What criteria should you use to define KPIs for a sales development rep in 2027?

The outcome you should expect

A well-built SDR scorecard changes what the rep does on Tuesday morning, not what a dashboard says on Friday. That is the only outcome worth measuring against. If you rewrite the KPI set and nobody's daily sequence of behavior changes, you did a reporting exercise, not a management one.

Concretely, a correctly defined KPI set produces three observable shifts within one to two quarters. First, dispersion between your top and bottom quartile SDR narrows on activity but widens on conversion — because the metric stopped rewarding raw dials and started rewarding whether the account was worth dialing. Second, meeting-to-opportunity acceptance by account executives climbs, typically from a chaotic 45–60% range into a stable 70–85% range, because the rep is now paid on meetings that survive AE qualification rather than meetings that merely appear on a calendar. Third, the volume of meetings booked usually *falls* by 15–30% in the first full quarter after you attach a held-and-accepted gate. That drop is the system working, not failing, and you have to socialize it with the VP of Sales before you ship the change or the first monthly business review will kill the program.

The wrong outcome to expect is a clean linear relationship between activity and revenue. It does not exist at the individual-rep level over a single month. A rep can make 1,400 touches in a month, book nine meetings, have six held, produce two opportunities, and close zero — and be an excellent rep who happened to work a slow segment. The KPI set has to be honest about that noise. It should reward the inputs the rep controls at a monthly cadence, and reserve outcome-based judgment for a rolling 90-day window where the sample is large enough to mean something.

There is a second-order outcome that matters more in 2027 than it did five years ago: a defensible KPI set is what lets you tell the difference between a rep who is good and a rep whose tooling is good. When a large fraction of first-touch sequencing, research, and drafting is automated, activity volume stops distinguishing anyone. Everyone can send 600 personalized-looking emails a week. What separates reps is judgment about which 200 accounts deserve a human, and whether their conversations convert. Your criteria have to be built to surface that, or you will promote the rep with the best automation configuration and wonder why they struggle as an AE.

Finally, expect the scorecard to become a hiring and coaching instrument, not just a comp instrument. The most valuable output of a good SDR KPI definition is a ranked diagnostic: this rep has connect-rate problems, that rep has a conversation-to-meeting problem, that third rep books fine but their meetings die in AE qualification. Those are three different coaching interventions, and a scorecard that only reports "meetings booked: 12" tells you none of them.

What drives that outcome

The criteria themselves are what drive it, and they are not interchangeable. Rank them in this order when you have to trade one against another.

Controllability. The rep must be able to move the number through their own behavior within the measurement period. Meetings booked is controllable. Closed revenue from sourced pipeline is not — it depends on the AE, the product, procurement, and a sales cycle that outlasts the quarter you are grading. This is the single most common failure in SDR scorecards: leadership wants the KPI to be revenue because revenue is what the business cares about, and the result is a rep whose comp is decided by someone else's close rate. Use revenue as a *bonus accelerator* on a 90-day or annual lag, never as the primary monthly metric.

Predictiveness. The metric has to correlate with pipeline that converts. You establish this empirically, not by assertion. Pull the last four to six quarters of sourced opportunities, and regress the SDR-stage attributes against win rate. You will usually find that one or two attributes dominate — most often whether the meeting was held (not just booked), and whether a specific qualification signal was captured before the handoff. Those become KPIs. Attributes that show no relationship to win rate are reporting metrics, not KPIs, and they belong on a diagnostic dashboard rather than in comp.

Gaming resistance. Every metric will be gamed; the criterion is how expensive the gaming is relative to doing the real work. Dials are nearly free to game (dial-and-hang-up, auto-dialer padding). Meetings booked is cheap to game (book a friendly contact, book someone with no budget, book and let it no-show). Meetings *held and accepted by the AE within five business days* is expensive to game because it requires a second party with a competing incentive to sign off. The general design rule: any KPI that a rep can satisfy without a third party's cooperation will eventually be satisfied that way.

Statistical sufficiency. A metric needs enough events per measurement period to be signal rather than noise. The rough working threshold is 30 events per period. If your SDR books 8 meetings a month, monthly meeting-to-opportunity conversion is not a meaningful KPI — the difference between 25% and 50% is two opportunities and pure luck. In that case, either lengthen the window to a rolling quarter or move the KPI up the funnel to a higher-volume event like conversations held.

