Page 26 - GS220802

Co v er St o r y
CoverStory

Analytics can even be used to Tackling merchant attrition
identify problems before they affect Customer attrition is a challenge for any business, but it
can be a real buzz kill in merchant acquiring. "Churn is
customers, for example, finding and something this industry is always chasing," said Johnny

remediating erroneous payments Stevning, vice president of business intelligence at Fiserv's
CardConnect.
before customers complain.
In fact, it's not uncommon for an ISO to lose 20 percent of
its merchant accounts each year. Goldman Sachs Equity
Research estimated that the merchant acquiring industry
"We've built up our analysis to monetize that data and loses $2 billion a year to merchant attrition and spends an
create value for clients." And that helps clients sell more, additional $1 billion a year acquiring new merchants to
which in turn boosts customer loyalty and revenue poten- replace those that have defected.
tial for ACI and its partners, he explained.
Old tech, new applications "Very few industries would tolerate attrition rates of
20 percent, or more. But many ISOs and agents in our
The technological underpinnings of what companies like industry have come to accept this as a reality that cannot
ACI do are not new. They are similar to the tools card is- be changed," said James Shepherd, president of CCSales
suers use to predict accounts that are likely to close, for ex- Pro. "The truth is that we can change this reality, and
ample. "The technology has been around for a while; we're many ISOs have figured this out and are seeing significant
just finding different use cases for it," said Tim Sloane, vice reductions in attrition."
president of payments innovation at Mercator Advisory
Group. "Even Amazon does it." It's all about being proactive, rather than reactive, Builes
said. Most ISOs, if they address churn at all, do so by
As anyone who has shopped on Amazon knows the com- tracking the general characteristics of merchants who
pany's recommendation engine plays an enormous role in leave, or they may have customer service reps call to
driving sales. By collecting data from individual customer inquire why they left and perhaps offer an enticement to
preferences and purchases, the company can create pro- come back. With predictive analytics, ISOs can identify
files and extrapolate those to find other people with simi- merchants before they leave, understand why they are
lar preferences and make purchase recommendations. dissatisfied and take steps to remedy the problem. "You
get an opportunity to heal the relationship," Builes pointed
The same type of predictive analytics can be applied to out.
merchant acquiring. The key is being able to access as
much information from as many sources as possible. And Arcum uses predictive analytics to help ISOs address
there is no dearth of available information, Sloane noted. churn. But Arcum is not alone. CardConnect has taken a
Merchants have a lot of data about customers, the items similar approach to identifying the problem. "We've been
they purchase and how and when. Acquirers have pay- working on our churn model for two years," Stevning said.
ments data from all the merchants they serve. Networks
and processors have information gleaned from acquirers Using a machine learning algorithm, CardConnect creates
and transactions. And banks have all kinds of information churn scores for each merchant in a portfolio, accessible
on customers. via an ISO management tool it calls copilot. "They use
this [copilot] every day, and can access the churn scores
Predictive analytics can be used to predict just about any- there," said Kyle Aceto, CardConnect director of business
thing: weather patterns, customer shopping habits and development. Stevning added, "It gives our partners the
merchant churn among them. It uses statistical algorithms, opportunity to go in and take action."
combined with internal and external data and paired with
machine learning (a type of artificial intelligence), to glean Each merchant account gets analyzed and assigned a score
insights that can forecast future actions. The more data of between one and 100. "It's not just transaction data that
sources that are available the more accurate predictions we analyze," Stevning said. "We also look at how engaged
can be. This means tapping into outside sources. "If you're merchants are with our products and customer service
only using the data you have access to, you're probably tickets as well. The closer that score gets to 100, the more
missing out on things," Sloane said. likely the account is to churn."

Sebastian Builes, CEO at Arcum, agreed. Arcum created It's akin to a credit score, only with churn scores the lower
a predictive algorithm that can be used to identify mer- the number the better, he added.
chants at risk of leaving their acquirers. Builes said Ar-
cum's algorithm can identify at-risk merchants up to 12 A score of 50 or above suggests the merchant account may
months before they defect with better than 90 percent ac- be at risk, alerting the ISO that they need to do something
curacy. to improve the relationship, Stevning said. Aceto agreed,
26