Orita helps you send the right message to the right person at the right time. We build causal models trained on an average of 250 million customer engagement touchpoints per brand. Brands get the answer to one specific question: will emailing or texting this person, today, actually move her toward engagement or a purchase. If not, messaging often adds risks like wasted spend, unsubscribes, and/or a negatively-impacted sender reputation.
Most customers use Orita Engagement Levels daily. But Orita Scores are a more precise tool that the most advanced brands and agencies use to engage their customers.
What are Orita Scores?
Every profile that has received an email gets a 1-100 Orita Score, in every channel (email and SMS score separately, as they're different models). The score is synced into Klaviyo as a plain profile property. And it's updated daily.
The five Orita Engagement Levels - Highly, Moderately, Slightly, Rarely, Not Engaged - are just named bands of the score. Mutually exclusive, but continuous underneath. E.g. for one brand, Highly Engaged might be a score of 92-100, while for another it's 88-100.
These update daily, so different people will be in different scores - and different Engagement Levels - every day.
Tapping into Scores allows for Meta-like testing of the frequency-distribution of your sends
In the world of ads, a shopper who adds to cart and bounces might get shown a retargeting ad a dozen times over the next week. A visitor who glanced at the homepage and left gets capped at zero or one impression, because she's not worth the spend. This is the platform (e.g. Meta, Google, TikTok, ConnectedTV providers, AdRoll) correctly reading who's still in-market.
For Orita Email and SMS customers, inside a single Engagement Level, performance can differ by 3x or more between the top of the tier and the bottom. If you're sending to the whole tier as one block, you're averaging those two very different audiences together. That's good for efficiency - and much better than relying on 30/60/90 or RFM - but can still leave money on the table.
On a real account (see the table below, cleaned up from Klaviyo for ease of reading), click rate drops from 0.75% for Highly Engaged to 0.12% for the top of Moderately Engaged, then to 0.02% a few bands lower, while total recipients in each of those lower bands stays just as large.
Send widely, and you're sending to hundreds of thousands of people for a rounding error of revenue.
Segment | Score range | Recipients | Click rate | Revenue |
|---|---|---|---|---|
Highly Engaged | Always-include | 362,694 | 0.75% | $34,021 |
Moderately (top) | 76–82 | 125,483 | 0.12% | $1,294 |
Moderately | 66–75 | 183,068 | 0.06% | $817 |
Mod/Slightly | 55–64 | 184,217 | 0.03% | $375 |
Slightly | 45–54 | 186,544 | 0.02% | $81 |
Slightly/Rarely | 40–44 | 94,148 | 0.02% | $0 |
Rarely | 35–39 | 91,581 | 0.02% | $76 |
Rarely | 30–34 | 91,296 | 0.02% | $0 |
Rarely | 25–29 | 94,199 | 0.01% | $48 |
Not Engaged | 1–24 | 435,560 | 0.01% | $107 |
The score is what lets you see that curve and stop before you hit it, instead of learning it the expensive way.
How do I use it?
Basics
For day-to-day sends, Engagement Levels are enough: five buckets, no extra setup. The score matters when you're asking one of two harder questions:
Depth: how far into a tier should I send today?
Frequency: how often can your best people hear from you before it stops paying off?
How to use Orita Scores to explore the Depth and Frequency with which your customers want to hear from you
Depth: walk down the scoring ladder
Say you're sending to Highly Engaged and want more reach. Don't flip a switch and open all of Moderately Engaged at once.
Test multiple score ranges at a time, and send to them altogether over the course of a week or more. Email performance fluctuates so much - day of the week, type of email, content, offer - that you'll never get to the exact profile cut off, but testing can meaningfully increase precision.
Then inspect your scores and figure out where diminishing returns are happening, i.e. "I'll maximize my program - revenue and efficiency over time - by regularly sending to this score and above".
Of course, don't never email below that threshold; but now you'll have much better data to understand the possible tradeoff of click rate versus revenue.
A few ways to slice a tier instead of sending the whole thing:
Top slice by score. "Score 71–81" gets the strongest performers in Moderately Engaged without the weaker half.
Build track segments. Using Score ranges to send to mutually exclusive segments: "Score 96–100", "Score 91–95", "Score 86–90"
Purchasers vs. prospects, same score. Two profiles can share a score and behave very differently depending on purchase history.
SMS bands. Since texts cost per send, break a tier into smaller bands and send the top few to see where performance actually drops.
The same logic works in reverse for exclusions: break Not Engaged into bands before suppressing the whole tier, rather than assuming it's all dead weight.
Two things to watch for: small score bands are noisy (don't trust one week's result on a 200-person slice), and depth tests read differently during a promotion than a normal week! Test in a normal week where you can.
Frequency: freeze a slice, test different cadences, find where it bends.
Freeze a slice of Highly and Moderately Engaged for a defined window.
Split it into groups on different frequencies (three sends a week vs. five).
Track click rate, revenue per recipient, and on-site visits per group.
Find where sending more stops producing proportionally more: that's your real ceiling.
One rule of thumb that holds up often: one or two extra sends to High Score ranges or even just Highly Engaged usually beats one more send deeper into the list, with less risk. And test depth and frequency separately, as changing both in the same week means you can't tell which one moved the number.
Examples from brands
Here's a real depth curve from one brand's Highly Engaged/Moderately Engaged split, going deeper using Engagement Scores.
The cliff shows up almost immediately: click rate is already down 6x by the top of Moderately Engaged, and revenue is basically gone by the 66–75 band - while every one of those lower bands still has 90,000–435,000 recipients in it. A walk-down test on this list would have stopped at the top of Moderately Engaged; everything below it is volume without return.
FAQ
Q: Does "timing" mean time of day?
A: No - Klaviyo's send-time tools handle the hour. The score answers whether she's listening today at all, not what hour today.
Q: Does the score know what the email is about?
A: No, on purpose. It answers "should she hear from you today," not "does she care about this product." Build your content segment first, then layer the score on top as a filter.
Q: Why did a profile show as "NA" on the campaign report?
A: Either the email bounced at send time, or she's new enough that scoring hasn't finished: a first score takes about 24 hours.
Q: Our open rate went down but click rate went up. Is something broken?
A: No. We weight opens less than clicks and purchases, since Apple Mail inflates opens that were never really opened. Judge the model by click rate, revenue per recipient, and deliverability.
Q: Does a score of 70 mean the same thing at every brand?
A: No. The model is bespoke to your account, so compare your own scores over time, not your raw number against someone else's.
Q: We have a VIP list that should always get emailed. Won't scoring them cause a conflict?
A: No. If you'd like, keep them as always-include and let the score run as a background read. A low email score with a high SMS score just tells you which channel she'd rather hear from.
Q: If we suppress someone, does Orita stop watching them?
A: No. The model keeps tracking her behavior in the background, and she becomes a reactivation candidate the moment her signals cross the bar.