Leading-indicator spread. The set as a whole should span the funnel so it diagnoses, not just scores. A practical spread: one volume-and-quality input (accounts worked at target depth), one efficiency ratio (conversation rate per touched account), one primary output (qualified meetings held), and one downstream quality gate (AE acceptance or stage-2 progression).

Benchmarks and realistic ranges

Treat every external benchmark as a sanity check on your own baseline, never as a target imported wholesale. Segment, motion, and price point move these numbers more than rep quality does. The following ranges are the ones worth arguing about internally.

Meetings held per SDR per month. Outbound SMB/mid-market typically lands in the 8–15 range; enterprise outbound runs 4–8 because the account depth per meeting is far higher. Inbound-heavy or hybrid roles run 15–30. If you are targeting more than 20 held enterprise meetings a month per rep, you are almost certainly measuring booked rather than held, or your qualification bar is decorative.

Booked-to-held rate. Healthy is 65–80%. Below 60% you have a confirmation-process problem (no reminder cadence, no calendar hold, booking too far out). Above 90% is suspicious — it usually means the rep is only booking warm inbound or is re-labeling casual conversations as meetings.

Held-to-accepted (AE qualification). Target 70–85%. This is the single most diagnostic number in the SDR scorecard. Under 60%, either the qualification criteria are not written down or the AE team is rejecting for reasons unrelated to fit (capacity, territory disputes, preference for their own sourced pipeline). Investigate before you punish the SDR.

Accepted-to-opportunity/stage-2. Usually 60–80% in organizations where "accepted" already implies a real qualification bar. If yours is much lower, "accepted" is a rubber stamp and should not be a comped gate.

Sourced-opportunity win rate versus AE-sourced. SDR-sourced pipeline commonly wins at a somewhat lower rate than inbound or AE-sourced, and that is expected — colder entry point. What matters is the trend and the gap. If SDR-sourced wins at less than half the rate of AE-sourced, your qualification definition is the problem, not the reps.

Activity ranges, with a strong caveat. Cold-call connect rates in most B2B segments sit in the low single digits per dial, and a "conversation" (a real two-way exchange, not a gatekeeper) is the unit that matters — not the dial. Email reply rates on genuinely researched, targeted sequences land in the low single digits to around 10% depending on segment and list quality. Because these ranges vary so widely by industry, the right move is to measure your own connect and reply rates for one full quarter, then set floors at roughly the 25th percentile of your own team rather than at an industry number you read somewhere.

Ramp. A new SDR typically reaches 50% of full quota by month two, 80% by month three, and full productivity around month four to six depending on complexity. Bake this into the KPI definition explicitly: a ramping rep should be graded on leading indicators (accounts researched, conversations held, call quality scores) and only transitioned to the full output KPI at a defined date. Grading a month-one rep on meetings held is how you lose good hires.

Accounts worked. For a depth-based outbound motion, 30–60 active accounts per SDR at any time is common, with 8–15 touches per account across 3–4 weeks. Higher account counts almost always mean shallower work and lower conversion; that trade-off should be an explicit decision, not an accident.

Cost per meeting. Fully loaded SDR cost divided by held meetings is a useful executive-level sanity metric. Compute it, watch its trend, and keep it off the rep's scorecard — it is not controllable at the individual level and it invites the wrong behavior.

Risks, edge cases, and failure modes

Too many KPIs. The most common failure. A scorecard with nine weighted metrics is a scorecard with zero priorities; reps optimize for whichever one is easiest that week. Cap comped KPIs at three or four, with one clearly dominant — a reasonable weighting is 60% on the primary output metric, 20% and 20% on two leading indicators. Everything else lives on a diagnostic dashboard the manager uses for coaching and the rep is never paid on.

Measuring booked instead of held. If comp triggers on booking, you get bookings. No-shows become someone else's problem, and the rep's incentive is to book anyone who says yes. Always attach the held gate, and define "held" precisely: both parties present, minimum duration (10–15 minutes is a common floor), logged in the CRM with the qualification fields completed.

The AE-rejection loophole in both directions. Tying SDR pay to AE acceptance gives the AE a lever over someone else's paycheck. AEs under capacity pressure reject good meetings; AEs who like a rep accept marginal ones. Mitigations: publish written acceptance criteria, require a rejection reason from a fixed picklist, enforce a 5-business-day rejection SLA after which the meeting auto-accepts, and route disputes to a manager who reviews the recording. Track rejection rate *by AE* — an AE rejecting at 3x the team median is a management issue, not a data point.

Segment inequity. Two SDRs on the same scorecard working different territories are not on the same scorecard. A rep assigned to a saturated vertical with 40% of accounts already in an active opportunity has a structurally lower ceiling. Either normalize quota by addressable-account count and prior-touch penetration, or accept that you are rewarding territory assignment. Refresh territory balance at least twice a year.

The automation-inflation trap. With heavy sequencing and AI-drafted outreach, activity counts inflate without any increase in real work. Two defenses: measure *conversations* and *replies* rather than sends, and add a quality gate that a human must satisfy — call recording scores sampled at 3–5 calls per rep per week, or a required research note per account that a manager spot-checks. If your activity KPI can be satisfied by a scheduled job, it is not a KPI.

Deliverability as a hidden variable. An SDR whose reply rate collapses may have a domain reputation problem, not a skill problem. Before you put a rep on a performance plan for low reply rates, check inbox placement, domain warm-up, bounce rate, and whether sending volume per mailbox crossed a threshold. This misdiagnosis is common and expensive.

Attribution disputes. When an SDR touches an account that later converts through a marketing channel, who gets credit? Decide the rule before the quarter, not during the dispute: a common workable rule is that the SDR is credited if they had a meaningful two-way interaction within a defined window (often 60–90 days) before opportunity creation. Write it down, apply it mechanically, and do not relitigate individual deals.

Changing the definition mid-period. Never. Even an obviously better definition, shipped in week six of a quarter, destroys trust more than the bad metric was costing you. Announce changes at least one full period ahead, run the old and new definitions in parallel for one period, and show every rep what their number would have been under both.

Small-team noise. With three SDRs, ranked leaderboards are meaningless and actively corrosive. Use absolute thresholds against a documented standard instead of relative ranking until you have enough reps for the ranking to carry information.

Data quality undermining everything. If meeting outcomes are logged manually and inconsistently, every downstream KPI is fiction. Before shipping a scorecard, verify that the held/no-show status, the source attribution, and the qualification fields are captured automatically or enforced as required fields at the stage transition. A KPI you cannot audit from system-of-record data is not a KPI.

A practical rollout plan

Ship the change on a defined schedule, in parallel, with the old numbers still visible. The sequence below takes roughly one quarter and is deliberately slow at the front.

Weeks 1–2: baseline. Pull four to six quarters of history. For every SDR-sourced opportunity, record: touches before first conversation, booked date, held/no-show, AE accepted or rejected with reason, stage reached, closed-won or lost. This is where you discover your data is worse than you thought — budget time for it. Compute your own distributions for every candidate metric, and note the event counts per rep per month so you know which metrics clear the statistical-sufficiency bar.

Week 3: choose the set. Apply the four criteria in order. Draft a one-page definition document per KPI: exact name, precise definition including edge cases, system of record, query or report that produces it, refresh cadence, owner, and the weighting. If a KPI cannot be written down unambiguously in five sentences, it is not ready.

Week 4: pressure-test with the people affected. Walk the draft through two AEs, two SDRs (one top performer, one struggling), the sales manager, and RevOps. The specific question to ask each SDR: "How would you hit this number if you were being lazy?" They will tell you, accurately, and you will fix two or three loopholes you did not see.

Weeks 5–8: shadow period. Run the new scorecard in parallel. Pay on the old plan. Publish both weekly. Track the delta per rep and investigate any rep whose ranking moves more than one quartile — that is either a real insight or a definition bug, and you need to know which before it affects pay.

Week 9: recalibrate targets. You now have eight weeks of real data on the new definitions. Set the floor near the 25th percentile of observed performance and the target near the median, then move the target up deliberately over the following two quarters. Setting the initial target at the top performer's number is the fastest way to make a new scorecard illegitimate.

Week 10: announce. Publish the definition document, the shadow-period results, the effective date, and — critically — the appeal process. Every SDR should be able to reproduce their own number from a report they can open themselves.

Weeks 11–13: cut over with a floor. Go live at the start of a full period. Guarantee a comp floor for the first period (commonly the rep's trailing three-month average) so nobody is financially punished by a transition they did not choose. This costs a little money and buys the credibility that makes the whole thing work.

Ongoing: quarterly review, annual redefinition. Each quarter, re-run the correlation between the KPI and closed-won to confirm predictiveness still holds — this is the check most teams skip, and it is what catches a metric that has quietly decayed. Redefine at most annually, and always with a full period of notice.

Related questions

How many KPIs should an SDR be paid on?

Three or four at most, with one dominant. A workable split is 60% weight on qualified meetings held, 20% on a leading efficiency ratio, and 20% on a quality gate. Additional metrics belong on a coaching dashboard the rep is never compensated against.

