New
Pre-built Claude skills are here. Set up Claude and Kolvera end to end — then run your whole workflow just by chatting.
Get the skills →
All Updates
TL;DR
We found that the same metric meant different things on different screens, so we fixed the definitions rather than leave them disagreeing.
Open and click rates will go down. They used to count every open, so one prospect opening the same email four times counted as four opens — which is how an open rate could read over 100%. They now count opened emails.
Reply rates will go up. On your dashboard and in the assistant they used to divide replies by every email sent, so adding a follow-up step to a sequence halved your reply rate even though the same people replied. They now divide by the people you actually reached.
Your underlying data has not changed — no opens, clicks or replies were lost or added. The same events are being counted a better way, on the four screens listed below.
Where this applies
- The four surfaces that now share one set of definitions — the campaign page (including its A/B variant block and per-step table), the campaign list, your dashboard's campaign card, and the assistant. Each shows a different subset — the list card shows counts, not rates; the dashboard card shows an open and reply rate. Wherever two of them show the same figure for the same campaign, it is now the same number.
- The reply rate on the Replies card at the top of the campaign list is a different measurement, and stays that way — it is a 30-day figure across your whole desk rather than one campaign's, and it counts every reply (out-of-office included) against every email sent. That is what its own footnote already says it counts. It is not the campaign reply rate and will not match it — reconciling the two is a separate piece of work we have not done, and we would rather say so than quietly imply otherwise.
What changed
- Open rate and click rate now count emails, not events — Opened means the email was opened, however many times it was reopened. Expect these rates to drop, most noticeably on campaigns with engaged prospects who reread your emails. Nothing got worse; the old number was inflated.
- Reply rate is now replies per person reached — not per email sent. A 5-step sequence no longer reports a reply rate five times lower than a 1-step one for the same result. Expect this to rise on multi-step campaigns.
- Bounce rate on the campaign page now counts against the contacts you actually reached — not everyone enrolled. A part-sent campaign used to show a flattering rate. Expect this to rise slightly on campaigns still working through their queue. It is close to the figure that can auto-pause a campaign but is not always identical to it: that safety check deliberately counts every attempted send, because a campaign whose sends are all failing is exactly what it has to be able to see. The two differ only on campaigns with failed sends.
- Replies we already knew about are no longer hidden — a campaign could show fewer replies than we had on record when a reply arrived but the contact's status hadn't caught up. The campaign page, the campaign list and your dashboard now all show the higher count — which, for mail sent through Kolvera, is the complete one: a reply is recorded before the status that could hide it is set. The campaign page's own reply rate rises with that count, since it divides by the same contacts as before — a separate reason from the denominator change above, and the one that moves this particular number.
- The campaign list card counts the same mail as the campaign page — it used to count an email as opened even when the send itself had failed, and to miss replies whose status hadn't caught up. Its sent, opened and replied figures now match the campaign page for the same campaign.
- The per-step table now adds up to the campaign totals — and the permanent footnote explaining why it didn't is gone. It now appears only when the totals genuinely can't match, and says which of the two reasons applies: a reply we can't tie to the specific email that drew it, or mail sent before we recorded which step it belonged to.
- A/B variant stats use the same definitions — the variant block and the campaign card no longer disagree about the same mail.
Why
- A number should not depend on which screen you read it from — The campaign page counted “open rate” from open events, while the dashboard and the assistant counted opened emails over emails sent — a different number — and the campaign list counted opened mail its own way again, treating an email as opened even when the send had failed. Each was defensible on its own; they did not all agree, which made them untrustworthy together.
- A rate that can exceed 100% is not a rate — Dividing open events by emails sent could, and did, produce open rates above 100%. Counting opened emails cannot.
- Follow-ups should not punish your stats — Replies are people writing back. Dividing them by emails sent meant the act of adding a follow-up made your campaign look worse at the exact moment it was working harder.
- We would rather move your numbers once than keep several versions of them — This is a one-off correction, not a rolling change. The definitions are now written down in one place and those screens read from it.