
To personalize retention messaging in Braze, start with Liquid fields that always resolve, then add a churn prediction to decide who hears from you. Content Blocks keep the copy consistent across channels, and abort logic stops sends that would read badly. Build in that order, so each layer has something solid underneath it.
Key Takeaways
- Set a Liquid default value for every field a message uses, and check for whitespace-only values, which the default does not catch.
- Predictive Churn scores run from 0 to 100, with Low, Medium and High Risk bands you can use as filters in segments or campaigns.
- Some high-scoring users would not churn anyway, so we suggest keeping discounts for the tier that needs them.
- Content Blocks inserted with Liquid stay linked to the original, while drag and drop copies are not linked.
- Add explicit nil or empty checks before
abort_message(), and remember an aborted Canvas step lets the user continue. - Build in order: data, churn definition, risk tiers, shared content, then abort logic.
What does personalised retention messaging mean in Braze?
Braze describes Liquid as a way to build templates that change their content based on each recipient's profile data, behaviour or language. For retention, that means the person who has gone quiet sees a message that reflects what they did, instead of a broadcast.
The documentation also recommends testing every Liquid message before it goes out. We'd go further: treat a retention message as a set of branches, and test each branch with a profile that takes it.
Profile fields such as a first name or a language may count as personal data, and this post doesn't settle the legal question. Treat consent and data handling as a separate review with your privacy team.
How do you personalize retention messaging in Braze without blank fields?
A frequent failure is a message that opens with a missing name. If no default value is set and the field is missing on the profile, the field will be blank in the message.
A default value fixes this, but there is a catch worth knowing early. The default shows for empty values and not for blank values, and a blank value is one that holds only whitespace.
So a name field containing a single space will slip past your default. We'd check for that in the data before launch, not after the first send.
Hi {{${first_name} | default: 'there'}},Our case study on unifying guest profiles covers consolidating fragmented guest profiles so personalised journeys have one record to read. For the same groundwork on your own data, see our Customer Data Management service.
How do you use churn risk scores to decide who gets a message?
Predictive Churn gives each user in the prediction audience a score between 0 and 100 showing how likely they are to churn under your definition. Braze's stated purpose is that you can contact at-risk users before they churn. The definition is yours to set: something a user either does once or stops doing within a window you choose.
Braze labels scores from 0 to 50 as Low Risk. Scores from 50 to 75 and from 75 to 100 are labelled Medium Risk and High Risk. You can use the output as filters in segments or campaigns, so you need no extra plumbing.
| Risk category | Score range | How we'd approach the message |
|---|---|---|
| Low Risk | 0 to 50 | Leave in the regular programme, no retention push |
| Medium Risk | 50 to 75 | Helpful content, a reminder of value, no discount |
| High Risk | 75 to 100 | The strongest message, narrowly targeted |
The third column is our judgement, not Braze guidance. The ranges come from the Predictive Churn analytics page.
How should you handle false positives in a retention audience?
Some users may have a high score and never churn. Braze notes they would not churn even if you took no action.
For a valuable promotion, the documentation suggests keeping non-churners to a minimum while still reaching as many expected churners as the model allows.
That is the argument for holding back discounts from the Medium tier. A discount sent to a user who was staying anyway costs margin, and it also teaches them to wait for the next one.
A reminder costs almost nothing, so it's the safer first step for anyone the model is unsure about.
How do the prediction settings affect your targeting?
Two settings shape what the score means. The churn window can be set for up to 60 days, depending on the available data, and the definition has to match how your customers actually behave. Braze's own example is that within 14 days a user does not place an order.
The prediction also retrains automatically every two weeks, which is a separate process from when scores are produced. A preview prediction will not update scores regularly or let you target users on its output. Check the current Braze page for how many predictions can be active at once, because the documentation pages we read give different figures.
What does a hypothetical worked example look like?
Take a hypothetical subscription retailer that wants lapsing customers back. Braze gives the example of creating two predictions so that new subscribers can be messaged differently from longer-term subscribers.
The retailer builds one prediction for each group, and each group gets its own copy. A customer in their first weeks needs different reassurance from a long-standing one.
Within each group, the retailer sends Medium Risk users a message built around a Content Block with the product guidance, and High Risk users a narrower offer. A Liquid default covers the first name.
Any send that depends on a missing attribute is aborted and never goes out half empty. This is an illustration only, and it uses no client data and no results.
