Poor-quality transactional data can also have a substantial impact and ensuring transactional data quality can deliver real value quickly. Make transactional data part of your data quality business case.

Most data quality programs focus on master data – for example, customers, suppliers, and products – as these reusable data sets impact multiple business processes.

Given the value that can be derived from improving master data and the complexities related to addressing master data issues, transactional data may be overlooked.

What is transactional data?

Transactional data is information that is captured from an event – such as an interaction, purchase, or payment. It will typically include the type of transaction, the date and time of a transaction,  various amounts s, some form of unique identifier, and a variety of reference data elements such as the parties and items involved in the event.

Logistical transaction data covers the movements of an item from source to destination.

Financial transactions cover various financial and accounting elements such as invoices, billing, point of sales, quotes, orders, and inventory.

Operational transactions cover human resources aspects such as payroll, time tracking, and leave management.

Digital transactions involve interactions via digital channels – such as clicks, browsing actions, views, and downloads.

Errors in transactions have a direct impact on operations – for example, an incorrectly completed invoice may result in a late payment. Increasingly, transactional data is driving analytics for example to better understand customer behaviour for segmentation and marketing. Again, missing or faulty data at the transactional level can affect the ability to garner critical insights.

Reduce transactional errors to improve operational efficiency

In order to reduce errors and their impact, we need to identify and correct transactional data capture errors early, ideally at the point of capture.

One customer, for example, identified that users were capturing multiple copies of supplier invoices – with small variations in the invoice number such as leading or trailing spaces. In the worst case, payments are being made more than once to the same supplier for the same goods. Mostly, the errors result in costly manual reviews and corrections. In this case, simply standardising how critical information is captured, for example by stripping spaces, delivers quick returns.

Missing information can also cause headaches. For one large HR department, the monthly payroll run is a nightmare. Every month, new employees are rejected by the payroll run due to missing or incorrect data. Similarly, missing information on invoices can result in delayed payments.

Improving the quality of these operational transactions means ensuring that business logic is reflected in the data, by adding or correcting data, or triggering an alert for intervention before the payroll is run, or before the invoice is delivered. Active data governance and data quality can play a substantial role in revenue assurance, as discussed in our free whitepaper, “5 ways quality data increases revenue.“

Accurate transactions drive real-time actions

Increasingly, transactional data is driving activities. Most credit card users will, for example, have received a call or a message flagging a suspicious transaction at some point. To trigger an action, we typically would feed a machine learning algorithm with pertinent information from a transaction – for example, the amount, the location, or even the item being purchased. Similarly, retailers may want to identify a particular customer’s activity, such as a visit to a particular web page, or an in-store visit, to trigger a targeted marketing message.  These events are time-sensitive and need to happen while the consumer is completing the transaction.

While some of this information may be easily obtained, often critical data is missing, incorrect, or inconsistent. Machine learning is heavily dependent on the quality of the data supplied  – missing or inaccurate data means that the opportunity to act may be missed.

Companies wishing to deliver real-time alerts and campaigns need to be able to pull together related data, at scale, deal with data quality problems like incomplete, inaccurate, and inconsistent transactional details, and deliver standardised records to the machine learning products.

Transactional data quality also affects regulatory compliance. For anti-money laundering, for example, financial institutions must pull transactional and master data together, from a variety of sources, without counting on quality data, In fact, criminals may deliberately make efforts to hide their activity by deliberately introducing errors or misinformation into critical data fields. For banks dealing with tens of thousands of potentially criminal transactions daily, the challenge is to reduce the number of false positive alerts fed to investigators, without compromising compliance. Again, this can only be effectively done of transactional data quality is adequate.

High-quality transactions deliver accurate insights

 Last, but certainly not least, poor transactional data quality affects the ability to derive accurate insights from the data.

For example, accountants analyse general ledger information to track spending in different categories – typically by allocating costs to particular general ledger codes that may relate to particular expense categories and/or cost centres.  By analysing the free-format descriptions linked to individual transactions we were able to help one large company to identify incorrectly allocated expenses that were skewing cost analysis. Certain GL accounts were being treated as catch-all buckets e.g. a variety of costs such as electricity or telephone costs allocated against rent, inflating rental costs and reducing the visible cost of electricity or telephone calls. In this case, we had to define data quality rules to identify incorrectly allocated transactions, and even potential opportunities to add new GL categories for future spend.

Missing information in third-party transactional data. 

 Another banking client used a third-party to process certain investments. To correctly allocate costs we needed, again, to process free-format text descriptions in order to identify the transaction category, which was not captured.

These transaction descriptions included a variety of spelling errors, abbreviations, typing errors, and missing or extraneous information that meant we could not search for standard keywords or phrases as these were not consistently captured. We were able to accurately categorize over 95% of these transactions by applying data quality algorithms to the data

_______

Some errors in transactions can be linked back to improving related master data, but there are definite opportunities to be explored at the transactional level alone.

I hope that these examples will trigger some thinking on your side about how and where you can enhance transaction data quality in your environment. Or give us a call and let’s explore this together.

Now let’s look more broadly, what is the true cost of poor quality data? Understanding this unveils the hidden toll it takes on your organization.

In the era of big data, preparation is key: how to prepare for big data. Anticipate the challenges and seize the opportunities it presents.

Photo by energepic.com on Pexels.com