Factory
The SaaS engine that scales with ruthless efficiency.
The Pattern
What the Factory pattern is
Every step of a Factory is measured and has an SLA (service level agreement, a set standard it has to meet), and the whole system is tuned to get predictable results out of inputs that vary. On a real factory floor that's conveyor belts and robot arms running in sync with quality-control sensors, except here what goes in is signups and what comes out is expanding accounts.
Think: Toyota, Dell, Printful, Stripe, AWS, Shopify, HubSpot, Salesforce, Twilio, Atlassian, Figma (pre-Adobe deal).
The Engine
How it makes money
A Factory's keystone is its product-led growth loop, a cycle that feeds itself because the product does the selling. The product brings users in and gets them active, then keeps them and expands what they use, and the marginal CAC (customer acquisition cost) keeps falling as it goes. If the loop breaks, the product stops distributing itself. Expansion stalls and retention drops, and you're left with a services business that has software bolted on. Toyota and Dell run the same logic with physical goods through a productized supply chain, so it's operational rather than digital, but the economics are similar.
A Factory runs on repeatable systems, where the marginal cost of serving the next customer trends toward zero. Every interaction, product decision and marketing dollar becomes a measured input the team can repeat. Once the loop matures, growth compounds as CAC falls, expansion revenue comes from existing customers, and gross margin (revenue minus direct costs) sits at 70-90%. That turns the work the team puts in into infrastructure that keeps paying back long after the work is done.
The Discipline
Operating principle
In hindsight, every Factory looks like it prints money. But it spends the 18 to 36 months it takes to build the loop swallowing cash. The founder has to stay patient long enough for the efficiency to compound, instead of switching to a growth hack that breaks the unit economics (what each customer brings in compared with what they cost).
A Factory focuses on retention and unit economics, backed by operational excellence, for one specific persona (a clearly defined type of user) and a workflow they repeat often.
Healthy
You know a Factory is healthy when it can say "we are indispensable for X persona doing Y job," and it doesn't add a second persona until the first one is happily expanding.
Unhealthy
A Factory that tries to serve everyone usually ends up serving no one well.
We'd check every year that existing customers are spending at least as much as they did the year before, because once they spend less, every dollar of growth has to be won from new customers.
What Works & What Breaks
Healthy vs. breaking
What healthy looks like
What breaks it
The Profile
Profile
Best Fit Founder
A Factory fits an operator or engineer who's obsessed with process and product and is fluent in metrics. They'll give engineering and go-to-market equal weight. They can wait 18-36 months for the loop to show measurable lift, and they'll kill any feature that doesn't move retention.
Go-To-Market
Factories sell SaaS subscriptions with freemium-to-paid upgrade paths (a free tier that converts to paid), and product-led growth is the main motion. Self-serve onboarding converts users without anyone from the company stepping in. A sales-assist layer gets added later, once enterprise ACVs (annual contract values) cross $25K-$50K. The product does the selling, and people only step in to clean up edge cases.
Ideal Focus
We'd focus on operational excellence and customer retention, plus unit economics, for one persona and one high-frequency workflow, aimed at one outcome where the value delivered is clear and measurable. Start narrow and expand later into more personas, adjacent workflows and platform plays. Trying to serve everyone usually leaves a Factory serving no one well, because the loop never reaches the activation density it needs to compound (enough active users of the same kind in one place).
Revenue Pattern
Once the loop matures, the recurring revenue is stable and it compounds, with strong NRR still providing growth when new-logo acquisition slows. Early on the curve looks flat and unimpressive. Then the retention and expansion math kicks in, and the curve bends into compounding growth. Healthy Factories can forecast within 5% on a 12-month horizon.
Unhealthy Signs
Risk, Speed & Burnout
Sales Cycle
Self-serve customers make their first conversion in minutes to days. When sales leads the deal, it takes 30-90 days in the mid-market and 90-180 days for true enterprise. The first paying users arrive in 90-180 days. A product-market fit signal takes 18-36 months, and it's 36-60 months before the loop scales on paid acquisition.
Hiring Sequence (first 5 after founder)
- Founding engineer: or a technical co-founder, if the founder isn't the one building the product
- Second engineer / first product designer: protects velocity and UX (user experience) as the product expands
- Head of growth / product marketer: owns the loop instrumentation (the tracking on each step) and the retention math
- Customer success / support lead: comes in early, because retention is revenue
- First account executive: comes in only after the self-serve loop is producing qualified inbound
The stack is cloud-native (AWS/GCP/Azure), with modern observability (Datadog, etc.) and product analytics (Amplitude, Mixpanel, Heap). The CRM (customer relationship management system) is tied to product events, and Factories invest heavily in internal tooling.
The rhythm is sprint and iterate. Product and engineering carry a high cognitive load, and go-to-market runs at a moderate intensity that holds steady over time. It's less spiky than the work in a Lab or a Conversion Engine.
