Most sales plans coming together now for the upcoming cycle will rely on productivity assumptions that are already starting to break down.

And it's largely due to pressure coming from multiple directions at once.

AI search is changing how buyers discover vendors. Agentic workflows are compressing what was once hours of seller preparation into minutes. Account coverage models are being reconsidered. And yet, a faster seller does not automatically support a higher quota.

All of the above makes this planning cycle unusually consequential.

You need to distinguish realistic changes in capacity from temporary efficiency gains. You also need to understand where AI is removing friction versus where it's creating new constraints, and how current market shifts should influence resource allocation before setting next year’s targets.

To explore the new realities of revenue growth, Forma.ai Founder and CEO Nabeil Alazzam sat down with James Roth, Chief Revenue Officer at ZoomInfo, for the opening session of the 2026 Sales Planning Summit.

James is navigating these changes first-hand while leading a complex revenue organization. ZoomInfo also sits close to many of the signals shaping modern go-to-market strategy, from buyer intent and account intelligence to outbound execution and AI-enabled selling.

The central message? AI is moving the bottlenecks in the revenue engine. And now, planning teams need to update their assumptions about where demand comes from, how much work sellers can take on, what strong execution looks like, and when productivity improvements should influence quotas.

Below we get into the biggest implications from the discussion for leaders responsible for sales capacity planning, quota strategy, and sales compensation.

You need to rebuild sales capacity around how demand actually enters the business

As James made clear, the first disruption is happening before sellers even enter the picture.

For years, many software companies built highly predictable growth systems around search traffic, form fills, and inbound SDR motions. Thiscreated relatively stable assumptions for pipeline, staffing, conversion, and seller capacity.

But AI search has upended this pattern.

Buyers are now researching, evaluating, and forming vendor shortlists inside ChatGPT and Claude. Interest hasn't disappeared, but the journey has become harder for companies to see, attribute, and influence using the traditional inbound playbook.

James explained how ZoomInfo began seeing this change and why it prompted a broader reassessment of its go-to-market motion:

The silver lining here is that declining inbound traffic does not automatically mean demand has declined respectively. As James shares, buyers are still arriving. They may simply appear later in the journey, after conducting far more independent research.  

Further, visitors referred through large language models can arrive with stronger intent, translating into higher conversion rates, larger average selling prices, and stronger win rates.

But ultimately, it's not in your head. The volume, visibility, and timing of buyers has changed. And this matters enormously for planning.

A sales capacity model built around the historical supply of inbound demos will no longer cut it. When demo volume falls from four per AE per day to two, the business carries excess selling capacity even when rep productivity remains strong. Quotas, headcount, marketing investment, and territory assumptions can all drift away from the opportunity available.

For planning teams, this creates several immediate priorities:

• Reforecast pipeline supply by channel. Use recent three-, six-, nine-, and 12-month performance to understand which assumptions have structurally shifted.
• Separate demand deterioration from attribution loss. A drop in trackable form fills may require a different response than a true reduction in market interest.
• Translate channel changes into capacity decisions. Model how outbound, partnerships, events, and other channels affect demos per seller, conversion, cost, and sales-cycle length.
• Reallocate before adding. Historical headcount ratios should not automatically survive when the source and economics of pipeline have changed.

James shared that ZoomInfo has responded by increasing its investment in outbound, partnerships, networking, and in-person activity. Your organization may reach different conclusions based on your market, segment, and business model.

The larger lesson is that top-of-funnel assumptions must be treated as active inputs into sales capacity planning this upcoming cycle.

Redesign account coverage based on the work AI realistically removes

Now that AI has entered the sales workflow, headcount has become a less precise proxy for capacity.

The more useful question is how much customer-facing work each role can perform after preparation, research, administration, and follow-up have been compressed.

James described this in the context of account management.

Over the past several years, ZoomInfo had deliberately reduced enterprise account loads to create a stronger customer experience. Some account managers moved from approximately 25 accounts to four or five, supported by dedicated customer success and technical resources.

At the same time, AI began eliminating significant portions of the work surrounding customer conversations.

QBR development, presentation creation, account research, and point-of-view preparation could once consume hours or even days. But now, with the right data and workflows, much of that work can now be completed much faster, with a seller editing and refining the output.

This changes the coverage equation:

As James points out, the opportunity is meaningful, but the resulting math requires care.

Realistically, time saved does not automatically become usable capacity. It only improves the revenue model when it produces additional customer conversations, stronger account penetration, and better commercial outcomes.

Relationship complexity creates a natural ceiling here. An account manager may be able to prepare for twice as many QBRs without being able to sustain twice as many meaningful enterprise relationships. So the optimal account load will continue to depend on factors like:

• Customer segment and contract value
• Gross retention and renewal risk
• The complexity of the product and implementation
• Cross-sell and expansion potential
• The roles assigned to renewals, adoption, support, and growth
• The amount of human judgment and coordination each account requires

This is why coverage should be recalibrated at the role and segment level.

A practical starting point is to baseline how sellers currently spend their time. Identify the activities consuming the largest portions of the week, determine which can be reliably automated, and define how the organization expects recovered time to be redeployed. Then test account-load changes in controlled cohorts.

