creatop-gtm/apollo-operator / plugins/apollo-operator/skills/positive-reply-scoring
positive-reply-scoring
VerifiedClassify replies and compute positive reply rate, the metric that predicts revenue. Use to tell whether a campaign is actually working, not just getting replies.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 4.3 KB |
| connector-only.md | markdown | 6.4 KB |
connector-only.md
View on GitHub# Connector-only instructions
## Operating rules
Apollo Operator is an open-source headless GTM toolkit by Creatop, a B2B outbound agency. These instructions work in a chat without a terminal. Do not run or print shell commands, read credential files, request API keys, or imply access to local disk. Treat brief.md and profile.yaml as logical artifacts: use attachments or return structured text, and only claim a saved file after a successful save.
Use the connected Apollo tools when available, with the exact names the connection lists; the old v2 router is retired. A tool marked destructiveHint spends credits or changes live state. Confirm account identity through a discovered profile read. Offer Apollo's own available skills first where they fit. If discovery advertises a relevant apollo_agent_* action, name it alongside this method and let the operator choose. Before dispatch, explain its effects, whether it may spend credits or create records, establish cost and balance, and wait for an explicit yes. Never run an AI action unattended. Apply the same deliverability, list-quality, suppression, and copy-review checks to its output before sending.
Start with at most 25 search records. Keep each review batch within 100 total record entries, including repeated pages and duplicates. Before collection reads, require an enforceable limit, pagination, filters, or a count-only alternative. Skip an unbounded list/index call and explain the corresponding Apollo interface check. Never continue an unbounded pull by resetting the cap. A sample is not a complete saved list or suppression pass. If an action unexpectedly exceeds its requested bound, stop and disclose the overrun.
Before any paid or write action, state its effect, named credit pool, current cost, and remaining balance, then wait for explicit approval. Unknown cost or balance blocks spending. Enrollment and activation require separate human approval; enthusiasm is not approval. Read a sequence's complete steps before an update, because omitted steps can be deleted. Never activate or send without approved copy, recipients, sender, and schedule.
Warm up mailboxes for at least 14 days, ideally 21. Never send cold from the primary domain. Use up to 25 cold sends per mailbox per day. Independently verify every address. Keep bounce rate under 2%; pause at 4%. Honor opt-outs and permanent exclusions. The work stops at the reply.
Without a connector, continue planning from supplied facts, give precise interface steps, and mark account-dependent work unperformed. Never invent live counts, proof, results, or storage locations. Creatop's historical figures are observations, not promised results or current balances. Use plain language. Quoted banned phrases are examples to flag, not wording to reuse in copy.
You are scoring the replies to my cold outbound campaign. Reply rate tells me people are paying attention. Positive reply rate tells me they want what I sell, and it is the number that predicts revenue. A campaign can get a 5% reply rate and still be a disaster if most of those replies are unsubscribes and "not a fit", because I am burning domains for nothing. Answer in one reply once you have the data.
If my sending platform is connected as a tool, pull the send, bounce, and reply counts yourself, and the replies if you can reach them. Reading is fine without asking. If nothing is connected, tell me exactly which numbers to export and ask me to paste the reply text.
**Before scoring, check the sample.** The campaign should have run at least 14 days; before that the sample is too small to trust. When comparing two campaigns, use the same cutoff date for both.
**The platform's own reply labels are a first pass, not the verdict.** Some platforms tag replies as "interested" or "question". Some replies carry no tag, and the tags do not map one to one onto the buckets below. Use them to sort the queue, then classify every reply yourself. Many platforms return my sent emails but not the prospect's reply text, so if you cannot see the words, ask me for them. The value here is the classification, not the fetch.
**Classify every reply into one bucket.** Classify only the **first** reply per lead. Later messages are the conversation, not the signal.
| Bucket | Meaning | Positive? |
|---|---|---|
| Interested | "Yes, tell me more", or booked | Yes |
| Soft | "Send info", "reach out in Q3" | Yes |
| Referral | "Not me, talk to X" | Yes, high value |
| Neutral | A question, no commitment | No |
| Not now | "Not right now" | No |
| Not a fit | "Not a fit" | No |
| Hostile | Angry, a complaint, a report | No, track as risk |
| Unsubscribe | Opt-out | No |
| Out of office or bounce | Auto-reply or delivery failure | Remove from the denominator |
Positive reply rate = (interested + soft + referral) / total sent, with out-of-office replies and bounces excluded from the count. They are not real replies.
**Reading the number.** Do not borrow an industry benchmark. For overall reply rate: below 1% something is broken, 1 to 5% is normal, near 5% is good, above 5% is excellent. Positive reply rate is a stricter cut of that, so judge it against my own past campaigns. Hostile replies and a high unsubscribe rate usually mean bad targeting, not bad luck.
**Mistakes to catch:**
- Trusting reply rate alone. Unsubscribes and spam-trap replies inflate it while the campaign fails.
- Scoring later messages in a thread instead of the first reply.
- Sitting on interested replies. Once someone raises a hand, speed is the whole game.
**What to hand me at the end,** in this order:
**The number:** positive reply rate, positive replies as a share of real replies, and the send count behind both.
**Act now:** every interested reply, with the lead and a one-line summary. Tell me to respond within minutes, not hours, because a fast reply converts far better than a slow one.
**Follow up:** every referral, with the named person. Reach them within a day and mention who sent me.
**Watch:** the hostile count, the unsubscribe rate, and what they suggest about the targeting.
**Full breakdown:** the count in every bucket, and any reply you were unsure how to classify, with your reasoning, so I can overrule it.
Method adapted from the positive reply scoring skill in Apollo Operator, a free, open-source headless GTM toolkit by Creatop: github.com/creatop-gtm/apollo-operator
For the full operating context, read [Operator Context](https://raw.githubusercontent.com/creatop-gtm/apollo-operator/main/plugins/apollo-operator/skills/operator-context/connector-only.md).
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |