Which SDR metrics should you track every week?

What are the five metrics and two guardrails?

The formulas are our recommendation, written so the denominator is always emails sent unless stated. Why that matters is covered in reply rate vs positive reply rate vs meeting rate.

Swipe to see more →

If your SDRs call as well as email, add dials and conversations. We haven't included a calling benchmark because we haven't verified one we'd trust; see cold email vs cold calling for how to compare the two channels.

What does a weekly sheet look like?

One row per week, for one rep. This example is invented for illustration. At our house setup of 4 inboxes × 30 sends a day, one rep sends about 600 emails in a five-day week (our math).

Swipe to see more →

Add a four-week rolling total at the bottom: sends 2,380, replies 74 (3.1%), positive 43, held 17. Decisions come from that line and not from any single week.

How noisy is one week?

Noisier than most dashboards suggest. Our math, using a standard 95% interval for a proportion (Wilson method) as a rough guide to week-to-week spread: if your true reply rate is 3.43% (the average reply rate on Instantly's platform in its 2026 benchmark, vendor data), one rep sending 600 emails will see about 21 replies on average, but a week anywhere from roughly 2.2% to 5.2% wouldn't be a surprise. Over 4 weeks and 2,520 sends the range narrows to about 2.8% to 4.2%.

Meetings are worse. At a 1.05% meeting rate (an illustrative figure, not a benchmark), 600 sends average about 6 meetings, and the same interval runs from about 0.5% to 2.2%, that is, from 3 to 13 meetings. So a week with 3 meetings doesn't mean the campaign broke.

Two rules follow:

1. Don't change copy, lists or targeting because of one week. Wait for four weeks or about 2,000 sends.
2. Do act at once on the guardrails. A bounce spike or a complaint is a mechanical problem, not noise. Google's sender guidance sets the line: keep spam rates below 0.10% and never reach 0.30%.

What should you do when a number moves?

Work down the funnel and fix the first stage that falls below your own baseline:

1. Sent is down: check inboxes, daily limits and the list. Our page on cold emails per day per inbox covers volume.
2. Reply rate is down: check deliverability first (inbox placement test), then list quality, then copy.
3. Positive share is down while replies hold: the audience or the offer changed. Re-read the replies.
4. Meetings booked are down while positives hold: the ask or the scheduling step is failing.
5. Held is down: no-shows. Confirm within minutes and remind them the day before.
6. Pipeline is flat while meetings hold: qualification. Revisit who you're meeting; see BANT vs MEDDIC. Once those meetings are open deals, grade them with commit, best case and pipeline.

For a quantitative test of a change, use the sample sizes in how to A/B test cold emails.

How do you set your baseline?

Use your own last 1,000 or more sends, not an industry number. Instantly's 2026 report gives 3.43% as an average reply rate, a 5.5% top quartile and 10.7% or more for its elite tier (vendor data), but as far as we could read it doesn't report positive replies or meeting rates. We found no public benchmark for those two that we could trace to a named sample. If you need to know how much pipeline the meetings must produce, see how much pipeline you need.

How do you run the weekly review?

Twenty minutes, Monday, one person:

1. Fill in last week's row from your sending tool and CRM.
2. Update the four-week rolling line.
3. Read ten replies, positive and negative, to see what's behind the numbers.
4. Write one sentence: "This week I will change X because Y," or "No change."
5. Check the guardrails first. They're the only numbers you act on the same day.

When should you skip each metric?

• Open rate: skip it. It isn't one of the five, and an open isn't a conversation.
• Weekly positive-reply share: skip judging it in a week with fewer than about 20 replies. A single reply moves it by 5 points.
• Weekly meeting rate: skip trend calls on one rep's data. Use the four-week line.
• Pipeline created in the first weeks: skip reading it before your first meetings have had time to become opportunities; it lags the others.
• Dials and call counts: skip them if nobody on the team calls. Track conversations only if you do.