SDR call metrics managers should track every week

SDR call metrics worth a weekly review are dials, connects, conversations, meetings booked and talk time, plus the 3 rates that link them: connect rate, conversation rate and conversation-to-meeting rate. Add calls per number per day to catch spam-label risk early. Compare each rep with their own 4-week average, and fix the first rate in the chain that dropped.

The metrics and how to calculate them

Count everything per rep and per list for the week, then add up the team. Each rate is one count divided by another, so keep the counts next to the rates: a rate built on a small count moves a lot by chance.

Metric	Formula	What counts
Dials	Outbound calls placed	Every attempt: no answer, busy, voicemail, phone menu, call screener, wrong number
Connects	Calls answered by a person	Any live person, including gatekeepers and wrong people
Connect rate	Connects ÷ dials × 100
Conversations	Connects with the right person that meet your written rule	See the definitions below
Conversation rate	Conversations ÷ connects × 100
Meetings booked	Meetings set from calls this week	Counted on the day they are booked
Conversation-to-meeting rate	Meetings booked ÷ conversations × 100
Talk time	Total minutes connected to a person	Also: average per conversation = conversation minutes ÷ conversations
Calls per number per day	Dials from one caller ID number on one day	Review the highest value, not the average

To show the math only, not as a target: 500 dials, 40 connects, 25 conversations and 4 meetings booked give a connect rate of 8%, a conversation rate of 62.5% and a conversation-to-meeting rate of 16%.

Agree on 3 definitions before the first review, and keep them for the quarter. Changing a definition mid-quarter breaks every trend line.

• Connect. A person answered. Voicemail greetings, phone menus and call screeners are not connects. Screening is now built into phones: on iPhone, the “Ask Reason for Calling” setting asks a caller from an unsaved number why they are calling before the phone rings (Apple Support, as of September 2026). Log screener answers as their own outcome, so a rise shows up instead of hiding inside connect rate.
• Conversation. Pick one rule and write it down, for example “the right person, and the rep got past the opener”, or a minimum length your team agrees on. The Bridge Group’s research reports a similar count, quality conversations, but its published summary does not define one, so your own rule matters more than the label.
• Meeting. Count meetings booked on the day they are set, and meetings held on the day they happen. The Bridge Group’s 2025 SDR report gives a global median monthly quota of 10 Stage 0 meetings held, so if your quota counts held meetings, also track show rate: meetings held ÷ meetings booked × 100.

One more number helps when you compare dialing modes or call blocks: conversations per calling hour, which is conversations ÷ hours spent in call sessions.

The weekly review table: what a drop means and what to fix

Read the chain from left to right. A drop early in the chain shrinks every number after it, so fix the first rate that fell.

Metric	What it tells you	When it drops, check	What to fix
Dials	How much of the week went to calling	Meetings and admin inside call blocks; lists running out; contacts outside their calling hours	Fixed call blocks on the calendar; next week’s lists loaded by Friday; blocks placed inside each region’s calling hours
Connect rate	List quality, and whether people pick up your number	Wrong or switchboard numbers; calls at the wrong local time; one caller ID with a falling answer rate; more call screener outcomes	Remove bad numbers; call in the contact’s time zone; slow down or replace numbers whose answer rate fell
Conversation rate	Whether reps reach the right person and keep them past the opener	Gatekeepers and wrong people; calls that end during the opener	Better contact data (direct numbers, right titles); listen to the first 30 s of 5 short calls and rewrite the opener
Conversation-to-meeting rate	Whether conversations end with a clear next step	Reps who never ask for the meeting; one objection that keeps coming up; the wrong segment	Read transcripts for the ask; role-play the top objection; compare the rate by list
Meetings booked	The output of everything above	Which earlier rate moved first	Fix that rate; check show rate too
Talk time	How much time goes to live conversations, and how long each one lasts	Shorter average: calls end early. Longer total with flat meetings: long calls with no ask	Review the opener and targeting; review the 3 longest calls that ended without a meeting
Calls per number per day	Spam-label risk on each caller ID	Any number above your cap; one number’s answer rate falling while the others hold	Add numbers to match your volume instead of raising the cap; slow down or replace that number
Abandoned call rate (parallel only)	Whether answered calls reach a rep in time	More lines than free reps	Fewer lines per rep, or more reps in the session

Calls per number per day: the spam-label metric

Phone networks and caller-ID apps score numbers by how they call. In its June 2019 ruling that allowed phone companies to block likely unwanted calls by default using reasonable call analytics (FCC 19-51, paragraph 35), the FCC listed example signals. They include large bursts of calls in a short timeframe, low average call duration, low call completion ratios and a large volume of complaints. The FCC chose not to set rigid blocking rules, so the ruling gives no safe number of calls per day.

