database would result in £54,000 in wasted printing and postage, and wasted effort.
Conclusions
If data quality is poor, there are three key consequences for the business:
- Money is squandered
- Resources are wasted
- Conclusions are not reliable
For a data quality project to be successful, the entire business needs to have bought in to the reasons for doing it. Every department should have a vested interest.
Once data in the database is healthy, there are a number of benefits for the business.
- It gains a competitive advantage, since it can be more agile and able to adapt to change
- Decision making is more effective, since information is accurate and can be trusted as a basis for judgment
- The business increases its profit by bringing in more revenue and wasting less resource
Once you have a business incentive – preferably a financial one – it is much easier to proactively improve data quality across the business. As marketing becomes more profitable, the whole organisation has the opportunity to thrive.
