The automation maturity model: where does your team sit?

One of the most useful things about working with thousands of companies is that you start to see patterns that aren't visible from inside any single organization. The way companies adopt automation is one of those patterns — and it's remarkably consistent regardless of company size, industry or technical sophistication.

Over the past two years, we've mapped those patterns into a five-level maturity model. It's not a theoretical framework — it's an empirical one, built from observing what 9,000 teams actually do, what blocks them, and what unlocks the next stage of adoption. Here's the model, and where you probably sit within it.

Level 1 — Reactive

At Level 1, automation is something you reach for when the pain of not having it becomes unbearable. There might be a few Zaps running — a Slack notification when a new lead comes in, a Google Sheet that gets a new row when a form is submitted — but there's no deliberate strategy. Automation is a painkiller, not a system.

The defining characteristic of Level 1 is that every automation is a response to a specific incident. The lead notification Zap exists because someone missed an important lead six months ago. The spreadsheet automation exists because the person who used to update it manually left the company.

The primary blocker at Level 1 is almost always organizational, not technical. Teams at this level typically don't think of automation as something they own — it's something that happens to them, or something that requires a developer to implement. The path forward is a single successful automation that a non-technical team member builds themselves. Once someone sees that they can build a working workflow in an afternoon without engineering help, the mental model changes permanently.

Level 2 — Systematic

At Level 2, the team has identified its most painful repetitive processes and automated them deliberately. There are 5–15 active agents, they're mostly in one department (usually the one where the first enthusiastic automator works), and they're maintained by one or two people.

The blocker at Level 2 is scope — automation is still seen as a specialized skill rather than a general-purpose tool for everyone. Non-technical team members don't build their own workflows; they request them from the one person who knows how. This creates a bottleneck that caps how much automation value the team can extract.

The path forward is deliberate enablement. Identify the three to five processes in other departments that would benefit most from automation, and work with those department leads to build the first workflow together — with the explicit goal of leaving them capable of building the next one themselves.

Level 3 — Cross-functional

At Level 3, automation has escaped its original department and spread across the organization. Sales has agents. Support has agents. Ops has agents. There's usually a dedicated owner — a RevOps manager, an operations lead or a technically-minded department head — who maintains the stack, sets standards and onboards new users.

The characteristic challenge at Level 3 is governance. With multiple departments building their own workflows, you start to see duplication (three different departments have built their own lead notification agents), inconsistency (different departments use different field naming conventions for the same data) and fragility (nobody knows which workflows depend on which others).

The path forward is a lightweight governance layer — a shared naming convention, a documented workflow registry, and a monthly review where someone checks for duplication and inconsistency.

Level 4 — Intelligent

Level 4 is where AI steps enter the picture in a meaningful way. The team isn't just triggering actions — it's making decisions. Lead scoring, content classification, sentiment analysis, anomaly detection in financial data — all running as agents, replacing judgment calls that previously required human review.

Teams at Level 4 typically have 25–50 active workflows and are starting to think about agent reliability metrics. They track success rates, latency and error rates per workflow. They have alerts for critical workflow failures. They review run logs proactively rather than only when something breaks.

The path forward involves expanding AI step usage from a single use case to systematic coverage of any decision that involves pattern recognition in structured data — which is more of your business than you probably think.

Level 5 — Autonomous

Level 5 teams function like software engineering teams when it comes to operations. Agents handle the majority of repetitive work. Humans focus on strategy, edge cases and continuous improvement. The ops team maintains a workflow roadmap the same way engineering maintains a product roadmap. New automations are scoped, prioritized and shipped on a regular cadence.

The metric that defines Level 5 is leverage — the team can scale operations without proportional headcount growth. Every new workflow built compounds: a workflow built this quarter creates time savings every quarter from now on. Teams at Level 5 typically save 20+ hours per person per week and report that automation has fundamentally changed what roles on their team are responsible for.

Most teams reading this article are at Level 2 or 3. That's exactly the right time to be thinking about this framework — because the moves that take you from Level 2 to Level 4 are clear, achievable and significantly more impactful than anything you'll do at the margins of your current level.