Know which replies are worth your time
Every response is read and sorted by what the person actually means, so your team answers the real opportunities first.
Every reply is read and sorted the moment it lands.
No rules to build, no keywords to configure. AI reads what the person actually means and labels the reply accordingly.
A reply arrives, AI reads it
The full message is analysed for intent, not scanned for trigger words. Tone, context, and what the person is asking for all count.
It gets one of six labels
Interested, objection, referral, out of office, wrong person, or unsubscribe. Applied automatically, editable any time.
The label triggers the right action
Unsubscribes are suppressed everywhere. Referrals are flagged for handoff. Interested replies get pushed to the top of the queue.
Five things a raw reply count never tells you
Reply rate counts auto-responders and rejections alongside genuine interest. Classification is what turns it into a real signal.
Reply Rate Alone Is a Vanity Metric
Out of office messages, wrong-person replies, and unsubscribes all count as replies. So does someone asking to book a call.
Interested People Get Answered First
At volume, working an unsorted inbox top to bottom means hot leads wait behind out of office messages.
A Soft No Is Not a Real No
Plenty of replies contain a rejection and an opening in the same sentence. A keyword filter throws both away.
Referrals Never Get Filed as Rejections
Being pointed to the right person is one of the best outcomes in outbound, and it is easy to misread as a dead end.
Unsubscribes Handled Instantly
An opt-out is a legal obligation, not just a label. Classification connects the two automatically.
Positive reply rate is the number that predicts pipeline.
Because every reply is labelled as it arrives, positive reply rate is a real metric on your dashboard rather than something you estimate at the end of the month.
Reply categories
Rules to configure
Labelled on arrival
Editable by your team
Two things classification gets right that filters get wrong
Keyword rules are brittle. Reading intent is what makes the labels trustworthy at volume.
A "no" that is really a "maybe" gets labelled as an opportunity.
"We already use something for this, but send me details" contains a rejection and an invitation. Classification reads it as an objection with interest rather than binning it.
Referrals are treated as warm handoffs, not dead ends.
"I'm not the right person, talk to Dana" is one of the most valuable replies you can get. It gets labelled as a referral so nobody files it under rejected.
Reply rate tells you people wrote back. Classification tells you which ones matter.
Real opportunities first
Nothing to configure
Objections spotted as openings
Metrics you can trust
Everything you need to know
Common questions about Reply Classification in EVOX.
What is Reply Classification?
Every response to your campaigns is automatically read and given a label based on what the person means: interested, objection, referral, out of office, wrong person, or unsubscribe.
Is it based on keywords?
Do I need to set up any rules?
Can I change a label if the AI gets it wrong?
What happens automatically based on the label?
How does this change my reporting?
Know which replies are worth answering first.
Evox reads every response for meaning, sorts it into one of six categories, and acts on the label so nothing important waits behind an auto-responder.