Top 10 Highlight Film Tips for Football Recruits 2027

The outcome you should expect

When AI agents absorb roughly 40% of discovery, the first thing that breaks is not the quota number — it is the assumption underneath it. A traditional quota is a bet on one thing: a rep with a territory, a ramp curve, and a fixed amount of selling time will produce N dollars. Every input to that bet moves when agents run first-call qualification, initial needs assessment, and basic competitive framing before a human joins the deal.

The outcome you should expect is a bifurcated book of business. Some portion of an account executive's pipeline arrives pre-qualified, structured, and shallow — the buyer has answered the qualification questions, but nobody senior has been engaged, no business case exists, and the discovery is thin on political context. Another portion arrives raw and complex, where the AE runs discovery personally because the deal has too many stakeholders, too much custom scope, or too much risk for an agent to navigate. These two populations behave so differently that averaging them into one quota produces a number that is wrong for both.

Concretely, most teams working through this see three effects. First, throughput per rep goes up: the same AE can carry more open opportunities because the early-stage time cost per deal falls sharply. Second, average deal value on the agent-sourced track goes down, because agents are far better at surfacing well-defined, self-serve-adjacent demand than at manufacturing enterprise consensus. Third, win-rate variance widens — agent-sourced deals cluster tightly around a modest conversion rate, while human-led deals produce a long tail of both larger wins and expensive losses.

The practical consequence for quota setting is that you stop asking "what should this rep carry?" and start asking "what should this rep carry *of each type*?" That reframing is what most revenue operations groups are actually adjusting. The headline quota may not move much — in many cases it rises, because total pipeline volume rises — but its composition, its attainment curve, and the commission rate attached to each dollar all change.

Expect the transition to take two to three quarters. Quarter one is instrumentation: you cannot split a quota you cannot measure. Quarter two is a shadow model, where you run dual-track math in parallel with the existing plan and pay on the old plan. Quarter three is the live cutover, usually at a fiscal boundary. Teams that skip straight to a live dual-track plan without instrumentation almost always end up litigating attribution disputes deal by deal, which burns more RevOps time than the plan saves.

One outcome you should *not* expect is a proportional quota cut. "Agents do 40% of discovery, so cut quota 40%" misreads what was measured. Discovery is a slice of the selling cycle, not the whole of it; removing it frees capacity rather than removing output. The correct adjustment runs through capacity math — more capacity, similar or higher quota, lower cost of pipeline generation — not through a straight-line reduction.

What drives that outcome

Three mechanics drive the split, and understanding them is what separates a defensible plan from a spreadsheet that reps immediately game.

Time reallocation. Discovery is time-expensive and cognitively front-loaded. When an agent handles scheduling, initial qualification questions, note capture, and CRM entry, the AE's first touch happens later in the cycle and starts from a written record. The freed hours do not vanish — they flow into more concurrent deals, more multi-threading, or more expansion work in the installed base. Where those hours go is a design choice, and quota should encode that choice explicitly. If you want the freed time spent on multi-threading, the acceleration track needs to be large enough and paid well enough to pull effort there.

Selection effects. Agents are routed the leads that fit their competence: inbound, mid-market, well-defined use cases, single-threaded buying groups. That routing is not neutral — it systematically hands the agent track the deals with smaller contract values and shorter cycles, and hands humans the harder, larger, slower ones. So the observed performance gap between tracks is partly the agent's doing and partly the routing rule's doing. This matters enormously for quota fairness: a rep whose territory happens to be highly agent-penetrated is not a worse rep, they are a differently-fed one, and their quota must reflect the mix they were dealt rather than the mix a peer received.

Attribution ambiguity. In a hybrid cycle, discovery is rarely all-agent or all-human. A common pattern: the agent qualifies, the AE re-opens discovery when a second stakeholder appears, then the agent handles follow-up research. Without a rule for how that deal is classified, reps will classify opportunistically — pushing deals into whichever track pays better. Every durable dual-track plan has an explicit, mechanical classification rule, decided by system-recorded events rather than rep self-report.

The reclassification gate in the middle is where most plans live or die. Set the bar too low and everything drifts to the acceleration track, because reps will always find a reason to claim they "really" did the discovery. Set it too high and reps ignore agent-sourced deals that genuinely need human depth, because the extra work is unpaid. The workable middle is an evidence standard: reclassification requires new qualifying information that the agent did not capture — a new stakeholder, a changed scope, a business case built with the buyer — recorded in the CRM before the deal advances past a defined stage. Not a checkbox; a field with content that a manager reviews on a sample basis.