That makes 3 per-number checks part of the weekly review:

1. Calls per number per day. Dials from each caller ID number on each day. Look at the busiest number on the busiest day.
2. Answer rate per number. Connects ÷ dials for each number. When one number falls while the team’s other numbers hold steady, call a few phones on different networks and see what the screen shows.
3. Short and unanswered calls per number. Low call duration and low completion ratios are on the FCC’s list, so a number used for many quick hang-ups carries more risk.

Set a written daily cap per number. When the team’s volume grows, add numbers instead of raising the cap. To check and fix a label, see why calls show Spam Likely.

If your team dials in parallel, add abandoned call rate

A parallel or predictive dialer can connect a person when no rep is free. That call counts as abandoned, and it is the one metric on this list that calling rules address directly. As of October 2026:

• United States. Under the FCC’s rule (47 CFR 64.1200(a)(7)), a telemarketing call is abandoned if it is not connected to a live sales representative within 2 seconds of the person’s completed greeting. Abandoned calls may not exceed 3% of telemarketing calls answered by a person, measured over 30 days for a single calling campaign, and an unanswered call must ring for at least 15 seconds or 4 rings before it is dropped (64.1200(a)(6)). The FTC’s Telemarketing Sales Rule has a safe harbor at the same 3% (16 CFR 310.4(b)(4)), but that rule exempts most calls to businesses (16 CFR 310.6(b)(7)). Check with your counsel how these rules apply to your lists.
• United Kingdom. Ofcom’s persistent misuse policy, in force since 1 Mar 2017 and restated in a 22 Jan 2026 reminder, sets no safe abandoned call rate: any abandoned call within the caller’s control can count as misuse. The 3% figure in its 2010 policy was only a guide to which cases to pursue, and the current policy drops it. Ofcom calculates the rate per calendar day, per campaign or call center, as abandoned calls ÷ (abandoned calls + live calls) × 100. It expects an information message within 2 seconds when no agent is free, and it treats a call that answering machine detection wrongly hangs up on as a silent call, the more harmful kind.

A weekly total can hide one bad day, so check the rate daily when reps run parallel sessions. For how lines per rep drive this number, see what a parallel dialer is.

How to run the weekly review

1. Pull the week per rep and per list. Team totals hide which rep or which list moved.
2. Compare each rate with that rep’s own 4-week average. One team-wide target cannot tell a hard list from a skill gap.
3. Check the counts behind each rate. With 10 conversations, one meeting more or less moves the conversation-to-meeting rate by 10 points. Wait for more data before you act on a small count.
4. Find the first rate in the chain that fell, and pick one fix per rep from the table above.
5. Listen before you coach. Take 3 recordings or transcripts from the stage that dropped: short calls for conversation rate, calls with no next step for conversation-to-meeting rate.
6. Check every caller ID number against your daily cap and its answer rate.
7. Write down each fix and the metric it should move, and open next week’s review with it.

Use outside benchmarks as a sanity check, not as targets. The 2025 edition of The Bridge Group’s SDR research (published 6 Feb 2025) surveyed 351 B2B companies, 78% based in North America and 83% B2B SaaS. It reports 4.1 quality conversations per SDR per day, a median of 112 daily activities with 44 of them by phone, and phone-centric teams averaging 56 dials and 4.6 quality conversations a day. If your countries, deal size or buyers differ from that sample, your own 4-week baseline is the better comparison.

How DialSavvy handles this

• AI answer detection labels every answer as a person, voicemail, a phone menu or a call screener.
• Recordings and transcripts are on the call page, and after every conversation an AI summary gives the next step.
• Each number makes up to 50 calls a day and rings one person at a time.
• In parallel mode (2 lines solo, 3 with a team of 3 or more), the abandonment governor drops to 2 lines from 2.5% abandoned calls and to 1 line from 2.8%.
• Every call is itemized in your credits ledger.

Answer rates per number, as part of number health, and a live floor for managers are coming. See the DialSavvy dialer.

Before next week’s review, write your 3 definitions at the top of the sheet, pull 4 weeks of history per rep, and pick one fix per rep from the table.