Deletion Using TTL
13 Sep 2026 » Platform
In my post on Individual Profile Deletion I explained a surgical approach to deleting profiles. This approach requires you to actively request the deletion of the data. With the Time-to-Live (TTL) configuration parameters, you can let the Adobe Experience Platform (AEP) take care of the deletion for you in certain scenarios. You set it and forget it.
Edge
The AEP Edge Network has its particularities when it comes to managing its data lifecycle. The way to understand it is very simple. The Edge Network receives hundreds or thousands of requests per second and needs to respond in a few hundred milliseconds. If it had a copy of all the data, either it would take seconds, or we would need thousands of servers to manage a single customer.
The solution to the previous issue is obvious: limit the amount of data an Edge server keeps. This limitation comes in two flavors:
- Edge profiles are kept for up to 14 days after the last time that profile was active.
- Edge events are deleted after 24 hours, which I am sure you already know if you understand the edge segment qualification rules.
Events in RTCP
When we talk about the concept of “profile” in the Real-Time Customer Profile (RTCP), we tend to just think of profile attributes. However, AEP profiles are actually comprised of both profile attributes and events.
This has three major consequences:
- Profiles grow in size with events.
- Batch segmentation will take longer the more events it has to evaluate.
- Some profiles may only be composed of events, but they still count towards the number of profiles in the license.
To avoid profiles from becoming too big, AEP offers a simple solution: set a TTL for event datasets in RTCP. This is the span of time an event will be available in the profile store for segmentation.
One typical pushback that we get from customers when we talk to them about setting this TTL is that they want all events in the profile. In this scenario, I start by questioning the need: Marketing campaigns have a greater impact the closer to recent customer activities; as time goes by, the response is going to decrease. If there is a real need for old information, I then recommend converting major events, like transactions, into profile attributes. This can be done using Data Distiller or Computed Attributes.
Events in the Data Lake
You may have noticed that the UI for the configuration of the TTL for event datasets in RTCP also offers another setting, the TTL for the Data Lake. The idea is the same as above: you define an expiration for the events in the Data Lake, after which the Data Lake will automatically delete old records. This parameter is not only useful to prevent your event datasets from growing indefinitely, but to comply with your regulatory mandates for data storage.
Anonymous Profiles
Whenever a new identity is received by AEP, a profile is created and the profile count is increased. This means that all anonymous visitors to your website will increase the profile count, as each of them will have, at least an ECID. If an anonymous visitor never comes back or deletes their cookies, you have a profile in the system that is totally useless, but counts towards your license.
To solve this conundrum, AEP has a setting to delete old anonymous profiles: Pseudonymous profiles data expiration. You basically set an expiration, which is the period since the last visitor’s activity, after which the anonymous profile is deleted. This also means that, as long as an anonymous profile is active, it will be kept in the RTCP.
Let me clarify one detail. While I use the word “anonymous”, the documentation uses the word “pseudonymous”. I know there are subtle differences, but I will equate them and use my preferred word.
Finally, if you read this section and compare it with Events in RTCP, you will notice that there seems to be some overlap. If you set the event dataset TTL, you should be able to delete anonymous profiles that are only composed of events. So, why is this other feature needed? I am not 100% sure. However, I can guess that an anonymous profile may contain more than just events, through other AEP features, in which case an event dataset TTL would not be enough. I will get more details and update the post.
Typical Values
When setting these values in AEP, our customers ask us for help in setting them. There is no one-size-fits-all set of values. We need to review the use cases, the data, and the contractual obligations to make a suggestion. However, typical TTL values that we see among our customers are:
- RTCP: 30, 60, 90 days.
- Data Lake: 13, 25 months.
- Pseudonymous profiles: 14, 30 days.