Sales scoring model template with three worked examples

A scoring model turns mixed sales activity into one comparable number. Build it in four moves: pick three to five activities your CRM already records, derive a point value for each from your own stage conversion rates, write a guardrail for every metric that can be padded, then check the leading/lagging balance before anyone sees a board. This page gives three complete models you can copy — high-volume inside sales, outbound SMB, mid-market — plus the arithmetic to re-derive every weight against your own funnel. If you have not yet picked which activities to score, start with the KPI shortlist.

The derivation, in one formula

Every weight comes from the same calculation. For any stage:

Margin value of one unit = (period wins ÷ period volume at that stage) × gross margin per win

Then divide every stage's margin value by the smallest one. That quotient is the point value.

Two things fall out of this that matter.

First, the model is self-balancing across stages. Stage volume × per-unit value = volume × (wins ÷ volume × margin) = wins × margin. The volume term cancels, so every derived stage contributes the same expected point total over a period. That is not a coincidence to be fixed; it is the correct starting point, because each stage carries the whole funnel's margin exactly once.

Second, it means any emphasis you want is a deliberate override. Multiply your constraint stage by 1.25 to 1.5 and say so out loud in the announcement. Reps accept a stated tilt. They do not accept an unexplained one.

Model A: high-volume inside sales

Assumptions: $1,200 average deal, 60% gross margin ($720 per win). Trailing quarter: 2,000 connected calls → 400 demos → 100 wins.

Derivation: a call is worth 100 ÷ 2,000 = 5% of a win = $36. A demo is 100 ÷ 400 = 25% of a win = $180. A win is $720. Normalising on the call at 1 point gives $36 per point.

Two point classes are doing different jobs. Derived points track economic contribution. Behaviour points buy habits that have no clean conversion maths behind them. Keep behaviour points under roughly 10% of expected total, or the board stops describing revenue.

Model B: outbound SMB

Assumptions: $6,000 average deal, 70% gross margin ($4,200 per win). Trailing quarter: 1,500 unique outbound touches → 150 replies → 60 meetings booked → 45 meetings held → 12 wins.

Derivation: touch = 12 ÷ 1,500 = 0.8% of a win = $33.60. Reply = 8% = $336. Meeting booked = 20% = $840. Meeting held = 26.7% = $1,120. Normalising on the touch at 1 point gives $33.60 per point.

The booked/held split is the single most useful guardrail in outbound. Full credit on booking rewards a rep for a calendar entry that never happens; full credit on attendance alone punishes reps whose prospects cancel for reasons they cannot control. Splitting 10/23 keeps both signals live. More patterns of this kind are in the competition rules template.

Model C: mid-market

Assumptions: $30,000 average deal, 60% gross margin ($18,000 per win). Trailing quarter: 400 qualified conversations → 100 discovery meetings → 50 multi-threaded opportunities → 25 proposals → 10 wins.

Derivation: conversation = 2.5% of a win = $450. Discovery = 10% = $1,800. Multi-threaded = 20% = $3,600. Proposal = 40% = $7,200. Normalising on the conversation at 1 point gives $450 per point.

Multi-threading carries the ×1.5 tilt because in a mid-market funnel that is normally the stage where deals die. Apply the tilt to your constraint, not to the stage that is easiest to measure.

Checking the leading/lagging balance

Compute expected points per stage — point value × expected period volume — then take leading points as a share of the total. A workable range is 70% to 85% leading. Below that the board mostly reports luck; above it, the board stops connecting to revenue.

Model A as written sits below the range, which is what a two-stage funnel does: with only two derived leading stages the ceiling is 67% before behaviour points. The fix is one of three moves — halve the win weight, add a third leading stage such as qualified conversations, or accept 69% and tell the team the board leans on outcomes. Do not fix it by inflating behaviour points.

Adapting the weights to your own team

1. Re-derive, do not copy. Pull 90 days of stage volumes from HubSpot, Pipedrive or wherever your funnel lives, and run the formula. Borrowed weights encode someone else's conversion rates.
2. Round to something a rep can hold in their head. Snapping 23.4 to 23 is fine. Snapping 125 to 100 is not — you have just moved 20% of the model's weight.
3. Re-baseline every quarter. Conversion rates drift and a stale model quietly rewards the wrong stage. In Sales Gamification the Coach pass watches for exactly that drift and flags it before the board goes stale.
4. Cap what can be manufactured. If a rep can create a scoring event with no counterparty, the metric needs a cap, a duration floor or a uniqueness rule. Every row in the tables above has one.
5. Check fairness before launch. If tenure or territory spread is wide, an absolute points board is decided in advance; switch the ranking formula rather than the scoring model. The options are laid out in the leaderboard template.

If you would rather not do the derivation by hand, Blueprint reads your pipeline and produces the scoring model, the gaming-risk audit and the fairness correction in about four minutes, with a stated reason attached to every weight. You can start a 14-day trial with no card required, and pricing is flat per team rather than per seat — the full breakdown sits on the pricing page.

FAQ

How many activities should a sales scoring model include?

Three to five. Fewer than three and the board is a single-metric race that ignores how deals actually progress. More than five and reps cannot recall what earns points, which removes the behavioural effect the model exists to create. Each of the three models on this page scores four or five activities, with at most two of them behaviour-based rather than derived from conversion rates.

How do you stop reps gaming a points system?

Attach a written guardrail to every metric before launch. The four that cover most cases: a duration floor (calls under 90 seconds do not count), a uniqueness rule (one scoring event per contact per week), a counterparty requirement (an external attendee must be present), and a clawback window (win points reverse if the deal cancels within 60 days). Audit which metrics can be manufactured without a counterparty — those need caps.

Should closed revenue be worth the most points?

Per unit, yes — a win is the most valuable single event, so derived maths always gives it the largest weight. Across a period, no. Because there are far fewer wins than activities, closed revenue should contribute roughly 15% to 30% of total expected points. If it contributes more, the board rewards deal flow that was largely decided before the scoring period began.

How often should point weights be recalculated?

Quarterly, or sooner if a stage conversion rate moves by more than about a fifth. Weights are derived from conversion rates, so when the rates change the weights are wrong, and the board starts pointing reps at a stage that no longer limits revenue. Re-pull 90 days of stage volumes, re-run the derivation, and publish the change with the reason rather than adjusting silently.

Can the same scoring model work for new and tenured reps?

The model can; the ranking usually cannot. Keep one scoring model so the definition of good work is consistent, then change how scores are compared — percentage to individual target, tenure bands, or improvement against each rep's own trailing baseline. Changing the point weights per rep instead makes the programme unauditable and invites the accusation that the scoring is arranged around a favourite.