Close this search box.

Segmentation in emailing: how to take it to the next level?

On Thursday, November 25th (Happy Birthday Catherine!), we organized a live session on customer segmentation with Cédric LetessierDATA consultant and co-founder of Meetyourpeople. The objective was to answer these questions very concretely: how to go beyond a "prospects / customers / ex-customers" segmentation? Which strategic segments work well? How do you go about setting them up? With which tools (or not)?

Below are the main lessons learned from our discussion. But for those who prefer audio, you can listen to the replay (the image has very little added value ;-))

We quickly talked about affinity segments

The discussion quickly turned to the establishment of affinity segments. In our jargon at Badsender, we call them "segments by interests" but it's the same thing ;-). Cédric gave us examples with the FFF or Le Figaro.

  1. The important thing is to first make a list of affinities that is interesting to collect. This list should be fairly limited. You don't have to have 30 themes. The less the better 😉 Like 10 max if you have a large universe of products. Feel free to group themes into "Master Themes", otherwise it gets messy fast. You also need to have these themes correlated with the shopping categories.
  2. Then, from look in the available data, what exists to qualify the affinities.
  3. Finally, to automate processes in order to qualify better and better as time goes by

It takes time. You can't do it in a month. It's a long-term job. This logic of qualification of individuals must be anchored in all the "touch points" of the contact, whether it is a customer or a prospect.

Automate by categorizing emailing opens and clicks

To qualify the affinities, we will look for information in the openings and clicks of emails. A campaign can be categorized on a specific theme. Example: 1 campaign on women's soccer. If a contact opens this email, I can add "+1" in the affinity "Women's Football".
We can also categorize the links present in a campaign. Example: in 1 rather generic campaign, I have 3 links: 1 on women's soccer, 1 on basketball, 1 on tennis. Depending on the clicks, I will add "+1" to the right affinity.
Technically speaking, the categorization of clicks is done when the campaign is set up in the routing tool. Most of the routing tools proposes to add a tag to the campaign and to the links present in the email. You will then have to create the segments per tag that will feed themselves.

If I haven't done this categorization work, how can I do it?
We can rework the data and search the whole history. That is to say: take the last X months (12 months if we can, the maximum is the best). Re-theme each of the campaigns and URLs. It is always feasible. Depending on the situation, it can be worth it, sometimes not. But above all: you have to define processes where you will succeed in AUTOMATICALLY collecting affinities throughout the year.

Automate affinity collection

Once the affinities/themes have been defined, they must be included everywhere in order to qualify the database as quickly as possible: from the collection form, in the preference centerin the welcome scenario emails, in customer satisfaction collection emails...

How do you do it? Then either :

  • in survey mode by asking a question directly (declarative data)
  • in behavioral mode, by categorizing the links / visuals of the content blocks (behavioral data)
  • in the satisfaction questionnaires (declarative data)

You can also mix with other data!

If we have the data available, we can mix :
opening + clicks + purchase categories + reporting data + navigation data + timing!

Attention: We will not give the same weight to declarative data as to transactional or behavioral data. Timing is also important! Someone who hasn't bought in 12 months, or who hasn't clicked in 6 months, won't have the same score as someone who is more active.
At this stage, there is a phase of study, algorithms, data science to get out homogeneous populations (because there is the stake!) and management rules.
Then, these rules can be implemented from a technical point of view and automated in a database.

Do you need the help of a Data Scientist?

  • If we limit ourselves to the categorization of opens and clicks. No. The routing tool may suffice. This is more about aggregation and not scoring. You don't need a data scientist for that.
  • If you are in the pure player e-commerce business, it depends on your tools 😉 There are great routing tools for that!
  • If you are in another sector or if you mix several data sources (ecommerce, stores, call-center...), you need to call upon a Data Scientist. There are some tools that allow you to do scoring. But it's still something quite simple that you can't necessarily modify. So when you have a business that is not recurring, that is not in ecommerce, scoring is not necessarily very well adapted. So you have to call on a data scientist. You have to extract the data to make a study and analyze it. We get out of the business rules. When you have the rules, you can ask the IT department to implement them automatically. It's better to calculate the score in the CRM database (i.e., outside the routing tool).
Example of ready-to-use strategic segments in the tool Klaviyo. The criteria can be modified. When you have only one data source like the e-commerce site, it's great.
Anti-churn scoring embedded in Klaviyo.

Which segments are working well?

  • The Affinity ScoreThat we understood 😉
  • The Promophile score to know the degree of appetence of the contacts to the promotion...or not. This allows you to not target (or much less) promophiles during hyper promotional events such as Black Friday Christmas Winter sales
  • Seasonality score Identify contacts who buy from a certain period rather than another (particularly useful in the travel/hospitality industry... though)
  • The "Booking window" score : it's the time lapse between the purchase made and the date. More like travel/hospitality sector... although bis 😉 It can be done very well in other sectors. For example: those who do their Christmas shopping at the last minute vs. those who go for it in November.
  • Email channel appetence score This is the frequency of email openings. It is about grouping the contacts who open 10 times out of 10 the emails sent versus those who open 1 time out of 10. This can be useful, once again, in periods where the marketing pressure is enormous. You can't afford to send 1 email per day to your entire base. There is too much risk of generating complaints, thus harming the reputation of the sender and to message delivery. They use this score to detect those who are mega-appetitive to the email channel (they will be more targeted during the period) vs. those who are hyper-refractory to email (they will be much less targeted). By the way, we talked about it here in this article on how to target inactive email contacts during the Christmas period.
  • The channel appetence score to favor one channel over another. For example: not at all keen on the email channel but keen on the SMS channel.
  • The lead nurturing score HYPER cool in BtotB (although ter ;-)) We use it thoroughly at Badsender.

Why is navigation data difficult to collect?

  • Because people don't lodge! Therefore, few email-cookie reconciliations are possible, volumes are ultra low. If you really want to collect browsing data, you have to accompany the contacts when they connect: remind them in the emails of the reasons why they are interested in connecting.
  • Because technically it's not easy. Even if it is always less complicated than before, the tools are evolving.

So, it is better to focus on the base: clicks, openings, purchases, declarative data of the surveys... it is necessary to capitalize on it. When we have done all this, we will ask ourselves the question of navigation.

Are the strategic segments usually the ones that were anticipated upstream?

It depends. Sometimes beliefs are confirmed after the audit. Sometimes not at all. In any case, new populations are detected and we even go as far as defining the brand's personas.

So we summarize!

If you want to build your strategic segments tomorrow :

  1. Conduct a data audit
  2. Simulation of strategic segments until homogeneous segments are found
  3. Define the business rules and ask your IT department to automate them
  4. Make strategic segments accessible from within your routing tool
  5. Automate the data collection processes to be qualified

Keep in mind: you have to do according to the data you have at a given time. You have to adapt, otherwise no activation will be launched.

If you want help building your strategic segments, do not hesitate!

The author

Laisser un commentaire

Your email address will not be published. Les champs obligatoires sont indiqués avec *