How do we design comp for deal teams (AE + SA + Sales Engineer) where all three touch the deal but at different stages?

The outcome you should expect

A correctly structured deal-team plan changes three measurable things inside two quarters, and it is worth knowing what "working" looks like before you touch a single percentage.

The first is that credit disputes collapse. Teams that move from a verbal, negotiated split to a CRM-tag-driven split typically see arguments drop from a recurring monthly finance escalation to a handful of edge cases per quarter. The mechanism is boring: when the split is computed from fields that were stamped with a timestamp during the deal, nobody has a story to tell at close. Disputes shift from "who deserves this" to "was this field filled in correctly," which is an auditable question with a yes/no answer.

The second is that technical resource allocation gets rational. When the Sales Engineer's variable pay is tied to closed deals they were tagged on, SEs start triaging their own calendar. They push back on demos for unqualified opportunities and they chase the AEs with real pipeline. Before the change, the SE queue is usually first-come-first-served and the loudest AE wins. After, the SE queue is driven by deal quality because the SE's paycheck depends on the deal actually closing, not on the demo happening.

The third is that AE ownership survives. This is the outcome most teams accidentally destroy. If you split commission evenly across three people, the AE — the one person accountable for close rate, forecast accuracy, and the customer relationship — is being paid the same as two people who cannot influence whether the contract gets signed. That plan reads as fair and behaves as poison. AEs stop pulling deals across the line, they stop doing the unglamorous procurement and legal chasing, and they start looking for a role where closing pays like closing.

What you should not expect is a plan that eliminates judgment entirely. Somewhere between 5% and 10% of deals will have a genuinely weird shape — an SE who sourced the opportunity from a conference, an SA who saved a deal that had already been lost, a partner-influenced deal where the AE did almost nothing. Build an exception path that runs through a comp committee, cap it at roughly 10% of deals per quarter, and require written justification. If more than 10% of your deals need an exception, the base model is wrong and you should fix the model rather than keep granting exceptions.

The financial outcome to target is a total deal-team cost that sits at a defensible multiple of the revenue it produces. For a team of one AE carrying $1.5M in new ACV, with fractional SE and SA coverage, total loaded variable-plus-base cost in the $450k–$550k range yields roughly $2.75–$3.30 of ACV per dollar of comp. That is a healthy ratio for a mid-market or enterprise motion with meaningful technical validation. If you land under $2.00, either your quotas are too low or you have over-staffed the technical roles relative to deal complexity.

What drives that outcome

Three mechanisms do the actual work, and each one maps to a specific design choice you make on purpose.

Stage ownership is the unit of credit, not the deal. The deal is a sequence of gates: prospecting, discovery, technical demo, solution design, security and legal review, and signature. Each gate has exactly one owner. The AE owns prospecting, negotiation, and signature. The Sales Engineer owns the technical demo and any proof-of-concept execution. The Solutions Architect owns use-case design, feasibility assessment, and implementation scoping. When you write the plan, you are not deciding "how much is an SE worth" — you are deciding how much each gate is worth, and then paying whoever owned that gate.

A weighted point model makes this concrete. Assign prospecting 10 points, discovery 20, demo 30, solution design 20, and legal-through-close 40 — 120 total. Tally the points each role owned on a given deal, divide by the total awarded, and you have the split. On a typical deal where the AE prospects, runs discovery, and closes (10 + 20 + 40 = 70), the SE demos (30), and the SA is uninvolved, the AE takes 70/100 = 70% and the SE takes 30%. On a complex enterprise deal where the SA runs discovery alongside the AE and co-owns design, the AE lands near 55–60% and the technical roles take the rest. The math is objective, time-stamped, and auditable, which is the entire point.

Different accelerator thresholds keep the roles from optimizing the same lever. If every role's accelerator kicks in at 100% of quota with the same multiplier, you have effectively built one plan with three names on it. Stagger them: the AE accelerator at 100–110% attainment, the SA at 110–115%, the SE at 115–120%, all at roughly a 1.25x multiplier on incremental commission. The AE's threshold is lowest because close volume is directly her lever. The SE's is highest because technical wins are lumpier and less controllable, so the accelerator should represent genuine over-performance rather than normal variance. The SA sits between.

A participation floor prevents starvation. Because SE and SA variable pay depends on being tagged into deals, an AE who never books them can zero out their comp through pure neglect. Set a minimum guaranteed variable component — commonly 30–40% of the SE's target variable, paid quarterly regardless of AE demand — so the technical roles are insulated from routing failures they cannot control. Pair it with a capacity metric: an SE covering three AEs should be tagged into roughly 60–75% of their deals; an SA covering the same three should be in 40–55%, weighted toward enterprise. If actual participation is far below that, the problem is deal routing or headcount, not the comp plan, and no amount of percentage tuning will fix it.