A second driver worth naming: agent performance is not static. Agents are configured, prompted, and connected to data sources, and all three drift. Product changes, competitive shifts, and new objection patterns degrade qualification quality over time unless someone owns updating them. Any quota model built on last quarter's agent conversion rate will be wrong next quarter if nobody is tuning the agent. That is why the recalibration loop belongs in the plan design, not in a backlog.

Benchmarks and realistic ranges

Public benchmarks for agent-mediated discovery are thin and mostly vendor-published, so treat any single number with suspicion and instrument your own. What follows are the ranges teams tend to land in and, more usefully, how to derive your own version of each.

Track split. Most orgs that adopt a dual-track structure put 55–75% of the quota dollar target on the base track and the remainder on acceleration. The split should mirror your actual pipeline mix, not an aspiration. Compute it from trailing four quarters: what share of closed-won dollars originated from opportunities where the agent completed qualification? If that is 30%, do not set the base track at 70% — you will hand every rep an unreachable base number and a trivially attainable acceleration number, and the plan will pay out on the wrong behavior.

Conversion differential. Expect agent-sourced opportunities to convert at a meaningfully lower rate than human-led ones, often on the order of half. The differential is real but heavily confounded by the selection effect described above. Before you bake it into quota, run a fair comparison: hold out a small random slice of agent-eligible leads (5–10% is enough for a directional read within a quarter at reasonable volume) and route them to humans. Compare conversion *within that eligible population only*. That number, not the headline gap across all deals, is what your quota model should use.

Deal value differential. Agent-sourced deals typically land smaller. How much smaller depends almost entirely on your routing rules — if you route everything under a size threshold to the agent, you have defined the gap yourself. Report ACV by track and by segment simultaneously; a track-level average that mixes SMB and enterprise tells you nothing actionable.

Cycle time. The reliable, well-attested gain is time-to-first-qualified-conversation. Agents respond immediately and at any hour, so the lag between inbound signal and qualification collapses from days to hours. Downstream cycle time — qualified to closed-won — usually does not compress much, because that stage is governed by the buyer's procurement calendar, not by your responsiveness.

Capacity per rep. The straightforward model: if discovery consumed X% of an AE's selling hours and agents absorb 40% of discovery, the freed capacity is roughly 0.4X% of selling hours. If discovery was a quarter of selling time, you have freed about 10% of capacity. That is a meaningful gain and not a transformational one, and it is nowhere near a justification for cutting headcount 40%. Measure X from your own calendar and call data before modeling anything; the number varies enormously by segment.

Commission multipliers. Common practice puts the acceleration track at 1.2–1.5× the base rate and the base track at or slightly below the historical single rate. Two guardrails: total on-target earnings at 100% attainment should be at or above the prior plan, or you have quietly given the team a pay cut and will lose people; and the blended cost of sale should be modeled at several attainment levels — 70%, 100%, 130% — before the plan is approved, because multiplier structures compound at the top and can blow through the comp budget in a good quarter.

Attainment distribution. Watch the shape, not just the mean. A healthy plan produces a distribution where roughly half to two-thirds of reps land between 80% and 120%. If the dual-track plan bunches everyone at 100% on base and near zero on acceleration, the acceleration track is either too hard or not worth chasing, and reps have rationally abandoned it.

Ramp. New AEs on a heavily agent-fed territory ramp faster on the base track, because the qualification skill they lack is being supplied. They ramp *slower* on the acceleration track, because complex discovery is exactly the skill that takes time to build. Ramped quota schedules should therefore be track-specific: bring base quota to full faster than before, and acceleration quota slower.

Risks, edge cases, and failure modes

Reps stop reading agent notes. The most common failure. If agent-sourced deals pay less, the rational response is to spend less time on them — including skipping the discovery record and re-asking the buyer questions they already answered. Buyers notice, and it reads as disorganization. Mitigations: make the notes genuinely good (a summary a rep will actually use beats a transcript dump), require an explicit acknowledgment step before the first human meeting, and sample-audit calls for re-asked questions.

