How many HelloFresh accounts do you have?
I am a regular, happy, and well-fed HelloFresh customer, but let's just say that I have become a “new” customer more than once in the past. As a result, I probably have two or three different HelloFresh accounts under different email addresses.
This might seem like a non-issue, but for companies like HelloFresh, that are based on consumer subscriptions, these multiple accounts per customer can lead to them wasting millions of dollars in marketing costs.
Voucher Fraud
Firstly, there is the “new customer” voucher abuse. HelloFresh is pretty generous in giving free recipe boxes to new customers if they are introduced by an existing customer. Fast food delivery companies like Wolt or DeliveryHero, similarly will give generous vouchers to new customers to encourage that first purchase.
The downside of this generosity is that if I am an unscrupulous character, I will just keep on deleting their app, reinstalling it and making fresh accounts with new account details so I can use another new customer voucher. If I am determined enough, I might never have to pay for another pizza again.
Customer Suppression Lists
The bigger cost is arguably wasted marketing dollars. If we go back to my HelloFresh account(s) - I am currently a paying customer using my current, regular personal email address. However, my previous account was created using my previous email address, which I only use for social media sign-ups. You might not be surprised to learn that the former is a Gmail address, and the latter a Yahoo! email account.
HelloFresh will send my Gmail address to its advertising networks (e.g. Meta) and tell them not to bother showing me any HelloFresh adverts - the so-called “customer suppression list”. After all they have already won my custom, so why waste advertising dollars.
However, that other Steven with the Yahoo! Email address, who used to be a HelloFresh customer and left us - of course HelloFresh should spend money trying to win me back. So I see HelloFresh adverts on Facebook and Instagram. But of course that is a waste of money - I am already a customer - they just don’t know that the paying Steven Renwick with a Gmail address, is the same as that former customer Steven Renwick with a Yahoo! Email address.
Identity Resolution
This is where identity resolution comes in. Identity resolution is the process of determining that two separate identities in a dataset actually belong together - they are the same person. Using identity resolution technology, like Tilores, HelloFresh would be able to identify that Steven Renwick with a Gmail and Yahoo! Email address are the same person, and thus neither should see any adverts via Meta or Google.
When extrapolated to the millions of customers that a company like HelloFresh has, this can result in preventing millions of dollars being wasted on adverts to existing customers.
Similarly, identity resolution technology can help identify that the new customer that has just signed up and tried to use a new account voucher is actually a previous customer that has previously had an account and already used a voucher.
The Difficult with Identity Resolution
So you might be wondering what is so difficult about identity resolution? The issue is that to compare related, but non-identical data, you need to use fuzzy matching algorithms to allow for the fact that the two profiles might use non-exact data - i.e. I might have used “Steve” vs “Steven” as my first name, or I might have used a variation of the address - “Oxford Street” vs “Oxford St.”. I might be using a combination of fuzzy name matching, digital fingerprinting, and geolocation. For the particular use case, the right balance of precision vs recall must be found.
Then there comes the volume of data that needs to be compared - identity resolution is what’s known as a “quadratic problem” - which means the more data you have, the more comparisons that need to be made every time you ingest more data. The result is the system can get slower and more expensive to process as you grow. Especially for voucher fraud detection, it is essential that the entire identity resolution system can be checked and updated in milliseconds.
That’s why identity resolution systems that are built for scale and real-time data ingestion, such as Tilores, are essential for consumer-facing enterprises to make sense of their large volumes of customer data.
