Choosing your recommendation system

Tupperware meetings, one of the oldest forms of referral!

A few weeks ago, I published this article about the referral system of Growth Room, built with Mailchimp: Building a referral system with Mailchimp

In talking to those who have read it, I thought it would surely be useful to summarize the possible options for recommending and sharing. Hence this article 😀

Why a referral system?

When it comes to growth hacking, there are two major inflection points to work on:

• The organic growth loop: activation — retention — recommendation. In other words, this is called virality: getting your users to like your product, inviting their friends, inviting their friends to convert, and in turn liking your product, inviting their friends, etc.
• The acquisition: once this growth loop is in place, we still need to find a way to grow even more quickly. It's acquisition, or how to find techniques to open the floodgates and bring in a constant flow of customers.

The recommendation system is therefore important from the start of the startup's life. This is what will allow you to grow organically.

The k-factor

In order to build the right mechanism, you still need to know how to measure the recommendation. For this purpose, we use the K-Factor (or coefficient of Virality). This is the average of the number of additional users that each initial user brings in. It is calculated as follows:

k = number of people invited * registration rate of these people

So, if I invite 10 friends and 20% of them sign up In fine On the service, I have a k-factor of 2. Of course, the aim is to have a k-factor greater than 1;)

Maximizing the k-factor therefore involves two things:

• Maximize the number of people invited: To do this, no secret, you must already have very good retention. Indeed, if your product does not satisfy the customer, it is unlikely that the customer will recommend you. Then, you have to propose a system that makes sense for your users — which we will detail below.
• Maximize the registration rate of invited persons: It's essentially an activation topic. Once the person is invited, be sure to remind them why and by whom they are there, your value proposition. Take her to her AHA moment (i.e. the first positive experience with your product).

Word of mouth and NPS

Before building a complex recommendation system, you need to start by asking yourself if users don't naturally tell their friends about your product. This is the famous word of mouth, a very powerful recommendation tool but also very difficult to measure accurately. As a first approximation, we can therefore use the Net Promoter Score (NPS).

NPS is a marketing tool that involves asking a company's customers how likely they would be to recommend the product to their friends. Concretely, it often takes the form of an email sent after purchase or after use that asks this question and offers to answer with a score between 0 and 10.

While NPS is a good way to identify power users, it is still a fragile tool. On the one hand, it does not capture the entire word-of-mouth phenomenon (some do not answer the questionnaire, others will give a score of 9 or 10 but will not recommend). In fine, etc.). And on the other hand, it does not allow us to learn more about the motivation of users to share.

Plus, it's still a bit scary for the user. Nobody likes to be in a situation where they rate a service. Ask the question in binary form: are you satisfied/not satisfied, or, in a more visual form, 👍/👎.

The question of incitement

In order to encourage referrals, it is therefore necessary to create the most appropriate system to encourage natural sponsorship behavior. To this end, a commonly used technique is an incentive system: that is, offering something to the user in exchange for inviting a friend. This reward can be of three types:

• Social (when your product allows the user to value themselves socially),
• Non-monetary (when you offer exclusive content, a feature or a bonus)
• Monetary (when you offer a voucher, a coupon, etc.).

Social motivation

Let's start with social motivation. Not all products allow this type of incentive to be used, but they generally have the following characteristics:

• They are consumed by several people or are gifts: the aim is to ensure that the user invites friends to consume with him (example: sharing a trip in Uber).
• They don't work without inviting other people. For example, this is the case with social networks. The aim is to create a network effect: the greater the market share within a target, the more value the product creates for users.
• They give rise to a particular social status. This is the case of luxury products or exclusive clubs, which allow the user to belong to a privileged social category.

The reward

Sometimes, however, it is necessary to reward the user for sharing. Indeed, some products are not ideal for generating natural sharing. While the reward may be monetary, it's always a good idea to start by looking at whether it's possible to give something that's free for the business but offers a lot of value to the user.

Here again, it is possible to distinguish a few main models. We can offer:

• Content (a white paper in exchange for an email for example).
• Access to an exclusive group (it is about offering a VIP treatment, i.e. giving access to privileged events or an exclusive product).
• An additional feature (a method widely used in games: the more you advance, the more levels you unlock).

The monetary incentive

Finally, the last possible type of incentive is monetary incentives. This system, made very popular by Uber or AirBnb, uses a simple mechanism: offer €X in credit to your friends, earn €X when they register. It has the disadvantage of placing the user in a very utilitarian logic but has the advantage of being extremely effective and motivating.

How to properly determine the amount offered? Above all, it is a story of two parameters:

• The reasonable acquisition cost: you need to find a reward that is attractive enough to trigger registration but that does not cost you too much either, otherwise you may never make the money offered profitable.
• The rate of use of the product or service after sponsorship. That is why it is absolutely crucial to measure the life value (LTV) of sponsored users separately from others and to always check that it is not too much lower than the average LTV.

In order to define the right amount, there are a multitude of strategies. First of all, do not forget that the amount is what you offer to the sponsored person + what you offer to the sponsor. Ideally, this sum is recovered from the first transaction. If this is not the case, a fairly reliable method is to limit yourself to offering approximately 30% of the observed LTV.

Conclusion

Whatever the mechanism chosen, it is important to keep a few best practices in mind:

• Personalize the recommendation: The main advantage of the recommendation is that the user who receives it sees the message as coming from their environment and not from outside. This is why it is essential to play the personalization card: specify who the sponsor is (name, first name, photo and any other useful detail), why he invites the sponsored person and what the sponsor has to gain by registering.
• Always give priority to live recommendations: it is generally more effective to ask sponsors to target 2 or 3 people around them and convince them to register than to flood all their friends without putting any real personalization into them.
• Strictly punish opportunistic behavior: unfortunately there is no perfect mechanism but it is possible to monitor what users do. In case of suspicious behavior, actions can always be taken by hand at the beginning and then automatically afterwards.