← all playbooks

Sort campaign replies and put the ones that need you today at the top

Every reply to the campaign sorted into interested, not now, referral, wrong person, unsubscribe or out of office, with the ones that need a person today at the top and a suggested answer drafted for each. The agent never replies for you.

use when
a campaign is running and replies are arriving faster than anyone reads them
starts from
A list

The prompt

Paste it into Claude Code or the Claude desktop app with Hyreflow connected. The first line tells your agent to use Hyreflow, so it reads the play, asks before it spends anything, and hands the work back to you.

paste this into Claude
Use hyreflow and sort the replies to my <CAMPAIGN_NAME> campaign so I can
see which ones need me today.

The campaign runs in <SEQUENCER>. If you cannot read replies from it, I
will give you an export: <FILE>. It is a <CANDIDATE_OR_BD> campaign
offering <WHAT THE MESSAGE OFFERED>.

Read everything since <DATE> and put each reply in one bucket: interested,
not now, referral to someone else, wrong person, unsubscribe, out of
office. Give the reason in a few words.
- Put the ones that need a person today at the top, oldest first.
- Draft a suggested answer for those, in my voice:
  <TONE, OR PASTE A REPLY I SENT BEFORE>. For a referral, name who I was
  pointed to.
- For out of office, give me the return date where the message has one.
- List the unsubscribes separately and ask me before you remove anyone.

Run five first and tell me what the full run costs before you do the rest.
Stage it, do not send anything: the drafts are for me to send myself. If
you log outcomes to <CRM_NAME>, show me the changes before you write
anything to the CRM.

Replace every <PLACEHOLDER> with your own detail. Everything else can stay as written.

What you need first

  • A running campaign in lemlist or HeyReach, connected with your own account, or an export of the replies from any other sequencer
  • One line on what the campaign offered, so an interested reply can be told from a polite one
  • A Hyreflow workspace with credits, since sorting and drafting are metered

Tools it can reach for

The agent picks per step from what your workspace has. Nothing here is required by name.

What happens when you run it

Free steps are marked free. Anything that spends credits is marked, and the agent asks before the first paid run of any size.

  1. 1

    Pull the replies from your sequencer

    free

    Reply events are read from the lemlist campaign, and LinkedIn conversations from the HeyReach inbox, for the dates you gave. It is a read of your own account, so it is free, and an export of replies from any other sequencer does the same job.

  2. 2

    Match each reply to the person and the message

    free

    Each reply is joined to the lead it came from and to the campaign's own messages, so the sorting sees what was asked as well as what was said.

  3. 3

    Pilot five replies and price the run

    credits

    Five replies are sorted and answered end to end so you can check the buckets and the tone of the drafts. The agent then states what the rest will cost and waits for your go-ahead.

  4. 4

    Sort every reply into a bucket

    credits

    Each reply lands in exactly one of six buckets, with the reason in a few words: interested, not now, referral to someone else, wrong person, unsubscribe, out of office. Anything the agent cannot place with confidence goes to a seventh pile marked for a person to read.

  5. 5

    Rank what needs a person today

    free

    Interested replies and referrals go to the top, oldest first, because a warm reply cools by the hour. Not now carries the date the person named, and out of office carries the return date where the message gives one.

  6. 6

    Draft a suggested answer

    credits

    For the replies at the top, a short answer is drafted in your voice that responds to what the person wrote. It is a draft for you to edit and send from your own inbox.

  7. 7

    Hand over the queue, and remove unsubscribes on approval

    free

    You get the sorted queue as a file. With your approval an unsubscribe is taken out of the lemlist campaign, and nothing else in the sequencer is touched: the agent does not reply, pause or re-flag anyone.

  8. 8

    Log the outcome where your CRM takes notes

    free

    Where your CRM connection can write a note, the bucket and the reply are logged against the person after you approve the list, one record first. Where it cannot, the outcomes stay in the file.

What a run costs

Nothing in this play spends credits.

