How to increase retention with incentives, not features

The problem

One of the first things I looked at upon joining Deel was churn reasons for my product. I was hoping to find some quick wins and close some feature gaps, but instead, I found a less juicy picture: #1 reason for churn was client disengagement (they stop using product, responding to emails, and/or paying their bills).?

As a PM - what can you build to address this? We’ll get to this in a moment, but a fair short answer can be “nothing”.

Now at the same time, majority of multi-product companies, and especially feature factories, struggle from having too many features released with <1% adoption. All the cases when the feature was built to win a big deal, and all the shiny objects start to bite back. I’ve seen this too at my other jobs and companies I’ve advised. It turns out both of these are very common scenarios.

These two facts put together didn’t make sense to me. How can we be losing clients due to disengagement if we have such a vast array of great value (aka features) available for them? Luckily I got a chance to solve these at Deel. I found out that Head of Customer Success, Luke F. is looking to fix this problem too, and so we partnered up.

I think what we came up with is quite an elegant management practice that I’m very proud of. Essentially, we incentivized CSMs to increase adoption of retention-driving features among their customer portfolio. There are 2 key hypothesis' behind this:

1. Adopting certain features leads to higher client retention (classic thesis in the product world)
2. Tying CSM compensation to the adoption of these features will increase the adoption?

Now, let’s dig into how to set this up.

Pick retention-driving features

A valid approach to lower your risk is to run a retention correlation analysis, i.e. compare clients who churned with clients who didn’t churn and slice them by the adoption of certain features, geographies, job titles, interaction with customer support, etc. As a result you’ll see some data like: “clients who used feature X retained at 90% while those who didn’t retained at 70%”. Now, there are a lot of caveats here that you need to be careful with: correlation doesn’t mean causation; clients using ANY feature usually retain better than those not using any feature (self-selection bias); volumes in the B2B enterprise world don’t give you stat significance, etc etc. But it can be a good starting point, and to get the budget approved. Hint, Reforge has a great course on this retention correlation analysis (in b2c world though).

The main danger with this approach is that it’s very easy to get lost in analysis paralysis, spend weeks or months without making progress, and kill the momentum and entire initiative. Thus I recommend taking a lesson from Deel’s book, turning on “Deel Speed” , and considering a second approach: use your best judgment and some of the best practices. First of all, integrations are usually a safe bet. I’ve seen it increase retention by up to 30% in Miro and other companies. Second, a combined team of founder, product, and CSM leaders probably have a good idea about which features have a good chance of increasing retention, based on their intimate knowledge of their clients and industry.

Lastly, number of features is important. If you go too broad, you’ll have a hard time promoting them to CSMs and clients respectively as it’ll get overwhelming - you can’t pitch 100 features on a call or in an email. On the other hand, if you go too narrow, you’ll make it very hard for CSMs to do their job. They need to have some room in picking the features to recommend, as different clients will have different best matches (based on industry, client size etc).

A good number we landed at Deel was 20 features. Remember, for each feature you’ll need to prepare a pitch, a target client profile, definition of adoption, and add them to a dashboard (see an example below). And you’ll need to work with many PMs to nail this for the variety of functionality you’ll need to cover.

Define the adoption of such features

The nitty gritty part of this process comes down to meticulously defining the adoption of each feature. A common mistake would be to stay generic and say something like “we consider the feature X adopted if the client used it once”. But does "used" mean? And used when? By how many people in client’s org? And how many times? It’s critical to nail these right, as people’s livelihoods will depend on these definitions, and you need a transparent and fair way to communicate their goals to them.

Let’s take an example of the Equipment feature at Deel - it allows remote workers to rent their work laptops or phones (while Deel allows them to be hired remotely in the first place). This feature is a solid candidate for retention boost - it’s harder for clients to churn if a lot of their workforce has equipment on their hands provided by Deel (or their partner). A good definition of adoption in this case would cover a) how many workers in the company rented equipment b) time period over which equipment was rented. You could also add number of items rented, or specific type of items, if those have shown higher retention.?

Then you need to put some thresholds. If you’ve done retention correlation analysis - you’re in luck and can borrow data from that, if not, you can get creative. Continuing with the same example, my thought process was this: clearly, 1% of the workforce renting equipment is not enough to retain a client. 100% is unrealistic to achieve, so I’m shooting for something in the middle. Then I looked at actual data - out of clients who use the feature, what % of their workforce rent it on average? Then I increased it a bit to create some stretch goals for CSMs, and landed at 30%. So the final definition would be: “30% of client workforce rented 1+ piece of equipment in the last 6 months”

Once we rolled out the process, I naturally got a big pushback from CSM to change thresholds for some of the features. Sometimes they were right - we miscalculated. In some cases they were wrong - they just wanted an easier win. It’s a natural part of the process, but I would lean on the more aggressive side, as you’ll be able to find fair ground with the pushback, while you won’t find it if you start too low.

Set goals & incentives for CSMs

This part should be done primarily by CSM leadership, as it concerns compensation of their reports. But product can help. Here’s few key considerations:

• You’re balancing achievability vs aggressiveness. If goals are too high, noone will achieve them and you’ll seed demotivation. If they’re too low, you’re losing money and not making an impact you can
• Basic mechanic we followed is CSMs get X bonus in USD/month once they get Y% of their portfolio to use at least Z features. E.g. If 20% of their clients have 10+ features adopted in a given month, they get $400 USD bonus/month on top of their base pay (actual numbers were different).
• As the initiative is novel and just gets off the ground, it’s very important that CSMs feel it’s achievable, and a good approach is to create cascading goals: e.g. with above example, if they get 20% of their clients adopt 10+ features, they get $400 USD bonus/month. If they get it to 50% of clients, they get $1000 USD bonus/month.
• Calculating ROI is quite hard, and even though it’s possible, this still requires a leap of faith from leadership, as it’ll take several quarters to see actual retention lift (but it’s looking good at Deel so far!)

Launch tactics

There are a few tactics worth highlighting for the launch:

• Obviously, the core of it is training CSMs to mention the right features to the right clients in the right context (usually Quarterly Business Reviews, QBRs, but also during ad hoc check-ins and esclations), understanding all the features, learning to use the dashboard, etc. But this can be improved by the items below
• Automated email campaigns pitching the top 3 features for this client’s profile
• In-product promos of the same features to the same clients?
• Smart recommendations of features based not only on client profile, but their activity. If they have X adopted, we can recommend Y. Or vice versa, if we know they use X, we might wanna mention it in the promo campaign, so they feel more familiar (“ah, these are new feature recomendations from Deel. I’m already using X, so maybe they understand what I need. I should check out other recommendations too)
• Auto-generated QBR presentations based on selected features
• Braze and Gainsight are good tools for the above.

Why product team needs to be involved?

In my view, target features should be co-owned by product and CSM teams, and feature definitions should be owned by respective PMs (or product leaders). Because without this CSMs might be increasing adoption of features PMs have already killed (it’s a real-life story), or moved away from. At the same time, for thresholds and goals, CSM needs to have a big say.?

Conclusion

In itself, this is a great tactic to increase retention, which is showing promise in early results (Deel team multiplied the adoption of key features and is seeing a correlation with higher retention). But there’s also another bonus benefit: this results in a beautiful org setup, where two quite distant teams have the same objectives (increase adoption of certain features), and work together to achieve them, each using their own strengths: CSMs by increasing awareness and finding a personal way to persuade clients to adopt the features, and PMs by improving the feature experience and its in-product discoverability. At Deel, this became company-wide OKR which trickled down to hundreds of CSMs and PMs, and that’s when magic happened.?