Classification gaming. Reps route deals to the higher-paying track by manufacturing thin justifications for reclassification. The fix is structural: classification triggers off system-recorded events with timestamps, decisions are frozen at a defined stage boundary, and disputes go through a small standing review rather than manager discretion. Publish the rule and the appeal path before the quarter starts; a rule invented mid-quarter reads as a comp change and destroys trust.

Territory inequity. Agent penetration is uneven — some segments and geographies get far more agent-handled volume. Two reps on identical quotas with 20% and 60% agent penetration face completely different jobs. Set track mix per territory rather than globally, and review the mix whenever routing rules change. This is the single most common source of mid-year comp escalations in dual-track plans.

Silent agent degradation. Agent qualification quality can decay without anyone noticing, because the leading indicator is a slow drift in downstream conversion that looks like normal noise for weeks. Instrument it directly: track agent-qualified-to-first-meeting-held and first-meeting-to-stage-two rates weekly, with an alert threshold. Keep a permanent human-routed holdout so you always have a live control group.

Over-cutting headcount. The freed capacity is real but partial. Cutting the team to match a theoretical productivity gain before the gain shows up in closed revenue leaves you unable to cover the human-led track, which is where the large deals live. Let attainment data, not projections, drive headcount decisions — and prefer holding headcount flat while raising quota over cutting heads.

Comp plan complexity. Two tracks, two rates, a reclassification rule, and a bonus is roughly the complexity ceiling for a plan reps can hold in their heads. Every additional dimension reduces the plan's ability to actually direct behavior. If a rep cannot compute their own commission on a given deal in under a minute, the plan has stopped being an incentive and become an accounting exercise.

Edge case: the expansion motion. Renewals and expansions frequently have no discovery phase at all in the classic sense. Do not force them into either track — carve them out with their own target and rate, or you will spend the year arbitrating whether a renewal conversation counted as human-led discovery.

Edge case: partner and channel-sourced deals. These arrive with discovery done by a third party. Treat them as their own classification with a defined rate, decided before the plan ships.

Edge case: the agent qualifies a deal the AE would have disqualified. Agents are tuned to a qualification bar; a rep with context may know the account is a poor fit. If disqualifying costs the rep base-quota credit, they will work bad deals to protect the number. Provide a disqualification path that returns pipeline credit when a manager agrees, so accuracy is not punished.

Legal and disclosure exposure. In some jurisdictions and industries, buyers must be told they are interacting with an automated system, and recorded conversations carry consent requirements. This sits outside comp design, but a compliance change that forces agents off certain conversations will instantly change your track mix — and therefore your quota model. Keep the plan flexible enough to absorb a routing change without a full redesign.

A practical rollout plan

Sequence matters more than sophistication. The plan below assumes a fiscal-year or quarter boundary as the cutover point and works backward.

Weeks 1–3: instrument. Add a single required field on the opportunity object recording discovery ownership, populated by the system where possible. Define the reclassification event and log it with a timestamp and a text justification. Backfill the last four quarters as best you can — even an imperfect backfill gives you a baseline distribution. Do not design the plan yet; you do not have the data.

Weeks 4–6: measure the honest differential. Stand up the holdout: route a random slice of agent-eligible leads to humans and leave it running permanently. Compute conversion, ACV, and cycle time by track *within the eligible population*. Compute agent penetration per territory. This is the input set for every downstream decision.

Weeks 7–9: model. Build the dual-track quota per territory using actual mix. Model on-target earnings at 70%, 100%, and 130% attainment against the prior plan. Model total comp cost at the company's forecast. If OTE at 100% is below the prior plan, adjust rates until it is not.

Weeks 10–12: shadow. Run the new model in parallel and pay on the old plan. Publish each rep's shadow attainment weekly. This surfaces classification disputes while they are cheap, and it lets reps argue with the model before it controls their income — which is the single highest-return step in the whole sequence.

Weeks 13–14: socialize and cut over. Present the plan with the reasoning, the territory-level mix, and the appeal path. Train front-line managers first; they will field every question. Ship at the period boundary.

Ongoing: recalibrate monthly, redesign quarterly. Monthly, review track conversion, agent health metrics, and the reclassification log for gaming patterns. Quarterly, revisit the track mix per territory and the multipliers. Freeze rates within a quarter — changing the rate mid-period is the fastest way to lose credibility with a sales team.

