Expansion Mechanisms

Three expansion paths for agent products — seat-based, workflow-based, capacity-based; applicable scenarios and combined strategy.

Expansion refers to existing customers paying more within their contract — the source of SaaS NDR > 100%, with major LTV impact.

Agent products have three expansion paths, of which workflow-based is unique to agents — traditional SaaS has no exact analog.

Three expansion paths

Seat-based expansion

Expand by seat (user count). Customer deploys agent in one team, expands to more teams / the whole company.

  • Suitable when: agent is a personal productivity tool (writing, coding, research assistant)
  • Trigger: internal word-of-mouth; sales reaching out to new departments
  • Per-seat value is typically stable; expansion curve mirrors SaaS experience
  • Ceiling: the customer company’s total headcount

Workflow-based expansion

Expand by workflow (task type). Customer initially uses the agent for one task type, expands to having the agent handle other task types as well.

  • Suitable for: all enterprise-targeted agent products
  • Trigger: users discover new scenarios; sales demonstrates new capabilities; product iteration adds new workflows
  • Per-user ARPU increases continuously, unbounded by company size
  • Ceiling: total tasks the user can offload to the agent — usually far exceeds seat-based ceiling

This is agent products’ most important and most underestimated expansion path — traditional SaaS has no such growth curve, so agent product early NDR may appear unusually strong (130-180%).

Capacity-based expansion

Expand by capacity (task volume cap). User upgrades from Lite to Pro to Ultra — same user, same workflow, larger task ceiling.

  • Suitable for: subscription / hybrid models with tier caps
  • Trigger: user hits tier cap; product auto-prompts upgrade
  • Per-user ARPU increases, but each individual expansion is small (one tier at a time)
  • Ceiling: user’s total task volume demand

Combined strategy across three paths

Three-axis expansion flywheel Agent products grow on three reinforcing axes — workflow drives capacity, capacity triggers seat outreach Seat 1 team Few depts Company-wide Enterprise more seats → more workflow surface area Workflow 1 workflow 3 workflows 8 workflows Depth ceiling more workflows → higher token usage Capacity Lite · 200/mo Pro · 1.5K/mo Ultra · 10K/mo Custom Time → The three axes compound multiplicatively, not linearly. A mature customer advances on all three simultaneously. Capacity upgrade lifts ARPU, which triggers sales attention, which drives seat expansion back to top — closing the loop.

Mature agent products run all three expansion paths in parallel:

New user enters → Activation (Lite tier)
                  ↓
               workflow count = 1 → monthly tasks < 200    → Retained
                  ↓ Workflow expansion                      ↓ Capacity expansion
               workflow count = 3 → monthly tasks 200-1000 → Upgrade to Pro
                  ↓ More workflow expansion                 ↓ More capacity expansion
               workflow count = 8 → monthly tasks > 1500   → Upgrade to Ultra
                  ↓ Internal word-of-mouth                  ↓ Sales outreach
               Other departments adopt → Company-wide → Enterprise contract

The three curves reinforce each other: workflow expansion drives capacity demand → capacity expansion raises ARPU → high ARPU triggers sales attention → seat expansion to whole company.

Early signals of likely expansion

Which customers will expand? When the following signals appear together, expansion probability rises significantly:

SignalInterpretation
Per-user workflow count > 3User has discovered agent multi-scenario value
Monthly tasks approaching 80% of capCapacity expansion window has opened
Task completion rate > 90%Agent quality is stable on this user’s workflows
7-day return rate > 60%User has formed a usage habit
Pattern of “repeated use of same workflow”Workflow embedded in user’s daily routine

Customers showing four or more signals together: sales should proactively pitch upgrade. Customers showing fewer than three: avoid pitching; improve product experience first.

Expansion anti-patterns

  • Pitching upgrade too early — pitching right after activation when the user has not yet perceived value; typical low conversion + trust damage
  • Pursuing seat expansion while neglecting workflow expansion — seat is a ceiling-bounded path; workflow is agent’s unique growth engine
  • Capacity expansion via “hit the cap then upgrade” — passively waiting for users to hit caps; should proactively prompt at 80% (captures expansion while avoiding hard-stop churn)
  • Workflow expansion relying entirely on user discovery — most users will not spontaneously try new agent capabilities; product-side recommendation + sales-side education are required

Differences from traditional SaaS NDR

NDR (Net Dollar Retention) = same-cohort customer revenue this period vs prior period.

Traditional SaaS: NDR is driven primarily by seat expansion + price increases; 120% is considered excellent.

Agent products: 150-200% NDR is achievable — but a significant portion comes from workflow + capacity expansion, not seat. Investors evaluating this need to decompose:

  • High NDR primarily from capacity expansion: risk is customer-hits-cap-then-upgrades, but per-task gross margin may deteriorate (see metrics/unit-economics)
  • High NDR primarily from workflow expansion: healthiest form, reflecting deep agent penetration into customer operations
  • High NDR primarily from seat expansion: traditional SaaS pattern; if agent product has only this curve, it suggests insufficient product penetration

Cross-section connections

Was this page helpful?
Edit on GitHub