We recently hosted an engaging AMA session and gathered some great questions from our Unify webinar. Below are the highlights and key insights from the discussions. We encourage you to share your thoughts and continue the conversation in the comments. Your feedback is invaluable as we strive to meet your needs and improve our offerings.
Editable Profiles
Q: I have situations where I'd like to be able to manually remove identifiers and traits from a profile without replaying events. Is Segment planning on supporting manual updates to profiles?
A: Profile Editability a granular set of data management controls including the ability to update and delete profile identifiers and traits. With this feature, Segment users would gain the flexibility to directly update and delete outdated and incorrect elements of the profile while preserving the golden profile. For example, Profile Editability would allow for the removal of incorrect email addresses from a profile, removal of outdated computed traits, and the transfer of a phone number from one profile to another. By being able to edit profiles directly, users can avoid costly and time-consuming operations related to event replays and decrease time to value while ensuring clean and accurate profiles for activation.
Q: Does the Profile Editability feature allow for editing individual customer profile traits or will it only be used to add/remove traits across all customer profiles?
A: Profile Editability would likely enable editing of individual customer profile traits at first but then we're likely to enable bulk updates and deletions as well.
Identifier Normalization
Q: Are you considering standardizing email addresses and phone numbers in order to reduce the number of duplicate profiles?
A: Yes. We plan to normalize identifiers as they are received in the identity resolution system to account for variations in phone number and email address formats. This would in fact improve profile merge rates and data quality, as well as making it easier to find profiles by identifier.
Merge Limits
Q: With internet browsers increasing their focus on privacy yielding more profiles to merge as cookie storage rules are tightening, what is the priority of increasing/removing the profile merge limit?
A: We do not have near-term plans to increase or remove profile merge limits. However, we are in the process of making major improvements to our backend architecture. Following this, we may consider revisiting the merge limits strategy. Additionally, we are actively researching methods for identity resolving anonymous profiles. The goal is to increase known profiles and profile merge rates in order to enable customers to reach more profiles with personalized messaging through the ad tech ecosystem.
Increasing 14-day Event History
Q: Are there any plans to get more than 14-day event history? The current look back window is pretty limiting.
A: With the new backend architecture mentioned in response to another question, we indeed have the foundational that will allow us to increase the 14-day event history in some cases. Additional development would be required to enable it and ensure that the API meets performance expectations. This investment is under consideration at the moment. Please work with your CSM to file a product feature request if you're interested.
Observability
Q: Is Segment going to provide any new troubleshooting or monitoring tools for ID resolution? It's hard to know what's going on when there is an issue or even before there's an issue.
A: New observability features we are considering would provide insights into the health of profiles with out-of-the-box metrics like profile count, profile over-merging or mapping limits and more. These dashboards and tools would enable users to proactively monitor profiles quality and proactively take corrective actions.
Probabilistic ID Resolution
Q: Are you going to build probabilistic identity resolution?
A: A common ask is to increase the number of identity resolved profiles. We know that flexible deterministic identity matching rules, and the ability to account for variations within profile data can help significantly increase the number of identity resolved profiles. Probabilistic identity resolution can help as well, particularly in cases where there is sparse, inconsistent or low-quality data from a variety of sources and when accuracy does not need to be 100%. For example, retargeting use cases but not email or SMS engagement where consent is required. We are considering a variety of new identity resolution capabilities including Boolean logic, nesting, composite keys, fuzzy matching, and probabilistic modeling. Please let your CSM know to file a feature request if these capabilities would benefit you.
We value your input and would love to hear your thoughts on these topics. What features or improvements would you like to see? How can we better meet your needs? Let’s keep the conversation going in the comments!