Blog post

Meet OfferFit — dump A/B tests, use AI to personalize promotions!

Meet OfferFit — dump A/B tests, use AI to personalize promotions!
Written byGeorge Khachatryan
Published7 Oct 2020

Drum roll, please… We are very excited to announce the launch of OfferFit!

Our mission: Make it easy for marketers to personalize promotional offers

You’re a B2C marketer. You want to grow your company’s revenue, so you send promotional offers to customers. Maybe you run special sales, or give gift cards in exchange for referrals, or offer special discounts to win back customers who canceled. Your challenge: How can you find the “sweet spot” of the discount levels, so for each customer, you’re not over-discounting or under-discounting?

Your solution is to personalize offers: send the profit-maximizing offer to each customer. That way you’re offering “just enough” to get the purchase.

Today, to do that you’d need to divide all your customers into segments, and then run A/B tests to find the best offer for each one. That sucks. Here’s why:

  1. It’s a TON of work. And if you want to add a new offer later, you’d better be ready to REDO all of your A/B tests again. If market conditions change (like, I don’t know… a global pandemic?)… yeah, you guessed it — redo again.

  2. It’s not really personalized. Sure, it’s “personalized” to segments, but segments aren’t people. You’re throwing away almost everything you know about each customer. Real personalization should be “1:1” — i.e., should really look at all the relevant things we know about each person.

All you want is an easy way to do 1:1 personalization. Is that really so much to ask??

OfferFit’s software uses self-learning AI — that’s the secret sauce

OfferFit’s AI sends offers to customers, then learns from their responses:

meet2 1

That makes it SO MUCH easier for marketers:

  • No segments!

  • No A/B tests!

  • If you add a new offer, no further action needed — OfferFit just runs with it.

  • If market conditions change, OfferFit adapts automatically.

It also personalizes better — a true 1:1 fit.

And it’s all wrapped in a nice interface that’s fun to use and gives you helpful insights:

Cropped-tablet 1

Who are we, and how did we get ourselves into this??

This is me (George, on the right) and Victor (on the left) at OfferFit’s global headquarters — a.k.a., my kitchen:

meet1 1

Victor and I met in 2008 while studying for our PhDs in math at Cornell. After that, we worked together for several years at Reasoning Mind, an education technology company now part of Imagine Learning. From Reasoning Mind, I headed off to McKinsey while Victor joined BCG.

I really enjoyed working at McKinsey, and was really lucky to have awesome coworkers and clients. One thing I noticed at all the companies I advised was how BUSY everyone was. One implication of that: the technologies people need aren’t the ones that are effective — it’s the ones that are effective AND save you time.

At BCG, Victor did a lot of work using AI to improve offers. He realized that for a lot of companies, the best way to personalize offers isn’t a bespoke machine learning model, but rather a software-as-a-service product.

We hope OfferFit will make marketers say, “Wow — that was easy!”

Why are we so excited to be doing self-learning AI for marketing?

We believe that promotional offers are just the beginning.

In the next decade, self-learning AI will change how marketing happens.

Broadly speaking, there are two aspects to marketing:

  • Creating: defining the brand, selecting the message, writing copy, etc.

  • Tinkering: experimenting to find the optimal parameters for marketing actions — which asset to deliver, when, and to whom.

Marketers will want to delegate tinkering to self-learning AI, so they can focus on creating. And we will see self-learning AI solutions appear for all aspects of marketing: ads, website optimization, lead nurturing, loyalty programs… the list goes on.

Let’s make it happen together!

Experimentation unleashed


Ready to make the leap from A/B to AI?