Benchmarks and realistic ranges

Here is a fully worked plan for a three-person deal team supporting a mid-market-to-enterprise motion. Treat the numbers as a calibrated starting shape, not gospel — adjust for your ACV, cycle length, and market.

Account Executive. Base $100k, target variable $150k, OTE $250k at a 40/60 base-to-variable split. Annual new-ACV quota of $1.5M, which is a 6x OTE-to-quota ratio — the standard band is 4x–6x, with 5x–6x appropriate where the AE has dedicated technical support. Commission rate of 10% of ACV on the AE's credited share. At 100% attainment the AE earns $150k variable; at 120% attainment, with the accelerator kicking in above 110% at 1.25x, she earns roughly $180k–$187k variable.

Sales Engineer. Base $80k, target variable $60k–$65k, OTE $140k–$145k at a roughly 55/45 to 60/40 base-to-variable split. The heavier base weighting is deliberate: the SE cannot control close rate, so the plan should not put half their income on an outcome they only influence. Commission rate of 2% of credited ACV. Supporting three AEs at $1.5M each is $4.5M of team quota; at 70% participation the SE is credited on roughly $3.15M, producing about $63k of variable. Guarantee floor of $30k paid quarterly against that target.

Solutions Architect. Base $85k, target variable $35k–$45k, OTE $120k–$130k at a 65/35 to 70/30 split. The SA carries the highest base ratio because the role is the most consultative and the least tied to a specific transaction. Commission rate of 1.5% of credited ACV. At 50% participation across the same $4.5M — the SA is pulled in mainly on enterprise and complex deals — that is $2.25M credited and roughly $34k of variable.

Total team economics. Loaded comp across the three roles for one AE's coverage is approximately $250k + $143k + $119k = $512k against $1.5M of ACV, or $2.93 of ACV per comp dollar. If you are running an SE and SA shared across three AEs, the fully-loaded picture is $750k of AE comp plus $263k of technical comp against $4.5M — a considerably better $4.44 per dollar, which is why shared technical coverage is the norm rather than dedicated pods.

Split ranges by deal shape. AE sources, AE and SE run the demo, AE closes: 70/20/10. AE sources, SA and SE own discovery and demo, AE closes: 60/15/25. SE or SA surfaces the opportunity from an existing account and the AE closes: 65/20/15. Genuine three-way equal contribution: 50/25/25 — this should be rare, and if it is showing up on more than one deal in ten, your tagging discipline has drifted toward everyone tagging everything.

Carve-out alternative. If a points engine is more machinery than you want, run a simpler carve-out: the AE takes a flat 50–60% and the remainder is paid as fixed per-event bonuses. The SE earns a set bonus — typically $500 to $2,000 depending on ACV band — for every demo that leads to a closed-won deal. The SA earns a comparable bonus per completed solution design or feasibility study. Sized as 10–20% of the AE's commission on the deal, a $500k deal paying the AE $10,000 would pay the SE and SA $1,000–$2,000 each. The trade-off is real: carve-outs are trivially easy to administer and explain, but they do not scale with deal size, so on a $2M deal the technical roles feel underpaid relative to the effort. Use carve-outs under roughly $250k average ACV and the percentage model above it.

Cadence and clawback. Pay all three roles on the same cycle — monthly or quarterly, following cash collection or invoice, whichever your finance team already uses. Apply clawback symmetrically: if a deal churns inside the clawback window (90 days is typical, 180 for annual prepay), all three roles give back proportionally. Asymmetric clawback, where only the AE is exposed, tells the technical roles that deal quality is not their problem.

Risks, edge cases, and failure modes

Equal splits. The 33/33/33 plan is the single most common failure in this design and it fails in a specific, predictable way. It reads as egalitarian and behaves as a diffusion of responsibility. The AE, who alone controls whether the paper gets signed, has the same upside as two people who do not. Meanwhile every role has an incentive to claim involvement in every deal, because presence equals payment. Within two quarters you get inflated tagging, an SE who attends thirty low-value calls to maximize surface area, and an AE quietly interviewing elsewhere.

Untagged deals and retroactive assignment. If stage tags are not captured during the deal, they get assigned at close by whoever argues most persuasively. The fix is mechanical: require the tag within 48 hours of the milestone or the points for that stage go unawarded. Enforce it with a gatekeeper rule — no deal advances to Closed-Won without a populated Demo Owner and Legal Lead field, and the Demo Owner field cannot be self-set by the AE. Audit the top 10% of deals by size every month; when a deal is found with missing or inaccurate tags after the fact, reduce the AE's credited share by 10–15% and redistribute. The rule is blunt, but it converts tagging from an administrative chore into a paycheck-relevant behavior, and accuracy climbs fast.

