With Gmail alone, you can automate most of the sorting of cold email replies and none of the judgment. Set up a label for each reply type, let a Gmail filter catch the mechanical ones (out-of-office, bounces, unsubscribes), stop your follow-ups for anyone who answered, and read the rest yourself once or twice a day. The six types below cover nearly everything that comes back.
Replies are where outbound pays or fails. A campaign can send 300 emails and produce 12 replies, and what you do with those twelve in the next hour decides whether the 300 mattered. The setup below keeps the mechanical replies from burying the ones a person has to answer.
The six types of reply
| Type | What it looks like | Automate? |
|---|---|---|
| Interested | "Yes, send details" or "Can we talk Thursday?" | Notify you at once. Reply by hand or with a template you approved. |
| Question | "What does this cost?" "Do you work in Texas?" | Notify you. A saved answer helps. |
| Not now | "Reach out in January." | Log the date. Schedule one follow-up. |
| Not interested | "No thanks." "Remove me." | Stop all emails to them and suppress the address. |
| Wrong person | "I do not handle this, try Dana." | Reply by hand to thank them. Add Dana. |
| Automatic | Out-of-office, bounces, delivery notices | Fully automate. |
If you only automate one row, make it the last. Out-of-office and delivery notices are noise, and they are the reason a real "yes" sits unread under forty of them.
Step 1: Make a label per type
In Gmail, create labels such as Cold/Interested, Cold/Question, Cold/Not now, Cold/Done and Cold/Auto. Nested labels keep the sidebar tidy. Do the labeling by hand at first, so you learn which phrases show up in your own replies before you write any rule.
Step 2: Write filters for the automatic ones
Google's help page for filters describes the process. Open the search options in the search box, enter your criteria and click Search to check which emails match, then choose Create filter and pick what it should do. The actions include applying a label, archiving, starring, deleting and forwarding. You can also start from a real message by choosing "Filter messages like these." Google adds that a filter only affects new messages, so it will not clean up your old inbox.
Three filters worth building:
- Out-of-office. Match subjects such as "automatic reply" or "out of office," apply Cold/Auto and archive them. Read cold email auto-replies and out-of-office first, because an out-of-office often gives a return date and a colleague's name, both of which you want to keep.
- Delivery failures. Match mail from "mailer-daemon" or subjects containing "delivery status notification," apply Cold/Auto and mark the address as bad in your sheet. What the codes mean is in email bounced back.
- Unsubscribe words. Match "unsubscribe," "remove me" and "stop emailing" in the body, apply Cold/Done, and star it so you see it. A filter can catch these but a person should confirm each one and add the address to the do-not-contact list.
Do not build a filter that archives anything containing the word "interested" or "yes." A reply that says "not interested" contains both. The risk of a missed buyer is worse than the effort of reading them.
Step 3: Stop the follow-ups the moment someone answers
This is the single most important automation. If your sequence emails a person who already replied "send me details," you look careless. Make sure your sending tool stops the sequence for a contact on any reply, even an out-of-office, then restarts it deliberately. The details are in how to stop follow-ups when someone replies.
Step 4: Set a reading schedule
Two slots a day, say 9:30 and 3:30. In each, open Cold/Interested and Cold/Question first and answer within the slot. Anyone who asks for a meeting gets time offers in the reply, not a question about their availability. Then move the replies to Not now or Done and clear the inbox. Fast replies matter because the interest is fresh. What a good answer looks like is in what to do when someone replies to your cold email.
An example morning
Say an outdoor-lighting company in Raleigh sends 120 emails to property managers on Monday. By Tuesday at 9:30 it has 19 messages:
- 7 out-of-office or delivery notices, already filtered into Cold/Auto.
- 3 not interested, one of them a "please remove me," confirmed and suppressed.
- 2 not now, with dates logged for January.
- 1 wrong person, with a new name to add.
- 4 questions, mostly about price and service area.
- 2 interested, one asking for Thursday.
The owner spends about fifteen minutes on twelve of those, mostly the six that need a human, and offers Thursday and Friday times to the two interested replies. Without the filters it would be 19 emails in a single pile, and the two warm ones would be mixed in with the rest.
What to measure weekly
- Replies per 100 emails sent.
- Share of replies that are interested or questions.
- Hours from reply to your answer. Aim for under four during work hours.
- Meetings booked per interested reply.
Where this breaks down is volume. Past a few hundred emails a week, filters catch the easy cases but someone still has to read, judge and answer. That is the job Rocketship's AI salesperson does: it reads each reply, answers it, stops the follow-ups and books the meeting on your calendar, and you can set it to show you drafts first. The safety answer explains the controls, and pricing lists the sending limits.