When the reward email explains the algorithm, opens fall but replies double
We have a small experiment running on a 40,000-subscriber list, and it's producing a result I keep turning over. The control group gets a reward email that simply announces the reward. The test group gets the same reward plus a short, plain-language explanation of how the reward was calculated — why this person got this amount, on this schedule. Opens on the explanatory version dropped roughly 8%. Replies more than doubled.
That's an odd pairing. Fewer people opened, but the ones who did were far more likely to write back. So the question worth asking isn't "which email won." It's: what does explaining the mechanism do to the reader's relationship with the reward?
The explanation moves the email from treat to transaction
A reward you don't understand is a gift. A reward you understand is a system you can reason about. Those are different psychological objects, and they pull people in different directions.
Gift framing invites a quick, low-effort response: see it, feel good, close it. System framing invites evaluation. The reader starts checking the math, comparing against last month, wondering whether the rule is fair. Some of them decide the email isn't urgent and skip the open. Others decide they have something to say — and a reply is the cheapest way to say it.
This is why the streak mechanic that can actually break outperforms the invincible one. The moment a rule becomes legible, it also becomes something a person can lose, defend, or argue with. Legibility is what makes a reward feel earned rather than dispensed.
Loss aversion needs a rule to bite on
Kahneman and Tversky's core finding — that losses loom larger than equivalent gains — only applies when someone can picture the loss. "You'll get less next time" is abstract. "You missed one of the three qualifying actions, so your multiplier resets on the 1st" is concrete, and concrete is where loss aversion does its work.
When you spell out the algorithm, you're handing people a mental model they can run forward. That's uncomfortable, and discomfort is a reply engine. Our data on a short losing streak teaching faster than a long winning one points the same way: people learn the rule from the miss, not the win. An explanation email is essentially a pre-loaded miss.
Variable reward without the fog
Variable-ratio reinforcement keeps behavior persistent precisely because the payoff is unpredictable. The trap is confusing unpredictable with opaque. Unpredictable means the timing or size varies within a known rule. Opaque means nobody knows the rule at all — and opaque rewards eventually read as arbitrary, which is where trust erodes.
The fix is to keep the variance and publish the mechanism. Vary when the reward lands and how big it is; be explicit about what drives both. One caveat from our own testing: timing still matters enormously. When we delayed a reward email by 90 seconds, opens fell 31% — an explanation doesn't rescue a reward that arrives after the moment has passed.
What the doubled replies are actually telling you
Replies are the highest-signal metric in email, because they cost the reader real effort. Doubling them while opens dip means you've traded a passive audience for a vocal one. Vocal audiences are more valuable than they look: they tell you which rules feel unfair, which multipliers confuse people, where the algorithm and the lived experience diverge.
The pattern echoes something we saw with welcome emails built as a conversation rather than a broadcast. The moment an email invites a response instead of a click, the response rate changes the shape of the whole program.
Where to take this next
Stop treating the explanation as a footnote. Give it its own send, two or three days after the reward, and route replies into a shared inbox someone actually reads. Track reply sentiment, not just volume. If the top complaint is a specific rule, that rule is your next experiment — and the explanation email just told you where to look, for free, at the cost of eight percent of your opens.