agentleFS
manypicommanypicom/sales-skills/skills/outbound-metrics-benchmarks/SKILL.md

outbound-metrics-benchmarks

Skill1 starsChanged 2 months agoMITView on GitHub

Measure a cold email programme with the metrics that matter and diagnose which stage of the funnel is broken. Use whenever the user mentions open rates, reply rates, conversion rates, outbound benchmarks, cold email metrics, campaign performance, pipeline from outbound, or asks whether their numbers are good. Also use when outbound is underperforming and the cause needs isolating.

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Treat these as rough bands.

    line 30
  2. Distinctive:

    Don't measure at one week: when the cycle is months.

    line 135

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Outbound Metrics

Most outbound reporting measures the wrong things carefully. Open rate is unreliable, click rate is irrelevant for cold email, and "emails sent" is an activity number that rises when things are going badly.

Two metrics matter: positive reply rate and meetings booked. Everything else is diagnostic — useful for finding which stage is broken, not for judging the programme.

The funnel

Each stage isolates a different failure, which is the point of measuring them separately.

Leads sourced          →  Is the list the right people?
  Verified             →  Was the data any good?
  Delivered            →  Did it reach an inbox?
  Opened               →  Unreliable. Directional only
  Replied              →  Was it worth responding to?
  Positive reply       →  Was it relevant?
  Meeting booked       →  Was the ask right?
  Opportunity          →  Was it a real fit?
  Closed               →  Everything downstream

Benchmarks

Treat these as rough bands. They vary enormously by market, deal size and how narrow the segment is, and anyone quoting precise industry figures is guessing.

MetricPoorTypicalGood
Bounce rateover 3%1-2%under 1%
Delivery rateunder 95%97-98%over 99%
Open rateunder 25%40-60%over 60%
Reply rateunder 1%2-5%over 8%
Positive reply rateunder 0.5%1-3%over 5%
Meeting rateunder 0.3%0.5-2%over 3%
Unsubscribe rateover 2%under 1%under 0.5%
Complaint rateover 0.3%under 0.1%under 0.05%

The most useful benchmark is your own last campaign, not an industry figure. Segment, offer and market vary too much for external numbers to mean anything precise. Track your own trend and compare campaigns against each other.

Metrics to be sceptical of

Open rate is unreliable. Apple Mail Privacy Protection pre-fetches images, inflating opens for a large share of recipients. Image blocking suppresses them elsewhere. The absolute number is close to meaningless.

What it's still good for: detecting change. A drop from 50% to 15% over a few days, with no content change, is a deliverability event and one of the most useful early warnings you have. Use opens as a trend alarm, never as a performance measure.

Click rate doesn't apply. A good cold email has zero or one link. Optimising for clicks pushes you toward more links, which hurts deliverability and gives people something to do other than reply.

Emails sent is an activity metric. It goes up when things are going badly, because the response to poor results is usually more volume.

Positive reply rate, defined

The metric worth building reporting around, and it needs a definition everyone uses consistently.

Positive — interested, asking a genuine question, requesting information, referring you to the right person, or asking to reconnect at a specific later date.

Not positive — "not interested", "remove me", out-of-office, or a bounce.

Classify every reply. Both raw reply rate and positive reply rate are needed, because a high reply rate with a low positive share means the email provokes responses for the wrong reason — usually annoyance, sometimes confusion about what you do.

Diagnosing by stage

The value of the funnel is that each stage points somewhere specific.

SymptomStageDiagnosis
Bounce over 3%DataUnverified or stale list. See email-verification
Delivered, opens under 25%DeliverabilityAuthentication, reputation, content. See email-deliverability
Opens fine, replies under 1%Copy or targetingUsually targeting. See below
Replies but mostly negativeTargetingWrong people, or the offer doesn't match
Positive replies, no meetingsThe askToo big, or the handoff is slow. See meeting-booking-cta
Meetings but no opportunitiesQualificationThe ICP is wrong, or the email over-promised
Unsubscribes over 2%TargetingWrong audience
Complaints over 0.3%Stop nowTargeting or consent. See bounce-recovery

The most common misdiagnosis is treating a targeting problem as a copy problem. Good opens and near-zero replies almost always means the right people aren't on the list, and rewriting the email for a month won't fix it. The test: read twenty replies. If they say "wrong person" or "we don't do that", it's the list.

Sample size

Small numbers produce confident nonsense. A campaign of 100 emails with 3 replies has a 3% reply rate with an error bar wide enough to include 1% and 7%.

Rough minimums:

  • Bounce rate: 200 sends
  • Reply rate: 500 sends
  • Positive reply rate: 1,000 sends
  • A/B test on reply rate: several hundred per variant

Below these, look at the replies themselves rather than at the rate. Twenty replies read carefully tell you more than a percentage from 100 sends.

Attribution over time

Cold email produces delayed conversions. Someone who says "not now, revisit in Q3" and books in September belongs to the original campaign, not to whatever happened to touch them last.

  • Track first touch, not just last touch.
  • Measure over a full sales cycle. Judging a campaign at four weeks when the cycle is three months measures nothing.
  • Attribute "not now" conversions back to the campaign that created them. Otherwise the follow-up looks like the whole cause and the original campaign looks like a failure.

The weekly review

Twenty minutes per campaign, same time each week:

[ ] Bounce rate                over 2% → pause and investigate
[ ] Complaint rate             over 0.1% → pause immediately
[ ] Open rate trend            sudden drop = deliverability
[ ] Reply rate vs other campaigns
[ ] Positive reply share
[ ] Inbox health per mailbox
[ ] Replies unanswered         should be zero
[ ] Read 10 actual replies

That last line is the most valuable one on the list. Metrics tell you something changed; replies tell you what. See reply-handling.

Reporting honestly

  • Report positive replies and meetings, not sends and opens.
  • Show the funnel, so the broken stage is visible rather than hidden in an average.
  • Segment the report by campaign. A blended number across three segments hides that one works and two don't.
  • State sample sizes. A 6% reply rate from 80 sends is not a result.
  • Don't report open rate as a headline. Explain why once and stop showing it as performance.

Don't do these

  • Don't optimise for opens. Unreliable, and it drives subject-line theatre.
  • Don't compare against published industry benchmarks. Compare against your own history.
  • Don't judge a campaign before it has volume.
  • Don't count out-of-office as a reply.
  • Don't celebrate a reply rate without checking the positive share.
  • Don't measure at one week when the cycle is months.
  • Don't blend segments into one number.

Getting the numbers

The practical obstacle is that sourcing, verification, sending and replies usually live in different tools, so the funnel has to be reassembled from exports and the stages can't be compared cleanly.

ManyPI keeps the whole chain on the same records — leads with their source and verification status, enrollments, sends, and replies classified by sentiment and intent attached back to the enrollment — so the funnel is readable end to end and outreach_stats is available over the API and MCP:

claude mcp add --transport http manypi https://mcp.manypi.com/mcp -s user

Whatever the source of the numbers, read ten replies a week. No dashboard replaces that.

Related skills

outbound-campaign-ops, reply-handling, email-deliverability, email-verification, meeting-booking-cta, bounce-recovery, icp-definition

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.