How many accounts should one SDR actually work per week?
The honest answer is that it is a division problem, not a benchmark. Take the selling minutes a rep really has in a week, divide by the minutes one account takes at the depth you expect, and that is the number. Everything else is negotiation. The reason teams get it wrong is that they set the number from the size of the list rather than from the size of the week, and the list has no opinion about how long research takes.
The second half of the answer is that the number should be smaller than the list and larger than zero for a specific reason: the population of accounts with a live reason to talk is much smaller than rep capacity in most markets. Once you accept that, the question changes from how many accounts can we cover to how few can we cover properly, and that is the question that improves results.
Start with the week, not the list
A forty hour week is not forty hours of account work. Subtract pipeline reviews, one to ones, training, CRM administration, inbound handling, and the meetings a rep books and then has to attend. What is left, the genuinely uninterrupted account-working time, is the only input that matters, and it is a number you can measure rather than assume: ask three reps to log it honestly for one week.
Then measure the second input the same way. Time one rep doing one account properly, from opening the research to logging the touch: reading the primary source, writing the consequence sentence, identifying the role, finding the person, and sending. Do it for five accounts and take the median. Those two measured numbers are the entire model.
The capacity math
The table below shows the arithmetic at three depths of work, using an illustrative twenty five hours of real account-working time per week. Replace both columns with your own measured values. The point is the shape: depth and coverage trade against each other linearly, and no messaging improvement changes the trade.
| Row | Depth of work | What it includes | Minutes per account | Accounts per week |
|---|---|---|---|---|
| 01 | Deep, event-led | Read the primary source, write the consequence, identify the role, find the person, multi-channel first touch | 45 | about 33 |
| 02 | Standard, event-led | Verify one event, adapt an opener to it, single channel plus one follow-up | 20 | about 75 |
| 03 | Light, list-led | Template send with a merged field, no research | 5 | about 300 |
| 04 | Mixed, realistic | 20 deep accounts plus 40 standard accounts in the same week | 45 and 20 | 60 |
Now overlay the second constraint, which is the one that decides whether the coverage is worth anything. If you sampled your list and found that roughly one account in eight currently carries a dated event, then a thousand account territory contains about one hundred and twenty five accounts with a reason attached. A rep working sixty accounts a week at mixed depth covers that live population in about two weeks. Coverage is not the binding constraint. The supply of live windows is.
What to hold reps to instead of account count
Account count is a poor target because it is trivially satisfiable by working shallower. If you must have an activity target, pair it with a quality gate that makes shallow work impossible to log, which in practice means requiring the evidence fields to be filled before a touch counts.
- Accounts touched with a dated event attached, not accounts touched. A touch with a blank event field does not count.
- Share of touched accounts where a consequence sentence and an owning role were recorded before the send.
- Reply rate split by whether the touch referenced a public event, which tells you whether the research is paying for itself.
- Accounts correctly deferred: rows the rep looked at, found nothing on, and moved to the watch list. Rewarding this is what stops manufactured relevance.
That last one is the cultural change. In most teams a rep who works fifty accounts and defers forty of them looks less productive than a rep who touches all fifty. The first rep is doing the job correctly, and the metric has to say so or nobody will do it twice.
Compare your numbers to published research
Do not take a per-rep account number from a blog post, including this one. Do compare your structure against real survey research: The Bridge Group has run a biennial study of the sales development role for a decade, and the 2025 edition reports on organizational structure, ramp, retention, metrics, quota, and compensation based on survey responses from 351 B2B companies. Use that kind of source to check whether your team's shape is unusual, and use your own two measured inputs to set the number.
This is what the category calls sales development capacity and account coverage.
The category calls this capacity planning, account coverage, or the coverage model, and it shows up in vendor material as territory design or book size. When people argue about it they are usually arguing about two different things at once: how many accounts a rep is assigned, and how many a rep actually works.
Keeping those separate is the whole trick. Assigned book size can be large, because a book is a watch list. Worked account count has to be small enough that each touched account carries a dated event, a consequence, and an owning role. A team that conflates the two ends up with reps who have five hundred accounts and no reason to call any of them.
Questions people ask next
Is there a right number of accounts per rep per week?
No, there is a right method. Measure real account-working minutes and median minutes per account, divide, then check the answer against how many of your accounts actually carry a dated event. If capacity exceeds the live population, work fewer accounts more deeply.
How big should the assigned book be?
Large enough to contain a useful watch list, since watching costs nothing if it is automated, and small enough that a rep can be accountable for noticing when something changes. The book is a monitoring surface, the weekly worked list is the job.
What if leadership insists on a high activity target?
Agree to the target and attach the evidence gate: a touch counts only when the event date, source link, consequence, and owning role are recorded. The number then becomes self-limiting and honest, and the conversation moves to where it belongs, which is the supply of live windows.
Where to check this yourself
Primary records and published research, not vendor blog posts. Every link was checked on September 1, 2026.
- The Bridge Group, SDR Models and Metrics researchblog.bridgegroupinc.com
Biennial survey research on the sales development role. The 2025 edition is the tenth, based on responses from 351 B2B companies.
- The Bridge Groupbridgegroupinc.com
The research firm behind the study, with its other published work on inside sales structure and metrics.
- Census Business Dynamics Statisticscensus.gov
Official firm-level data, useful for sizing how many companies of your target profile actually exist in a territory.
- The Short Life of Online Sales Leads, Harvard Business Reviewhbr.org
Research on inbound response speed, relevant when part of the rep's week is spent on hand-raisers rather than outbound.
Intakra shrinks the working list to the accounts that currently carry a dated event, a named consequence, and a role that absorbs it, so a rep's week goes to accounts that have a reason attached.
Free scan, no signup, no card