Paying technical roles on deals they source. If the SE or SA is paid at the same rate on self-sourced deals as on supported ones, you have quietly created two more AEs with worse quotas. They will prospect instead of support, because prospecting is where the leverage is. Keep source credit separate: pay a flat referral SPIFF — $1,000 to $3,000, or a small percentage capped in absolute dollars — for sourcing, and keep the stage-based split for the actual selling work. The SPIFF acknowledges the contribution without redirecting the role.

The stalled-deal problem. An SE delivers an excellent demo, the AE mishandles the close, and the deal dies. Under a pure closed-deal commission model the SE earns nothing. This is defensible in principle — it aligns the SE to care about close rate — but it becomes corrosive if it happens repeatedly with the same AE. Two mitigations: the guarantee floor described above, and a quality SPIFF that pays the SE a modest bonus when their demo-to-close conversion clears a threshold such as 40% over a rolling two quarters. The second one rewards the part of the outcome the SE actually controls.

Multiple SEs or SAs on one deal. Split the allocated percentage among contributors, weighted by stage ownership rather than hours logged. Hours invite gaming and are miserable to audit. If two SEs both tag as Demo Owner, either one is wrong or the deal genuinely had two technical workstreams — in which case they split the 20% evenly. Cap the number of taggable technical contributors per deal at two per role; beyond that, someone is padding.

Unilateral override authority. If any single person — usually the AE's manager or the AE herself — can override a computed split without a second signature, the model is decorative. Require two approvals for any override: the RevOps owner of the comp engine and a finance controller. Log every override with a written reason and review the log quarterly. A rising override rate is the earliest available signal that the plan no longer matches how deals are actually worked.

Territory and account-transfer edge cases. When a deal spans a territory change or an AE departure, credit follows the stage tags that already exist. The stages already stamped stay with the original owner; remaining stages go to the successor. This is far cleaner than a "whoever owns the account at close takes everything" rule, which creates an incentive to sit on deals until a transfer lands.

Over-engineering. A points engine with fourteen stages, six modifiers, and a complexity coefficient will be understood by exactly one person in RevOps and trusted by nobody. If a rep cannot compute their own expected split on a napkin, the plan has failed regardless of how correct the math is. Five stages, three roles, one override path.

A practical rollout plan

Run the change over one full quarter, in parallel with the existing plan, before anything is binding.

Weeks 1–2: instrument the CRM. Create the stage-owner fields — Prospecting Owner, Discovery Owner, Demo Owner, Solution Owner, Legal Lead, Close Owner — as user-lookup fields with timestamps, not free text. Add validation that blocks stage advancement past the demo gate without a populated Demo Owner, and blocks Closed-Won without Legal Lead. Build the split calculation as a formula or a scheduled job that reads the tags and writes a computed split to three read-only currency fields on the opportunity.

Weeks 3–4: backfill and calibrate. Take the last two full quarters of closed-won deals and reconstruct the tags from activity history, calendar records, and call recordings. Run those deals through the new engine and compare the computed splits to what was actually paid. You are looking for two things: how large the deltas are, and whether any individual would have been materially worse off. Deltas above 15% on a meaningful share of deals mean your stage weights are wrong — adjust the point values, not the individual outcomes.

Weeks 5–8: shadow run. Keep paying under the old plan. Publish the computed split to all three parties on every deal that closes, labeled clearly as informational. This surfaces tagging failures while they are still free, and it lets each role see the model's logic on their own deals rather than in a slide.

Weeks 9–12: dispute-resolution dry run. Collect every objection raised during the shadow period and route it through the exception process you intend to use in production — RevOps plus finance, written justification, decision inside five business days. You want the process load-tested before real money is attached. Track the exception rate; if it exceeds 10% of deals, delay go-live and re-weight.

Quarter boundary: go live. Issue new plan documents to all three roles with the point table, the split ranges by deal shape, the accelerator thresholds, the guarantee floors, and the override policy stated explicitly. Every rep signs. Do not go live mid-quarter — split-plan quarters generate exactly the retroactive arguments the design exists to eliminate.

Ongoing: monthly audit, quarterly review. Finance audits the top 10% of deals by value each month for tag accuracy and pays out corrections. RevOps reviews four metrics quarterly: dispute count, override rate, SE and SA participation rates against the 60–75% and 40–55% targets, and ACV per comp dollar. Adjust point weights at most once a year, at plan reset. Mid-year changes to the weights destroy the credibility the auditability was supposed to buy.

