Why A/B testing with first-party data is key to uncovering customer lifetime value

Philip Lenhoff, Henrik Landquist / July 2020 / Digital Transformation, Sweden, Case Studies

Reaching and acquiring those customers that offer the best, long-term value is key to helping businesses grow. But in reality, sub-optimal ad spend has long been a major challenge in marketing, and not knowing which customers to prioritise is a key part of that.

It was for this reason that Qred, a Swedish fintech company specialising in small-business loans, started investing in customer lifetime value (CLV). And while it initially seemed a daunting prospect, the company was successful by adopting a test and learn approach to their operations. “Treat your business as one big A/B test,“ says Emil Sunvisson, Qred's founder and CEO. “Try things out. If it doesn’t work, fine. If it does, great — start from there and build around it.”

Flexibility is also important, particularly when it comes to handling unprecedented challenges such as those brought by the COVID-19 outbreak. “It definitely had an impact on our lifetime value (LTV) bidding,” admits Sunvisson. “But in times like these it’s key to stabilise, adapt the bidding strategy, and recover to previous levels as soon as possible.”

Qred were able to quickly adjust to the changes brought by the pandemic. But how has centring their advertising strategy on CLV changed how they run their business in the long-term?

Short-term versus long-term bidding strategies

When it came to reaching their customers, Qred used to have a simple strategy. “'Business loans' in Swedish was an extremely high-converting keyword,” says Björn Söderqvist, Qred’s performance marketer. “The cost-per-click (CPC) was around €50, which isn’t cheap, but our strategy was to spend that money to appear at the top. It worked for a while, until brokers drove up competition and made it much harder for us to profit from this search term. They took the first spot no matter the cost — telling people to keep comparing the lending options available to them. It ended up costing €100 for one click, which made us realise our original strategy was no longer viable.”

The high level of ad spend wastage prompted Qred to adapt their approach to bidding. “We needed to find a more cost-effective way of reaching and converting our most valuable customers,” explains Söderqvist. “That’s when we turned to automated bidding.”

Flexibility is also important, particularly when it comes to handling unprecedented challenges such as those brought by the COVID-19 outbreak.

Pairing first-party data with third-party tools

While their bidding strategy needed to change, the finance company was already doing a good job at measuring CLV in their own backend, using it to understand if granting businesses with a loan was worthwhile.

A prerequisite to this, according to Sunvisson, is to gather as much first-party data as possible: “We automatically collect all data we can find and feed it into a machine learning model that calculates not only the likelihood of non-payment, but also the estimated customer lifetime and potential fees, among others. All this results in a precise estimate of the CLV for each individual customer.”

A new discovery, however, was that they could use their existing first-party data to measure CLV directly in Google Ads using a tracking method called offline conversion tracking.

“Once our data scientists have done their calculations, we know the CLV for every paid-out loan on our back-end,” explains Söderqvist. “This knowledge can be used to verify which of these loans was actually from Google Ads.” It’s thanks to this attribution that the algorithm can tell Qred which click led to a conversion and how much it was worth.

Trying new things to stay ahead of the competition

“We’ve applied our new approach to Qred’s largest campaigns, which again handles that keyword 'business loan' amongst others,” says Söderqvist. “Following our old approach, the algorithms didn't care about long-term value. But now that we’re taking CLV into account, the algorithm puts potential customers with a high value in favour of those that, in the long run, will be less profitable to us.”

For Qred, now the fastest growing company in Sweden, being able to feed in-house CLV calculations into Google Ads and adapt their bidding accordingly has proven highly effective, even in a dynamic market. The same strategy has now been applied in their Finland and Denmark branches.

“It’s doing exactly what we want it to do,” explains Söderqvist. “We’ve seen a 25% higher return on ad spend, so we’re spending more effectively, and have more time to focus on valuable customers.”

Whether it’s developing in-house machine learning models or feeding first-party data into Google Ads, what ultimately helped Qred optimise their ad spend is a strong willingness to try new things. And that’s an approach many different businesses could learn from.

Key takeaways

  • Take an A/B test approach to your marketing strategy to discover the changes that lead to positive results.
  • Understand your most valuable customers to optimise ad spend.
  • Use first-party customer data to measure customer lifetime value in Google Ads.
How one retailer unlocked first-party data to better connect with their customers