How do Content Blocks keep retention copy consistent?
Content Blocks let you manage reusable, cross-channel content in one place. A block inserted with Liquid stays linked to the original and reflects later changes to the template. A block added by drag and drop is not linked.
Content Blocks can be nested, but only once. For a retention programme this matters in practice: a shared footer or offer terms block can be changed once and flow into every linked message, which removes a common source of mismatch between email and push.
Catalogs work alongside blocks, since they enrich messages through Liquid in the same way as custom attributes or custom event properties.
When should you abort a retention message?
The abort_message() tag stops the send for users who don't meet your condition. The message doesn't appear on their profile and doesn't count toward deliveries or frequency capping. Braze says to add explicit nil or empty checks before calling it.
Be careful with the else branch. An else branch that aborts when a value is missing stops the send for any user without that data, which can shrink your audience without anyone noticing.
In a Canvas, an aborted message step doesn't remove the user, who proceeds to the next step. Abort logic is evaluated at send time, so a user's data at that moment decides the outcome.
{% if ${language} == 'en' %}
Your next order is waiting, {{${first_name} | default: 'there'}}.
{% else %}
{% abort_message ('language was nil') %}
{% endif %}What order should you build this in?
- Fix the data first. Check which profile fields are empty or blank, then set defaults.
- Define churn in terms your customers recognise, then build the prediction.
- Segment by risk category and decide the message for each tier.
- Put shared copy into Content Blocks, inserted through Liquid.
- Add abort logic with explicit checks, and test each branch with a matching profile.
We put the data step first because every later step reads from it. A prediction can be excellent and the message still arrive with a blank name.
How does this connect to the rest of the programme?
Personalisation sits alongside several other levers. For the mechanics of the tags themselves, see Personalization at Scale: Using Braze Liquid Tags. If you're weighing Content Blocks and Catalogs, Braze Advanced Features: Content Blocks, Catalogs, Extensions covers both.
On targeting, Using Segmentation for Smarter Retention Campaigns in Braze is the natural next read. For timing, see How to Trigger Retention Messages at the Right Time in Braze.
Once someone has lapsed, How to Build Effective Win-Back Campaigns in Braze picks up the story. Before launch, How to Avoid Common Personalization Pitfalls in Braze is a useful checklist.
For a view of bringing lapsing users back through a lifecycle journey, our case study on interactive rewards for re-engagement covers a retail programme.
What does CustomerIK recommend?
As a Braze implementation partner, CustomerIK recommends treating personalisation as a chain of small safeguards. Get the data right, define churn carefully, match the offer to the risk tier, and abort anything that can't be personalised properly. The documentation doesn't publish a lift figure for any of this, so measure your own results before you scale a tactic.
Frequently Asked Questions
1. Can I reference the churn risk score inside Liquid?
We haven't found that documented on the pages we read. Braze describes using the prediction's output as filters in segments or campaigns, so that is the route we'd use.
2. Does Liquid default work for blank values?
It doesn't. The default shows for empty values but not for blank values, and a blank value contains whitespace and no other characters.
3. What happens to users when abort_message() fires in a Canvas?
The user doesn't exit the Canvas. They proceed to the next step.
4. Are linked Content Blocks updated automatically?
Blocks inserted via Liquid are linked to the original and reflect changes to the template. Blocks added by drag and drop are not linked.
5. Does an aborted message count towards frequency capping?
It doesn't. Braze says the message won't appear on the profile and won't count toward deliveries or frequency capping.
Sources
- Braze, Liquid reference. Retrieved 06 Nov 2026.
- Braze, Set default values. Retrieved 06 Nov 2026.
- Braze, Conditional messaging logic. Retrieved 06 Nov 2026.
- Braze, Abort messages. Retrieved 06 Nov 2026.
- Braze, Predictive Churn. Retrieved 06 Nov 2026.
- Braze, Create a churn prediction. Retrieved 06 Nov 2026.
- Braze, Predictive churn analytics. Retrieved 06 Nov 2026.
- Braze, Content Blocks. Retrieved 06 Nov 2026.
- Braze, Catalogs. Retrieved 06 Nov 2026.
- Written by CustomerIK, a Braze implementation partner.
CustomerIK is a Braze implementation partner working across onboarding, technical integration, marketing operations and customer data management.
If you're planning retention journeys in Braze and want a second view on the data, prediction and abort design, let's talk.