One organizational note: the group adjusting these numbers needs standing access to comp, CRM administration, and the agent configuration itself. Where those three sit in different orgs, the recalibration loop stalls — the RevOps team spots the drift, but the fix lives with someone whose roadmap is full. Resolve that ownership question before the cutover, not after.

Related questions

Should total quota go up or down when agents handle discovery?

Usually modestly up, because agents expand top-of-funnel volume and free selling capacity. A straight-line cut proportional to the discovery share misreads the measurement — discovery is one slice of the cycle, and removing it redeploys time rather than eliminating output.

How do you set quota for a rep whose territory is barely agent-penetrated?

Set the track mix per territory rather than globally. A rep with 15% agent penetration should carry a correspondingly small base track and a large acceleration track. Global mixes applied to uneven territories are the leading cause of mid-year comp disputes.

What single metric best signals the model needs recalibrating?

Conversion rate on the human-routed holdout compared to the agent track, watched weekly. When that gap moves materially in either direction, your quota math is stale — the agent has either improved or degraded, and the track mix and multipliers should follow.

Does this change SDR or BDR quotas too?

Yes, and usually more sharply. Agent-handled qualification overlaps directly with the traditional SDR remit, so those roles shift toward orchestration, exception handling, and outbound on accounts agents cannot navigate. Meeting-count quotas become poor proxies quickly.

How long before the new plan produces trustworthy data?

Roughly two full sales cycles. Anything sooner reflects deals that started under the old model. Resist redesigning after one quarter of partial data — that is how plans acquire the complexity that makes them ungovernable.

FAQ

Do we need a separate quota track, or can we just adjust the single number?

A single adjusted number works if your agent penetration is low and evenly distributed — under roughly 20% with little territory variance. Above that, or with uneven distribution, a single number silently overpays reps on easy agent-fed pipeline and underpays those carrying complex human-led deals. The split exists to make that difference visible and payable.

How do you stop reps from claiming credit for discovery the agent actually did?

Classification triggers off system-recorded events, not self-report, and freezes at a defined stage boundary. Reclassification requires new qualifying information the agent did not capture, entered in the CRM before the deal advances. Managers sample-audit a share of reclassifications each month, and a small standing panel hears disputes. Publishing that rule before the quarter starts matters as much as the rule itself.

What happens to the plan when the agent gets meaningfully better?

The track mix shifts and the base track grows. Recalibrate the mix quarterly, not mid-quarter, and never change a rate a rep is already selling against. If agent performance improves enough that the acceleration track shrinks below roughly a quarter of the target, revisit whether the two-track structure is still earning its complexity.

Should commission be paid on the full deal value even when an agent did most of the discovery?

Yes. Pay commission on revenue closed; use the multiplier, not the revenue base, to reflect effort. Splitting the revenue base creates disputes over percentages on every deal and makes reps' earnings unpredictable, which defeats the point of a commission plan.

How do we handle the first quarter, when nobody trusts the numbers?

Run a shadow quarter: compute attainment under the new model, publish it weekly, and pay on the old plan. Reps get to argue with the model before it controls their income, and RevOps gets to find the classification edge cases while they are free to fix. Skipping this step is the most common reason dual-track plans get abandoned.

Does this reduce the number of account executives we need?

Not proportionally, and probably not immediately. Freed discovery time is a partial capacity gain — often around 10% of selling hours once you do the arithmetic against your actual calendar data. The defensible move is holding headcount steady while raising quota, then letting two quarters of real attainment data decide whether the capacity gain is durable.

Sources

• Gartner — Sales Practice Insights
• McKinsey — Growth, Marketing & Sales Insights
• Harvard Business Review — Sales & Marketing
• Salesforce — Sales Cloud Resources
• HubSpot — Sales Blog
• Gong — Sales Research and Labs
• SaaStr — Sales and Go-to-Market
• Bessemer Venture Partners — Atlas
• Forrester — Research and Insights
• WorldatWork — Sales Compensation Resources

Related on PULSE

• [How should RevOps design commission plans when pipeline mix shifts mid-year?](/knowledge)
• [What territory design changes follow automated lead routing?](/knowledge)
• [How do you measure sales capacity when selling time per deal falls?](/knowledge)
• [When should SDR meeting quotas be replaced with pipeline-value quotas?](/knowledge)
• [How do you run a fair holdout test on an automated qualification workflow?](/knowledge)