Growth
How agent product growth curves differ from traditional SaaS — per-user task volume is cognitively bounded, expansion comes from workflow depth not user base. Activation, expansion, referral, onboarding quadrants.
Where 150-200% NDR comes from
Agent products commonly show a counterintuitive number to investors: NDR (Net Dollar Retention) reaches 150-200%, well above the SaaS “excellent” benchmark of 120%. The gap is not because agent products acquire customers more aggressively — it’s because they have two growth paths SaaS lacks: workflow expansion and capacity expansion. This section unpacks how the three growth paths reinforce each other, and the three specific ways agent product growth differs from SaaS:
1. Per-user task volume is cognitively bounded
SaaS tools have near-zero marginal usage cost, so a power user can use them 100 times a day. Agent products have a human cognitive bound — the things a user can “think to ask” the agent to do per day are limited, far below the usage frequency ceiling of tool-style SaaS.
Implication: growth cannot rely solely on “make active users use more” — that path has a low ceiling. Growth comes more from “bring in new users” and “expand the types of work each user delegates to the agent”.
2. Expansion comes from workflow depth, not user base
The SaaS expansion path is land-and-expand: deploy in one team first, then expand to the whole company. Agent products have one additional expansion path: expanding workflow types on the same user — originally only used the agent for email, expanding to calendar management, data analysis, internal information retrieval.
Implication: expansion opportunities first appear on already-activated users, then in the organization. See expansion.
3. The time-to-value window is shorter
SaaS tools tolerate slow “value realization” — a user spends a week configuring a Notion database, but the next 5 years of retention is normal. Agent products tolerate less — when a user makes their first request and the agent fails or produces poor quality, the probability of a second request drops significantly.
Implication: first task success rate and quality determine the activation funnel — far more important than signup conversion. See activation.
Growth four-quadrant
Split by “existing / new users” × “existing / new workflows”:
| Dimension | Existing workflow (same task type) | New workflow (new task type) |
|---|---|---|
| Existing users | Increase task frequency for one workflow | Expand workflow types per user |
| New users | Acquisition (on validated workflows) | Acquisition into new scenarios (risky) |
Each quadrant requires a different operational strategy:
- Existing users × existing workflow: increase frequency. Through product optimization (faster, more accurate) + reminders (scenarios users didn’t think of). Modest growth but stable margins
- Existing users × new workflow: unlock new task types. Through product iteration (adding new agent capabilities) + education (users don’t know the agent can do this). Large expansion potential but prone to quality issues
- New users × existing workflow: routine customer acquisition. Growth engine; focus on activation funnel optimization
- New users × new workflow: highest risk, invest cautiously. Most agent products should not let “new users first try new workflows” — first task should use the most stable, highest-completion workflow
Activation path
See activation. Core conclusion: agent product activation is defined as “first successful task completion”, not “signup” or “first login”. Conversion loss at any intermediate step is more severe than SaaS.
Expansion mechanisms
See expansion. Three expansion paths: seat-based (more users), workflow-based (same user, more task types), capacity-based (same user, same task, larger volume caps).
Cross-section connections
- Activation quality criteria: metrics/north-star
- Expansion paths coupled to tier design: pricing/tier-design
- Customer onboarding operational implementation: playbooks