Empower Marketing Incentives With AI

Nine Best Practices For Creating Marketing Incentives

As rates change and competition for deposits continues to increase, getting incentives right is a core aspect in the process of unlocking enhanced customer lifetime value (CLV). Here are a few best practices to consider when approaching incentive management. 


Measure Twice, Cut Once 

Any marketing venture needs to have measurable, targetable and relevant KPIs tied to the bank’s overall objectives. Set as program goals within the Amplero Personalization Optimizer interface, it is critical to get these right in order to engage with the libraries and leverage the machine learning to optimize for that result. Ask yourself, what are you optimizing for? Revenue, adoption, net revenue… Amplero’s reinforced learning will optimize for whatever objective function the marketer sets. If you set your optimization KPI to revenue, for instance, it will maximize for revenue. But beware, that will probably lead to giving offer dollars away because it is not optimizing for NET revenue.


Don’t Trust Your Gut

Over time, marketers get a sixth sense of what promotions work best based on past performance. “25% off has always been our best performing offer, so why try anything else?” Fight that instinct because you just end up regressing to the mean. No customer is an average. Be agnostic to your own preferences or preconceived notions and uncover what promotions perform best for each individual customer, not the average one.


Change Your Frame Of Mind

The heavy lifting and time required just to get a campaign out the door – let alone measure its performance – has always blocked marketers from experimentation, and limited their thinking to what they feel is the, one, best offer. Now, with machine learning capabilities, marketers are free to explore and test different offers – even ones they think won’t work – easily. Their mindset changes from narrowly focusing on the one-best-offer to exploring all the potential offers that could work for any given customer. Your new best practice mantra is to “test, test, test”.


Sometimes The Best Offer Is No Offer

The more versions and parameters you test, the smarter your marketing becomes, and one of the variations should be no offer. Think of it as a control group that gives you insight into the incremental lift you are getting from all the other offer variations. Without having a no offer control group, you could be leaving money on the table or, literally, putting money in your customer’s pocket when you didn’t have to. This is easily created within the Amplero platform.


Don’t Test On The Fringes

Test offer variances that are perceptibly or meaningfully different, to a customer. Start, for example, by testing standard numerical offer ranges, such as $5 vs $10 rewards against $5 vs $6 offers, and of course you can iterate over time by introducing finer grain amounts. However, you are likely to find that other offer variables, for example offer expiry or related perks, will drive greater performance. This can be part of your exploration process when building and editing your incentives library.

Remember: Customers don’t read marketing; they scan it, so testing on the fringe will likely result in scant perceptible or significant variances in performance because the customers doesn’t perceive the offer choices are any different.


Set Your Horizon Differently

Look beyond response rates to business impact. Way too often, marketers measure promotional offer success on whether the campaign drove immediate action. Ask yourself: Did the promotional offer change customer behavior for the positive or the negative? For instance, did it create a Pavlov response where customers now only act if they are incentivized to do so? Did it improve net revenue on a trailing 14-day period or, better yet, has the customer’s lifetime value gone up or down by offering this promotion?


Decompose An Offer To Its Atomic Elements

When people think of offer testing they often only think about the offer itself. Offers are really comprised of components of a deal: Test offer (in category like a discounted loan rate or out of category like a gift card), offer type (discount, bonus, free with purchase), value ($ vs % vs units), offer expiry (only good this week or while supplies last), condition (BOGO, Spend X to get Y). Think of these components as parameters that also allow you to set boundaries for how much (X) you are willing to give, to get (Y) in return.  For example, we’ll wave your Non-Sufficient Funds charges if you sign up for direct deposit.


Think Big Then Start Small

Amplero lets you test an incredible number of permutations, as our closed loop recursive learning capability automatically expands your testing. The idea of testing hundreds, let alone even dozens of variables can be daunting for marketers who are used to A/B testing or even rules-based journey solutions that require a lot of upfront set up. Start by testing a few variables, for instance one offer type with two offer levels, two offer conditions and two expiries and two message tones. This simple two by two scenario, alone, yields 16 different permutations. By starting small, you’ll get in the rhythm of learning fast, iterating faster and getting big.


Attitude Matters

When creating offer messages, marketers lose their focus on the message and think only about the offer and it’s parameters. What they miss is the tone of the message. Tone evokes a reaction that can be just as influential as the incentive itself. With Amplero’s contextual multi-armed bandit testing capability, marketers can uncover customers who react more favorably (or not) to a message’s tone. A simple best practice example is varying the call-to-action tone, for instance, a reward tone like “A special offer for you” versus a punishment tone like “Don’t miss out; offer ends today.” How you frame your messages will be a core part of how you approach your incentives.

  • Author
  • Want to go further?

    Contact us to learn more about how Curinos can help you navigate today and prepare for tomorrow.