How to find and describe the paying segment of users?

We all know that users are divided into groups according to user metrics and product market share (ABCDX segmentation). And this division tells us: do everything in the product for the A, maximum, B-segment – for the most paying and loyal customers. And do not pay too much attention to the “complaining” and low-paying segments C and D that do not reduce the economy of the product.

Today, on the eve of the start of the course “Product Manager of IT projects” together with Sergey Koloskov, Otus teacher and channel author Fresh product manager we wrote a post on how to find a paying segment for three different cases.

If there are metrics / analytics for the product

1. Analysis “Frequency – Money – Date of last action” (RFM – analysis “by the classic”): a table of scores for indicators Recency – age (how long ago did your customers buy something from you), Frequency – frequency (how often they buy from you), Monetary – money (total amount of purchases ). Knowing these three metrics in general terms, the leaders in terms of amount are found, and they become for us the identified paying segment. For instance,

R – Recency:

1.0-30 days: 5,100 buyers / subscribers

2.31-60 days: 12,300

3.61-90 days: 32,800

4.11-180 days: 75,000

5.181-365 days: 123,400

Group 1 hereinafter will include the most profitable clients. The “best” segment will always have the lowest number of buyers / subscribers, while the “worst” segment will always have the largest.

F – Frequency:

1.16+ purchases / actions

2. 11-15




M = Monetary

1. $ 1001 +

2. $ 601-1000

3. $ 351-600

4. $ 201-350

5. $ 0-200

Thus, we get almost 125 groups. Clients 5R-5F-1M could apply to your services only once, spending a considerable amount, but not counting on long-term cooperation. But 5R-3F-1M are more likely to become dissatisfied with the company’s services or lose interest in it. 1R-1F-1M is the cream of your customer list. If you have correctly chosen the limits of groups R, F and M, then this segment should be extremely small – no more than 5% of the address base. Whatever you do, you are unlikely to be able to ruin your relationship with such clients.

2. Problem-solving interviews (as part of CustDev) in order to identify common patterns and characteristics among paying (from 8 problem interviews per prospective segment with hypothesis testing) – we conduct a problem-solving interview in order to identify the connection between the hypothesis and the segment. There may be more than one segment, raw segments can be taken from a complete or simplified (by several indicators) RFM analysis. Ideally, when CustDev complements RFM analysis and vice versa. About how to properly conduct casts, what questions and how to ask in review from Fresh Product manager channel.

3. Study of the first session of the most active users (search for the moment of the highest happiness from using the product – Aha-moment) – a method of focusing on all the key actions of users who eventually became active or paying users. A correlation is sought between the action and the segment.

If there are no product metrics and analytics

1. Simplified RFM Analysis (by revenue, by average check) – collected from what can be collected from the back-end of the product or surveys by possible indicators. Sometimes they are enough to identify segments. That is, the simplest unloading from the database and logs.

2. Segmented polls – collecting answers to questions within the product that reveal patterns of behavior. Possibility of setting questions for users with certain behavior. Surveys are best used precisely in terms of checking the number of scenarios: the question is the formulation of the problem, the answers are the options for solutions. These numbers can be used for the Reach metric in a hypothesis prioritization framework (such as RICE). Those scripts that received few answers are not needed from a product point of view.

3. CustDev in order to identify insights to build analytics. And, of course, conducting problem-solving interviews, in which there will be the results of a simplified RFM – analysis or polls. The goal is still the same – it is necessary to identify, through open-ended questions, similar patterns of behavior and unifying factors. The topic is excellent Cindy Alvarez blog, author of the book “Product to Buy” and channel with my notes about CustDev

If there is no product and analytics

1. CustDev for the purpose of forming segments and testing hypotheses (From 40 problematic interviews). When there are no leads and analytics, hypotheses are collected for both products and segments – active users of products. The most daring hypotheses, the most adventurous test.

2. Informational ways to attract customers (exhibitions, webinars, articles, distribution). It’s more about the minimum required work products without the involvement of code (MVP), when it is not clear for whom the product is. All sorts of news feeds can attract the most useful segment, it is important to “get to know” everyone and find out what is common between the representatives.

3. Checking search queries in Yandex / Google (Wordstat and others) as monitoring monetization and the need for a product by semantics, checking the presence of a target audience by checking the pool of queries in search engines. It is important to dig precisely narrow target queries. For example, “Math Tutors” will show popularity for the “OGE / Grade 9” segments, cities and intersections with other subjects.

That’s all. We invite you to visit a free demo lesson on the topic: “CJM (customer journey map): from hypotheses to product scenario”

Read more:

  • No-code: products versus spending money

Similar Posts

Leave a Reply