Faced with growing demand for multi-channel retailing to both trade and consumers, the builders’ merchant and home improvement chain recognised a pressing need to improve its product data. David Reed finds out how better data is now driving better sales.
Travis Perkins is a builders’ merchant and home improvement business operating across four divisions and 21 brands, the best-known of which are Travis Perkins and Wickes. With revenues of £5.5 billion in 2014, 1,900 outlets and 24,000 employees, it has a significant footprint in its categories and also opportunities in the emerging multi-channel world of both consumer and trade retailing.
As David Todd, group data director, Travis Perkins explains: “It has been an acquisitive company over the last 20 years and has doubled in size every five years through acquisition. Not all of those purchases have been integrated, so we have a lot of data silos. When I joined three-and-a-half years ago, we had different databases for products, brands and channels. Products sold in-store were on a different database to products sold online. There were issues with data quailty, product details and hierarchies which were not supporting the web site particularly well.”
Customers were beginning to demand a better experience, especially in the online environment, where they expected improved product search and stock availability information. To revamp the group web site, it was recognised that a data quality and integration programme needed to be undertaken.
“We knew that meant we had to improve our product data, so the decision was made to invest in an ecommerce platform and product information management store (PIMS) which would allow us to build a better web experience. As a starting point, we realised the need for data quality management and data integration tools, which is when we selected Talend,” says Todd.
As a first step, his team undertook data profiling. Travis Perkins has 500,000 products across the group - not all of which are on the website - which range from a few screws right up to bespoke, one-off lintels. Data had never been standardised or validated on the way in, so it was not good quality and often missed key elements, such as dimensions.
Using Talend, the company made a big effort to improve things, for example, by identifying when a product description contained fewer than ten characters - this would not be sufficient to support an informed buying decision. In the first six months following implementation, over 30,000 product updates were made.
Says Todd: “We implemented Talend as a data quality solution first, putting in place rules for our product data to ensure standardisation of what entered the PIMS and establishing those standards within the business. We have built a library of rules to deliver that data governance.”
This meant setting rules for what needed to be present in order for a product record to meet the agreed standard. Variables such as dimensions, image and even colour needed to be in place as this would ensure customers on the web site would be able to make an informed choice. Data is entered either one-by-one or in batch from spreadsheets, with commercial and other internal teams now being enabled to make uploads, subject to these rules.
Ciaran Dynes, VP products at Talend, explains: “Travis Perkins has grown up with a whole suite of in-house software and bought-in solutions which have been operating alongside each other. It needed help to standardise its data, building rules and data quality routines. That is where we have been able to help with data profiling, standardisation, cleansing and integration. It is an evolving project across multiple applications and divisions.”
“We looked at their data and identified where there were gaps and values that were out of line with the norms. In any sequence of numbers you will find some which are not in line with others in the range,” he adds. “One complexity we found with Travis Perkins’ merchant SKUs was differences in price depending on the customer.”
Problems with data were fixed in agreement with the business. That is important because, while IT understands the technology, it is the commercial user who understands the data. For example, in a single transactional row with 200 or 300 fields, only the subject matter experts will understand what a particular value means, especially if it appears twice in different fields.
Buy-in for the programme and investment was achieved through demonstrating to the business what the problems were with the data and showing how, if they were fixed, it would ensure the web site performed better for merchants, consumers, and trade customers. That involved having data stewards and the IT team working together to establish rules and fix problems. Once they had done that, the customer experience was improved and, as a result, conversions from web site visits rose significantly.
A culture change has also been achieved, according to Todd. “Data governance is becoming more important and we now have a data governance team in our data management function looking after data quality. There has been a mind shift from standard data stewardship towards a more technical, rules-based quality approach to ensure we sort out non-standard data. We are also going to improve our metadata and hierarchical taxonomies,” he says.
Part of that process has involved educating stakeholders, showing them the importance of data quality and how it supports data-driven marketing. That has helped to gain buy-in because through demonstrating the uplift achieved as a result of customers being able to find products. The new Travis Perkins web site has already been launched and 2016 will see improvements to the PIMS for the group’s plumbing supplies brand, CPS. “We will be looking at data quality and also product lifecycles to make sure there is nothing in there which fails to meet the standards,” says Todd.
The scale of the task is not to be underestimated. Out of its total product range across the group, Todd and his team have so far been able to improve the quality of 50,000 product records. Resource constraints mean it is unrealistic to attempt to do the same to all of the remaining range.
“The challenge this year is working out what we need to do next - how many of our products do we need to integrate and what quality are they at the moment? We are profiling that data at the moment to get an idea of the size of the job,” says Todd.
Continued growth and the expansion of multi-channel retailing means that scope will only continue to increase. As Dynes says: “Travis Perkins is a real-time business. Data quality and integration is not just for the purposes of those applications, it is for in-the-moment ecommerce, so you need to be able to apply rules in real-time. That is our bread and butter and why all the major retailers globally are using our solutions.”
He adds: “Travis Perkins’ aim is to get to an offering of over a million products, so it has to change the way it does business. That is why it is looking to standardise all of its data and automate data entry”
Thank you for your input
Thank you for your feedback
DataIQ is a trading name of IQ Data Group Limited
10 York Road, London, SE1 7ND
Phone: +44 020 3821 5665
Registered in England: 9900834
Copyright © IQ Data Group Limited 2024