Around 6% of annual revenue is being lost through poor quality data, according to new research by Royal Mail.
Given the positive impact well-maintained contact data can have on campaign performance, it’s perhaps surprising that around 70% of the 300 companies surveyed admit to having incomplete or out-of-date customer data. And the problem is getting worse given this figure is up 12% compared to 2014.
However, 34% of marketers fail to fully understand the financial impact of poor quality data, according to the study.
Jim Conning, managing director of Royal Mail Data Services: “CMOs and marketing directors all understand the importance of accurate customer data but I’m not sure more inexperienced members of the team understand the increased ROI of more accurate data.”
He reckons retailers alone could save £500,000 a year through data cleansing and argues that marketers need to address the “dysfunctional” silos within their businesses to capture data effectively.
Multiple departments have responsibility for data collection, regulation and strategy with varying levels of involvement. The marketing function tends to take responsibility most often for setting the data strategy (54%), data collection (49%) and regulation (42%). When it comes to strategy, the board (36%), data management teams (27%) and strategy (26%) also have a level of responsibility, while for data collection, behind marketing, it is the CRM department (40%), customer service (34%) and sales (33%) so these teams must work together more cohesively.
In a way life event data is the most important data of all, because there’s the potential to project future needs.
Gaelle Comte, Unbiased
Websites are the primary source of customer data collection (87%), according to the study, while less than half of data is collected via the mobile web (44%). However, only 44% of companies automatically validate data at the point of entry online and 19% have no validation at all.
Financial advice platform Unbiased captures the majority of its data via a form customers are required to fill in with their financial details and advice needs. The team also collects data from the website’s ‘find an adviser’ search tool and newsletter sign-up form.
“The consumer data we handle tends to be sensitive – not just their contact details but also the value of their assets, which is necessary in order to match them with the best available adviser,” explains head of marketing Gaelle Comte.
“The data captured is a mix of defined answers with regards to products held or wanted and free text, which provides the context of the enquiry and gives real insight into how an adviser may be best able to help.”
Social media is better for driving traffic rather than capturing data, says Comte, whose team developed its own data capture technology so it is in full control of the insight and how it can inform new products and services.
Ecommerce platform Thread is hoping to turn personal styling on its head by using its own algorithms to offer an ‘ultra-personalised’ experience. Led by head of applied research Ed Snelson, who joined Thread from Microsoft’s AI labs, the team has developed powerful machine-learning mechanisms that enable the website to scale the styling experience.
The system uses complex ‘decision trees’ to group users into millions of ‘micro-buckets’ across more than 100 criteria including age, style, brand, fit, colour and fabric. This means if one customer loves Ralph Lauren and Tommy Hilfiger, he will have a completely different experience to another who only loves Ralph Lauren, explains co-founder Kieran O’Neill.
“Signing into Thread is effectively like walking into a store where everything is in your size, in your budget, by your favourite brands and approved by a stylist. This is structurally impossible to do offline, as we have more than 3.7 billion outfit combinations to choose from,” O’Neill explains.
“We’ve developed our own proprietary, internal score for measuring the personalisation and quality of outfit recommendations. When we first began, it was frankly pretty poor. We tried pretty standard algorithm approaches such as collaborative filtering, and they yielded quite uninspiring results, like suggesting a grey T-shirt to a guy who likes black T-shirts.”
Thread therefore evolved its data strategy, drawing on a number of metrics including users’ ‘likes’ and ‘dislikes’ and ordering propensity. The algorithm is now 20 times more powerful and has helped push sales tenfold, increasing the propensity of first-time users to buy something and growing site visits.
Tracking life stages
Whether a customer is having a baby, moving or getting married, 61% of marketers consider life event data useful as it offers reasons to engage (61%) and creates sales opportunities (59%), according to the research.
As financial planning and legal advice is driven by life stage, Unbiased is keen to capture life event data to help deliver relevant content. The company recently mapped and profiled its existing data to Experian customer segments, which is tied to life stages.
Accurate life state data is also helping Unbiased tailor its email communications to specific segments. So customers who have just started a family will be sent one kind of message, whereas someone about to retire will receive a different tailored message.
“In a way life event data is the most important data of all, because there’s the potential to project future needs,” says Comte.
“For instance, someone aged 50 will probably need pension advice within five years, while someone starting a career may soon need mortgage advice, so you can see potential new clients coming on–stream even if they don’t revisit the site in the interim.”
Permission and the value exchange
The number of marketers using third party data sources is on the decline, according to the research. Only 47% of marketers use third party data, compared to 56% in 2014, which Conning attributes to a loss in confidence that third parties are compliant with GDPR regulations. Some 43% of marketers say they are worried about third party compliance, with 58% concerned about their own data compliance.
READ MORE: A third of marketers feel unprepared for new data laws
As it is difficult to get permission, social media is largely used as an advertisingchannel rather than a targeted ad channel, says Conning. “Engagement with customers on social media adds colour and updates on what people may be doing rather than definitive data.
“It’s important to build properly permissioned, insightful data along with social insight. For example, some 10% of people move once a year, which is a huge proportion of any database. Therefore social can add colour to the redirections data we collect,” he adds.
Achieving customer intimacy is crucial to reach true personalisation, as is appreciating the value exchange, says Conning, who recognises that marketers need to work harder to build value in the customer’s mind and stop them moving on to a different brand or provider.
We tried standard algorithm approaches such as collaborative filtering, and they yielded quite uninspiring results.
Kieran O’Neill, Thread
To achieve better data clarity brands should set up standard permissions to collect data and move away from siloed structures to achieve a single customer view, advises Conning, which will be key to generate customer loyalty. He also recommends integrating insight from social media and properly permissioned third party sources.
Comte agrees that data has to be earned and consumers are happy to share details about their lives as long as they feel they are getting a valuable service in return. The challenge for Unbiased is therefore to visualise the data and extract its full potential.
Forging an ongoing relationship with customers is the best way to earn trust, meaning they are happier to share data, adds Comte, who is working on using this insight to surface relevant data, showing past searches and predicting what help customers will need based on their previous behaviour.
Having a human element
For O’Neill and the team at Thread, while data is crucial to the business model, the human element of the brand is just as vital. “Our stylists also work closely with our data science team to build and improve our algorithm, making tweaks like adding new parameters to better capture an element of personal style, or changing the weightings of different elements of the algorithm to improve each user’s recommendations,” he adds.
The website teams its ten human stylists with the in-house proprietary algorithms, which assist the stylists in making recommendations. Consumers can also message their stylists directly for advice.
Brands that are committed to maintaining accurate data are set to reap the financial rewards, as long as they do not lose sight of the need to build customer intimacy and trust.