Related questions

Should the SE and SA have their own quota, or share the AE's?

Share a derived quota. Set the SE's credited-ACV target as the sum of supported AE quotas times expected participation — three AEs at $1.5M with 70% participation gives a $3.15M target. A separate independent quota turns them into sellers.

What happens when a deal closes with no SE or SA involvement at all?

The AE takes 100%. Unawarded stage points simply drop out of the denominator, so the split renormalizes automatically. This is correct and common on renewals, small transactional deals, and repeat purchases where no technical validation is required.

How do we handle comp when the SE supports pre-sales and post-sales work?

Split their plan by time allocation. If the SE spends 70% of their time pre-sales, weight 70% of variable to deal-based commission and the remaining 30% to post-sales metrics such as implementation milestones or technical CSAT. Do not pay one pool for both.

Does this design work for a two-person team without a Solutions Architect?

Yes. Drop the solution-design stage points and redistribute them — typically 10 points to discovery and 10 to demo. The AE lands near 70–75% and the Sales Engineer near 25–30%, which is the standard two-role shape.

Should managers get credit in the split?

No. Manager compensation should come from team attainment against a roll-up quota, not from participating in individual deal splits. Putting a manager in the split gives them an incentive to insert themselves into deals they should be coaching from a distance.

FAQ

What is the biggest mistake companies make when designing comp for a three-person deal team?

Splitting commission equally at 33/33/33. It removes ownership from the AE, who alone controls close rate, forecast accuracy, and the customer relationship, and it rewards every role for claiming involvement in every deal. A weighted split of roughly 60–70% AE with 15–25% each to the technical roles aligns pay with the stage each person actually owns.

How do we prevent arguments over credit at close?

Compute the split from CRM stage-owner fields captured with timestamps during the deal, and forbid retroactive changes. Require tags within 48 hours of a milestone or the points for that stage go unawarded. Block Closed-Won until Demo Owner and Legal Lead are populated. When the split is arithmetic on data already stamped, there is nothing left to argue about.

Should the split be tied to deal size or to role contribution?

Tie the percentage to role contribution and let deal size scale the dollars. A percentage model automatically pays more on a $2M deal than a $200k one while keeping the ratio between roles constant. Flat per-event bonuses are simpler to run but under-reward the technical roles on very large deals — use them below roughly $250k average ACV.

How do we stop an AE from starving the Sales Engineer of deals?

Two guardrails. First, a guarantee floor of roughly 30–40% of the SE's target variable, paid quarterly regardless of participation, so routing failures do not become the SE's pay cut. Second, a booking rule: the AE books the SE at deal creation rather than the week of the demo, and unbooked deals simply close without SE credit.

Can we mix base salary with the variable split?

Yes, and you should — but weight the mix differently by role. The AE runs 40/60 base-to-variable because outcome control is highest. The Sales Engineer runs closer to 55/45 and the Solutions Architect 65/35, because neither controls whether the contract gets signed. Identical mixes across three roles with different levels of outcome control is a design error.

How often should we change the stage point weights?

Once a year at plan reset, and no more. Mid-year weight changes destroy the auditability that made the model credible in the first place, and they invite exactly the retroactive renegotiation the design exists to prevent. If disputes, override rate, or participation rates drift badly mid-year, document it and fix it at the next reset — the RevOps discipline of holding weights stable is worth more than the incremental accuracy.

Sources

• https://hbr.org/2012/07/motivating-salespeople-what-really-works
• https://www.worldatwork.org/resources/publications/workspan
• https://www.alexandergroup.com/insights/sales-compensation/
• https://www.gartner.com/en/sales/topics/sales-compensation
• https://www.shrm.org/topics-tools/topics/compensation
• https://www.mckinsey.com/capabilities/growth-marketing-and-sales/our-insights
• https://www.bls.gov/ooh/sales/sales-engineers.htm
• https://openviewpartners.com/blog/

Related on PULSE

• [How should a 2027 sales org calculate partial-deal credit when multiple reps touch a deal?](/knowledge/q12436)
• [How do you calibrate interview expectations for different sales stages (SDR vs AE vs Sales Engineer)?](/knowledge/q360)
• [How do we execute take-out campaigns that convert competitive losses into wins on the second touch?](/knowledge/q479)
• [How do you balance automation and human touch when buying committees shrink in 2027?](/knowledge/q16467)
• [What are the top three operational challenges when integrating an AI sales assistant into an existing RevOps tech stack in 2027?](/knowledge/q13522)