Leading KPIs
- Net Revenue Retention (NRR): this is the single most important Factory metric, because below 100% the compounding logic is broken and growth has to come entirely from new logos.
- Gross margin %: every other unit-economics ratio is built on it, and a declining gross margin usually points to cloud cost creep or a loss of pricing power.
- CAC payback period: the ideal target is under 12 months and under 24 months is acceptable, and anything longer strains cash flow and signals weakening product-led acquisition.
- LTV/CAC ratio: this is the headline efficiency metric, where below 3:1 the model is broken and above 5:1 the company is likely under-investing in growth.
- Churn % and logo retention: these are the leading indicator of an NRR decline, because logo churn shows up before dollar churn does.
- ARR/MRR growth rate: this is the headline growth metric, but MRR (monthly recurring revenue) and ARR growth only mean something next to the efficiency metrics above.
- Activation rate (% of signups reaching the "aha" moment): this upstream metric controls everything downstream, so if activation drops, every other number eventually follows.
- Active user % (DAU/MAU): daily over monthly active users measures how sticky the product is, which is a leading indicator of expansion and retention.
- Magic Number / efficiency of growth spend: this is the lens public markets use to judge whether the loop is investable at scale.
- Support ticket resolution time: this quality signal often correlates with retention, and rising resolution time usually comes before churn.
Feature Focus
Processes and quality control let the business scale while ARR compounds.
The delivery model is rigid with little flexibility if the loop breaks, and churn stays invisible until it suddenly isn't.
CFO Watchpoints
What to watch
Defense
Defensibility
A true Factory's revenue comes from a long tail of self-serve customers, so concentration stays low. The risk goes up as the company moves upmarket, so we'd watch for the top-10 customers going over 20% of ARR.
The market is crowded and fast followers are a real risk, since the technology underneath is rarely defensible at the code level. Every promising Factory has 5-50 competitors within 18 months. Who wins comes down to how fast you ship and how deep your integrations go, and to how embedded you get before competitors catch up.
Switching costs, integrations and workflow lock-in are what protect a Factory, and sometimes network effects do too. Matching code is rarely what protects you, because competitors can ship the same features in 12 months. What they can't replicate is years of integrations and the habits a customer's team has formed. Most customers don't notice that lock-in until they try to leave and find they'd have to rebuild the integrations and retrain the team.
Switching cost is medium-high. Once the product is embedded in a customer's workflows, with their data and integrations in it and their team trained on it, they stay even when a competitor is nominally cheaper.
Financial Impact
The money
Impact
Factories create the most steady-state value in the Product Machines family. Once they're at scale, they support the highest revenue multiples in public markets and the most predictable cash compounding. Gross margins stay high as long as churn is under control, and the recurring revenue compounds.
Target P&L Allocation
| Category | Target |
|---|---|
| Client Acquisition / Ad Spend | 38% |
| Total Labor | 17% |
| Fixed Overheads & G&A | 18% |
| Legal & Professional | 3% |
| Operating Income (EBITDA) | 22-26% |
SaaS economics; watch churn and CAC payback religiously
Source: Compass archetype one-pager, target allocation as a percent of revenue
Ops and support get a base salary plus retention or customer-success bonuses. Engineers get salary plus equity, with the cash at market rate. Engineering comp is non-negotiable, because underpaying leads directly to worse product quality, and that compounds into churn within 12-18 months. Customer success is paid on retention and expansion, not on ticket volume.
Sales gets the classic SaaS split of 50/50 base and commission. AEs (account executives) are paid on new ARR booked, with accelerators above quota and clawbacks on customers who churn within 12 months. Multi-year deals pay out on year-one ACV, not on total contract value.
Risk is low for self-serve B2B SaaS, where card failures cause 1-3% involuntary churn, and it's higher in consumer SaaS. In the enterprise tier, prepaid annual contracts essentially eliminate bad debt.
Working capital runs negative, because Factories collect annual prepays before the monthly delivery costs come in. At scale it's a funding source here rather than a need.
Revenue Quality Score
9 / 10. The revenue recurs and expands on software margins, and since you can predict it, this is about as good as revenue gets. The missing point is customer concentration risk, which creeps in as the company moves upmarket.
Trajectory
Where this goes
The first ceiling is the total addressable market within the persona and workflow you serve, and after that it's how creative you can be with expansion loops. Past 200 employees, engineering org design becomes the binding constraint. The velocity that produced product-market fit gets lost in coordination overhead, and the rebuild takes 12-24 months. Beyond 1,000 employees the constraint becomes how extensible the platform is.
Optionality is high. A Factory can move upmarket or downmarket, into new geographies or into adjacencies, because the asset is the product loop underneath and not any single market position.
Strategic acquisition is most common, at $20M-$200M ARR. A mature Factory can also go to a private-equity rollup, or IPO at $100M+ ARR with strong growth and Rule of 40 compliance. Founders often keep a meaningful operational role through the exit and beyond it.
Common Founder Failure Modes