Everyone wants artificial intelligence (AI) in sales. Far fewer are willing to redesign sales around it.
That is the disconnect behind much of the disappointment with AI results. Revenue teams now have copilots, summaries, prompts, and workflow add-ons. They move faster in pockets of the sales process. But selling itself has not fundamentally improved. Gartner research found that 39% of chief sales officers say AI-focused sales initiatives have increased the percentage of sellers who meet or exceed quotas..
The problem is not that AI cannot help sales. The problem is that most organizations used it to optimize tasks when they needed to redesign decisions.
More Activity Still Does Not Mean Better Outcomes
In most sales organizations, AI has been layered onto the same operating rhythm that existed before. Reps update CRM fields, generate decks, summarize calls, and prepare deal notes more quickly. Managers still run forecast meetings built around status updates and subjective debate. Coaching conversations remain broad and repetitive. The same questions get asked, the same risks surface late, and forecast commitments are still made largely on instinct.
AI may reduce friction. But when it accelerates activity without improving judgment, performance rarely changes.
Usage is not strategy. Outcomes are.
Start with the Decision, not the Task
Status is infinite. Decisions are finite.
Sales organizations can always inspect more activity, review more pipeline, and ask for more updates. But only a small number of decisions determine whether revenue, margin, and risk improve.
One of the biggest reasons sales AI underperforms is that it is deployed as a task assistant rather than a decision engine. It helps complete work faster but does not improve the moments that actually determine results.
A more effective approach starts with decision points. Should this deal remain in commit? Does pricing require escalation? Is stakeholder coverage strong enough to move forward? Is the proposed next action the right one? These are finite, high‑value moments that determine revenue, margin, and risk. That is where AI belongs.
Consider a commit deal that has aged beyond benchmark, lacks executive engagement, and has no confirmed mutual action plan. In a task-first model, AI summarizes the risk and the manager asks for an update. In a decision-first model, those signals trigger a commit review. The manager must either validate the commit, move the deal out, or define the action required to keep it in. The difference is not the insight. The difference is the decision the insight forces.
Decision‑first selling replaces status‑heavy discussions with clear gates. AI can surface signals such as stage age versus benchmark, stakeholder coverage, or whether a mutual action plan exists. People then apply context and judgment around strategic importance, competitive credibility, and executive engagement. The goal is not a longer meeting. It is a better call, made earlier, with fewer surprises later.
Decision Rights Create Speed
A decision-first workflow should answer four questions: What signal does AI surface? What judgment must people apply? Who has the right to decide? What gets audited later?
AI does not create speed. Explicit decision rights do.
Revenue teams need to define who recommends, who decides, when exceptions escalate, and what gets audited later. In forecasting, AI can recommend while managers decide. In pricing, sellers and AI can propose within approved ranges, with escalation triggered only when thresholds are crossed. For next best action, AI can guide the default path while sensitive cases route to oversight. Metrics such as override rates, margin impact, forecast delta, cycle time, and conversion provide feedback on whether the model is working.
Many leaders worry that involving AI in decisions creates risk. In reality, ambiguity creates far more risk. When every recommendation requires extended debate or committee review, inconsistency grows and speed collapses. Clear guardrails allow organizations to push execution to the edge while keeping accountability intact.
The Real Unlock is the Manager
The most overlooked element of sales AI is the frontline manager.
AI does not change seller behavior directly. Managers do. That makes the frontline manager the highest-leverage conversion point between AI insight and commercial performance. Without that translation layer, AI becomes another dashboard: visible, interesting, and mostly disconnected from behavior. Gartner research shows that managers who use data-driven guidance to identify the highest-impact coaching opportunities are 4.3 times more likely to exceed expected profit growth. Yet only 15% of frontline managers primarily use data to guide the focus of coaching conversations. That gap helps explain why so many AI initiatives stall after rollout. The insight exists, but the translation layer is weak.
Managers need three new muscles:
- Signal literacy: Understanding what matters and why.
- Translation: Connecting data to the reality of the deal, account, or seller.
- Coaching conversion: Turning insight into a focused behavior change that is reinforced over time.
Design the Default Path
The default path is the route work follows when no one intervenes. In an AI-enabled sales organization, that path needs to be designed, not inherited. Leaders should decide where AI can recommend, where it can act, where sellers can override, where managers must intervene, and where governance must inspect patterns over time.
AI can create speed, but people still protect empathy, trust, and complex judgment. That makes operating‑model design more important, not less. The central question is no longer whether to adopt AI. It is where to automate, where to escalate, and where managers must intervene.
Organizations that focus too heavily on adoption often see quality erode. Customer interactions become less thoughtful. Coaching becomes noisier. The better approach is to design the default path intentionally. Clear decision rights accelerate execution. Guardrails protect margin and risk. Governance prevents drift. A simple discipline helps keep AI grounded in outcomes: inspect weekly and decide monthly.
One Workflow at a Time
The move from experimentation to impact should be concrete. Pick one workflow. Redesign one decision gate. Specify what the AI recommends, what people decide, when exceptions escalate, and what gets audited. Then equip managers to diagnose the signal, explain its importance, coach the action, and reinforce the behavior.
That is how AI starts to sell. Not by replacing sellers or automating another task, but by improving the quality and speed of the decisions that matter most.
The teams that win this next phase will not be the ones with the most tools or the most AI. They will be the teams that redesign work so AI sharpens judgment, strengthens coaching and improves the decisions that revenue growth depends on.
Doug Bushée is a VP Analyst in the Gartner Sales Practice who presented live on this subject and others at the Gartner CSO & Sales Leader Conference, May 19-20 in Las Vegas, NV.