Should SDR comp be tied to closed revenue?

Not as the primary metric — the rep cannot control the AE's close rate or the sales cycle. Use closed revenue from sourced pipeline as a lagging accelerator or annual bonus on a 90-day-plus window, where the sample size and time horizon make it fair.

How do you stop reps from booking junk meetings?

Trigger comp on held-and-accepted rather than booked, define "held" with a minimum duration and required CRM fields, require AEs to give a picklist rejection reason within five business days, and track rejection rate by AE to catch abusive rejection patterns.

What KPIs apply to a ramping SDR?

Grade months one through three on leading indicators: accounts researched to standard, conversations held, call-recording quality scores, and sequence completion. Transition to the full output KPI at a published date, usually month three or four, with a stepped quota of roughly 50% / 80% / 100%.

How often should SDR KPI definitions change?

Redefine at most once a year, announced at least one full measurement period in advance and run in parallel for a period. Targets can be recalibrated quarterly; definitions should not. Mid-period changes destroy trust faster than a mediocre metric costs you.

FAQ

Is meetings booked ever an acceptable primary KPI?

Only in a very early-stage team with no AE capacity to accept meetings and no historical data to establish predictiveness. Even then, treat it as temporary and instrument the held rate from day one so you can migrate to held within a quarter or two. Left in place, meetings booked reliably produces no-show inflation and low-fit bookings, because it is the cheapest metric in the entire funnel to game.

How do you handle an SDR supporting multiple AEs with different acceptance standards?

Publish a single written acceptance standard that all AEs are held to, require picklist rejection reasons, and monitor rejection rate by AE against the team median. If one AE rejects at three times the median, that is a management conversation with the AE. Where standards genuinely differ by segment, split the SDR's target by segment rather than blending them into one unachievable average.

Do activity metrics still belong on the scorecard?

Yes, but as floors rather than targets, and expressed as conversations and replies rather than dials and sends. A floor says "below this, we have a coaching conversation"; a target says "maximize this," which is exactly the behavior automation makes meaningless. Set floors from your own team's 25th percentile after a full quarter of measurement, not from a published industry number.

What is the minimum data infrastructure needed before shipping a new scorecard?

System-of-record capture of: meeting booked timestamp, held/no-show status, AE accept/reject with reason and timestamp, opportunity creation linked to the sourcing rep, and stage history with dates. If any of those five are manually entered and inconsistently populated, fix the capture first — a scorecard built on unreliable data will be disputed within the first pay period and you will never get the trust back.

How do you set the first target when there is no history?

Work backward from the revenue plan: required pipeline divided by average deal size gives opportunities needed, divided by your best estimate of held-to-opportunity conversion gives held meetings needed, divided by headcount and ramp-adjusted months gives per-rep target. Then sanity-check that number against the benchmark ranges above, and explicitly label it provisional for the first two quarters.

Should the same KPI set apply to inbound and outbound SDRs?

No. The output metric can be the same (qualified meetings held), but the leading indicators and the targets should differ substantially — inbound is measured on speed-to-lead and lead-to-meeting conversion, outbound on account depth and conversation rate. Blending them into one scorecard systematically overrewards whichever role has the easier funnel that quarter.

Sources

• https://hbr.org/2003/11/coming-up-short-on-nonfinancial-performance-measurement
• https://www.gartner.com/en/sales/topics/sales-performance-management
• https://www.mckinsey.com/capabilities/growth-marketing-and-sales/our-insights
• https://sloanreview.mit.edu/article/the-perils-of-bad-strategy-metrics/
• https://www.salesforce.com/resources/research-reports/state-of-sales/
• https://www.bls.gov/ooh/sales/home.htm
• https://hbr.org/2012/07/how-to-design-a-winning-sales-compensation-plan
• https://www.forrester.com/blogs/category/sales-development/

Related on PULSE

• [How do you define leading vs. lagging KPIs for a sales team in a one-hour workshop?](/knowledge/st812)
• [The Sales Rep Time Management Reboot — 60-Min Training](/knowledge/st152)
• [What is the best framework for defining sales KPIs for a new B2B team in 2027?](/knowledge/st807)
• [How do you align sales team KPIs with company revenue targets during a 60-minute training?](/knowledge/st810)
• [What is a typical sales compensation structure as a percentage of gross profit to the rep?](/knowledge/st0804)
• [60-Min Sales Training: Champion Development](/knowledge/st0449)