Maya has a segmentation drawback she will’t fairly identify.
She is aware of her prospects, or not less than, she is aware of what they appeared like three weeks in the past when the final information export landed. This contains:
- Complete spend
- Final buy date
- Metropolis
- Plan sort
The same old. And for broad campaigns, this information works too. However each time she wants one thing particular, comparable to “prospects whose buy frequency has dropped within the final two weeks” or “customers who’ve spent over $200 however haven’t opened an e mail in 30 days”, she must file a request with the information group.
The common turnaround is 10 days. By the point the section is prepared, the second has handed.
Maya isn’t the exception, however she’s the norm. And the issue isn’t that her group is sluggish, it’s that her platform was constructed to retailer buyer information, not compute it.
What Static Attributes Can’t Inform You
Most advertising and marketing platforms retailer buyer attributes as snapshots. A price will get recorded, it sits in a profile, and it stays there till one thing manually updates it.
That works effective for steady information comparable to a buyer’s nation, their account tier, their signup date, and extra. But it surely breaks down the second you’ll want to seize how a buyer’s habits is trending.
Take two prospects with equivalent static profiles who each bought twice within the final 12 months, with $150 in whole spend. On paper, they give the impression of being the identical. However one purchased twice within the final 30 days and is clearly warming up. The opposite purchased as soon as in January and as soon as in November and hasn’t been energetic since. Treating them the identical, like the identical section, message, or supply, is the form of mistake static information makes look rational.
The hole between what static attributes document and what behavioral patterns reveal is the place most personalization falls brief.
What Computed Traits Really Are
Slightly than explaining the idea after which illustrating it, right here’s the illustration first.
Maya desires to establish prospects whose buy momentum is constructing. She creates a trait known as Buy Velocity (the variety of purchases within the final 30 days). Each buyer now has a rating, so excessive scorers get a loyalty reward and low scorers (who have been beforehand excessive) get a win-back supply. The section updates mechanically as habits adjustments. No CSV. No information group. No 10-day wait.
That’s a computed trait.
Extra formally: a computed trait is a dynamic person attribute that summarizes a buyer’s habits over an outlined time window right into a single, actionable worth. Not like a saved attribute, which data a truth. A computed trait calculates a behavioral abstract and retains it updated as new occasions arrive.
Retail examples:
- Complete Spend Final 30 Days
- Variety of App Periods This Week
- Favourite Product Class (primarily based on browse frequency)
- Cart Abandonment Rely Final 14 Days
BFSI examples:
- Login Frequency Final 30 Days
- Variety of Transactions This Month
- Characteristic Adoption Rating (primarily based on the variety of product options used)
- Days Since Final App Open
Every of those is a metric a marketer truly wants, and none of them exist as a uncooked attribute within the database. They must be computed from occasion information.
How Computed Traits Work
Two methods to construct a computed trait: one requires no technical data, one offers information groups full flexibility.
Each sorts replace on a scheduled foundation, i.e., weekly or month-to-month, conserving traits present with out requiring handbook intervention.
How Computed Traits Feed into Segmentation and Personalization
As soon as constructed, a computed trait seems within the segmentation interface precisely like several common person attribute. That’s the sensible level as a result of entrepreneurs don’t have to understand how a trait was computed to make use of it. It simply exhibits up within the dropdown.
What adjustments is the precision of what’s doable.
As a substitute of “customers who made a purchase order,” Maya can now have interaction “customers whose Buy Velocity rating is within the high 20% and has elevated week-over-week.”
As a substitute of “inactive customers,” she will have interaction “customers whose Login Frequency has dropped beneath 2 within the final 14 days after averaging 8 the month earlier than.”
Computerized section strikes imply prospects shift between segments as their habits adjustments with out anybody manually updating a listing. A buyer whose Buy Velocity falls beneath a threshold mechanically strikes into an At-Threat section. The re-engagement marketing campaign fires with out Maya having to the touch something.
Dynamic message personalization takes this additional. Traits may be pulled instantly into message content material, comparable to this:
- “You’ve made 4 purchases this month: right here’s an early entry reward.”
- “It’s been 18 days since your final login. Right here’s what’s new.”
- “Your financial savings this quarter: $340. Maintain going.”
The message displays what the client truly did, not a generic assumption about what somebody of their demographic is more likely to do.
The Activation Hole: Why a Separate Software Breaks This
Right here’s the place most groups hit a wall they don’t see coming.
Some platforms compute traits however retailer them individually from the engagement engine. The computation occurs in an information warehouse or a standalone analytics layer. The trait worth then syncs to the advertising and marketing platform on a schedule comparable to hourly, day by day, or manually triggered.
That sync delay is the issue. And it’s not a small one.
When Maya’s “Buy Velocity” trait updates in a separate instrument after which waits 6 hours to sync together with her engagement platform, the section she acts on is 6 hours behind actuality. Principally, her buyer crossed the re-engagement threshold at 10 am, and will get their message at 4 pm in the event that they get it in any respect. And her buyer, who purchased twice for the reason that final sync, remains to be sitting within the win-back marketing campaign.
The rating was proper. The timing was improper. And timing is the complete level of a behavioral rating.
When computation and activation stay on the identical platform, the sync delay disappears. The second a buyer’s habits pushes their trait previous a threshold, the section updates and the marketing campaign fires. No intermediate step. No latency baked in.
That is what “native” means in observe – not simply that the platform has the function, however that the function and the motion it drives are a part of the identical platform.
How MoEngage Handles Computed Traits Natively
In MoEngage, computed traits are constructed and activated inside the identical platform. Principally, no separate computation layer, no sync dependency.
Commonplace logic traits can be found via a point-and-click interface. Entrepreneurs select an occasion, choose a calculation sort, set a time window, and the trait is able to use in segmentation. No SQL, no engineering request, no ready.
For extra complicated scoring, SQL Compute lets information groups write customized queries that produce trait values obtainable on to entrepreneurs (the identical method a normal logic trait would seem of their interface).
As soon as created, traits can be found throughout the platform:
- In segmentation: as a filter criterion alongside another attribute
- In journeys: as a set off or branching situation
- In message content material: as a personalization variable pulled instantly into copy
The replace frequency may be configured to weekly or month-to-month, relying on the use case. Day by day updates are in growth.
Conclusion – Maya: Six Weeks Later
Maya constructed her Buy Velocity trait on a Tuesday afternoon. No ticket. No ready.
Her re-engagement marketing campaign now targets prospects whose velocity has dropped within the final two weeks, fairly than a broad “inactive” section pulled from a three-week-old export. Her loyalty marketing campaign reaches prospects whose spending is trending up, not everybody who’s ever spent over $100.
Open charges are up. The information group hasn’t heard from her in weeks.
The information didn’t change. What modified was how rapidly she might act on it.
In case you’re prepared to maneuver past static segmentation, MoEngage’s Computed Traits give your group the behavioral scores they want with out the engineering overhead. See it in motion -> speak to our agent.