Free before anything is charged

  • Pull the replies from your sequencer
  • Match each reply to the person and the message
  • Rank what needs a person today
  • Hand over the queue, and remove unsubscribes on approval
  • Log the outcome where your CRM takes notes

What moves the number

  • How many records survive the free filters. Everything dropped before the paid steps costs nothing.
  • The scoring and drafting steps run on the metered agent, charged on what they read and write rather than per record, so they sit outside this table.
  • Providers you connect with your own key. Those calls bill your account, not your credits.

An estimate, not a quote, priced at the volume credit rate. Your agent sizes the run against your own workspace and tells you what it will cost before it spends anything.

The send was the easy part

Building a campaign takes an afternoon. Living with it takes every morning after that. Once a sequence is running, the replies arrive mixed together: three people who want a call, a dozen out of office notes, a few who say try again in the spring, one who names the colleague you should have written to, and two who want to be left alone. The problem is never how many were sent. It is reading that many live replies without dropping the one that mattered.

A dropped reply is the most expensive failure in outreach, because everything before it was already paid for. The person answered, and nobody came back to them.

What you get back

A queue, as a file, with one row per reply: who it is from, which campaign it belongs to, the bucket, the reason in a few words, and the date that matters: when they wrote, when they are back, or when to try again. The rows that need a person today sit at the top with a suggested answer beside each. Unsubscribes are listed on their own so they can be honoured in one pass. Anything the agent could not place is marked for you to read.

Variations worth knowing

Any sequencer, from an export. If your campaign runs somewhere the agent cannot read replies, export them or forward them into a file. Everything after the first step is the same.

Follow the referral. When a reply points you to a named colleague, that person can be looked up and added to the list. It is a paid lookup, so the agent asks first: work email for business development, personal email and LinkedIn for a candidate.

Count interest, not replies. A fast no is not engagement. The share of replies that land in interested tells you whether the message works. If it is low, the fix is in how the sequence was built, not in sending more.

Where this goes wrong

Reply events are not the whole thread. The lemlist connection reads reply events for a campaign, not the inbox behind it. A long back-and-forth is best opened in lemlist itself, and where a reply's text does not come through, the row says so and is left for a person.

Polite is not interested. "Thanks, I will keep you in mind" reads warm and means no. The buckets come with a reason so you can spot a pattern you disagree with and correct it once.

An unsubscribe sorted is not an unsubscribe honoured. Removal from one campaign does not reach your other campaigns, other tools or your CRM. Suppression across those is yours to carry through.

Sending the draft unread. The draft answers the words on the page. It does not know what you agreed with that person on the phone.

Replies that never reached the sequencer. A call, a text or an answer sent to a different mailbox is invisible here.

Questions

Which sequencers can it read replies from?

lemlist and HeyReach. The lemlist connection reads reply events for a campaign, and the HeyReach connection reads LinkedIn inbox conversations. The connections to Instantly, Smartlead, Reply.io and SourceWhale can load leads and report campaign figures, but they do not read replies. For those, export the replies or paste them in, and the sorting, ranking and drafting work the same way on the file.

Will it ever reply for me?

No. Nothing in this play sends a message. The drafts come back as text for you to edit and send from your own inbox or your LinkedIn account. The only change this play makes in the sequencer is removing an unsubscribed lead from a lemlist campaign, and the agent asks before it does that.

What counts as needing a person today?

An interested reply, a referral to a named colleague, and a question that blocks the conversation. Those cool quickly, so they go to the top, oldest first. Not now and out of office are dated and parked. Wrong person and unsubscribe need an action from you but not a conversation.

What does it cost?

Reading your sequencer, matching replies to leads, ranking and handing over the file are free. Sorting and drafting are done by the Hyreflow agent and metered by usage, so the cost follows the number of replies and how long they are. The agent gives you a figure after the five-reply pilot and before the full run.

Can it run every morning?

Yes, once you have run it by hand and agree with how it sorts. It converts into a scheduled workflow that reads the replies since the last run and leaves the sorted queue and the drafts waiting in your workspace. A scheduled run never sends anything, and removing unsubscribes stays a step you approve. Prove it by hand first, because a bucket you disagree with is cheap to correct on fifty replies and tedious on a month of them.