James advocates tracking whether additional capacity produces more customer meetings, pipeline, adoption, and expansion without creating declines in responsiveness, relationship quality, gross retention, or seller performance.

This turns AI productivity into something observable versus a theoretical efficiency estimate.

Build a hybrid motion that makes every human interaction more valuable

ZoomInfo’s response to declining inbound wasn't to replace its outbound organization with autonomous agents. Instead, they rebuilt the motion around a clearer division of labor.

Now, agents can prioritize signals, research accounts, identify likely members of the buying committee, retrieve relevant conversations, generate talk tracks, prepare a point of view, and automate follow-up. Wheras the human seller remains responsible for the moment that requires persuasion, judgment, credibility, and trust.

Here's James on why ZoomInfo sees this hybrid model as a more practical near-term design:

The result here is that ZoomInfo sellers can now enter conversations with stronger business acumen and a more relevant hypothesis.

But this also requires RevOps to do more than provide access to general-purpose AI.

James shared that after initially encouraging broad experimentation with agents and AI, ZoomInfo’s account management organization created approximately 8,000 agents. Many were duplicative or less effective versions of similar tools. The organization subsequently moved toward a centralized model where frontline teams could identify useful applications, while a dedicated group built, governed, integrated, and optimized the agents that truly belonged in core workflows.

Ultimately, consider how you're capturing AI use cases at your organization and preventing AI enablement from becoming an additional job sellers must perform.

The hidden requirement is context

A hybrid sales motion becomes far more powerful when AI can access a connected view of the customer. James described this as a context graph combining first-party and third-party information:

• CRM and account data
• Product usage and customer data
• Marketing engagement
• Emails, meetings, and calendar activity
• Conversation intelligence
• Contract and buying-committee history
• External company, contact, and intent signals

Connecting these sources allows AI to provide more than a generic account summary. It can identify patterns from similar customers, surface previous relationships, understand which messages created meaningful reactions, and formulate a point of view grounded in the company’s own experience.

Without this context, teams risk scaling incomplete or inaccurate assumptions. As Nabeil shared, AI doesn't repair a weak data foundation. It allows the organization to make mistakes faster and at greater scale, so it's something you need to address before agentic workflows can truly accelerate the work.

Let productivity earn its way into quota planning

Eventually, every conversation about AI efficiency reaches the same question:

If sellers can accomplish more, should their quotas increase?

The answer is likely yes over time. But the path to that answer is less straightforward.

James described a plausible future in which agents automate meeting preparation, follow-up, research, and administration, allowing an AE to conduct seven demos per day instead of four.

This increase creates theoretical capacity immediately, but it does not establish trusted capacity immediately:

As James notes, raising quotas based on the assumption that an agent will perform reliably transfers the risk of an unproven operating model to the seller. This can weaken trust, distort attainment, and obscure whether underperformance came from the rep, the territory, pipeline supply, or the technology.

AI productivity should therefore enter the quota model through evidence.

Before changing yourtargets, planning and compensation teams should determine:

1. Has the workflow been operationalized? A tool that strong sellers occasionally remember to use does not represent organization-wide capacity.

2. Is adoption consistent? Productivity assumptions should reflect actual usage across teams, managers, and segments.

3. Has customer-facing throughput increased? Administrative time savings matter when they lead to more demos, conversations, opportunities, or account engagement.

4. Is the lift sustained? Early novelty can temporarily increase usage. Planning requires evidence that the improvement persists.

5. Does productivity improve evenly? A workflow may create significant gains for one role, segment, or seller cohort and very little for another.

6. Can the organization explain the change to sellers? Credible quotas require a visible relationship between additional capacity, available opportunity, and the resulting target.

The organization also has more than one way to capture the benefit. Increased productivity could support higher quotas. But itt could also allow the business to redeploy headcount toward new markets or improve customer experience.

Each option creates different implications for cost, growth, attainment, and risk.

For 2027 planning, a scenario-based approach is likely more credible than a single aggressive productivity assumption. For instance, you might model:

• No material AI-driven lift
• Partial adoption with modest productivity gains
• Proven workflow adoption with measurable capacity improvement
• Uneven lift across roles or segments

Quota and coverage decisions could then be attached to observable triggers instad of optimism.

You'll want to consider that when productivity gains are uneven, a blanket quota increase can create avoidable fairness issues. Differentiated assumptions by role, market, and motion are harder to administer, but they will produce a more defensible plan.

The sales planning advantage will come from faster learning

As James underscored throughout the conversation, AI is changing revenue assumptions faster than annual planning cycles can handle them.

The strongest revenue teams will establish tighter connections between channel performance, seller activity, account coverage, workflow adoption, capacity, and compensation. They'll know which assumptions are established, which remain experiments, and what evidence should trigger an adjustment.

While annual planning sets the overaching direction, the organizations that outperform will keep learning after the plan's approved.

Because now the central planning question is where the revenue engine can create reliable, customer-facing capacity for generating the greatest return (far bigger a question than how many sellers the business needs).

Watch the full Sales Planning Summit conversation between James Roth, Chief Revenue Officer at ZoomInfo, and Nabeil Alazzam, Founder and CEO of Forma.ai, for more on how AI is reshaping enterprise revenue growth.