Building your growth engine.

The three stages, and why they are different

Growth is not a flat discipline. There are three stages with different physics, different tactics, and different failure modes.

• Launch - you are trying to get the first 50 to 500 users or 10 to 30 paying customers. No attribution model works yet. Noise dominates signal. The question is: can anyone be persuaded to care?
• Grow - you have product-market-fit signal. You are trying to turn one or two working loops into a repeatable motion. Attribution starts mattering. The question is: which loops compound?
• Scale - loops are identified. You are pouring fuel onto them and optimizing conversion. The question is: where does the CAC/LTV ratio break, and at what volume?

Running grow-stage playbooks in launch is a waste. Running launch-stage playbooks at scale is a waste. Figure out which stage you are actually in.

A decision tree for stage identification

Four questions:

1. Do you have at least 10 customers who pay real money (not pilots, not friends)?
2. Can you point at 3 identifiable sources for those customers (a channel, a referral loop, a specific content piece)?
3. Have those sources produced customers in the last 30 days, not just historically?
4. Do you know, roughly, the cost per customer acquired through each source?

Zero or one yes: launch. Two or three yes: grow. Four yes: scale.

Most founders overclaim their stage by one. A company with 3 paying customers, all from the founder's network, is launch-stage regardless of how much funding it has. A company with 40 customers from a mix of cold outbound, SEO, and referrals is grow-stage, not scale-stage.

Launch stage: positioning is the only lever

In launch, channel choice barely matters. Positioning dominates.

If you cannot finish the sentence “We are the [category] for [ICP] that [unique value] because [proof]” in one breath, no channel is going to save you. Spend the first month of launch writing and rewriting that sentence.

Once positioning is defensible:

• Founder-led outreach. 30 to 50 ICP conversations per week. Cold, warm, referral, it does not matter. Build the map of who cares.
• Three to five owned pieces of content that assert the positioning loudly. Launch essay, case study, demo, comparison page, founder bio.
• One owned channel where you show up consistently. LinkedIn or X usually. Twitter for developer-focused products; LinkedIn for enterprise.
• Forget paid. Paid ads in launch are the worst kind of feedback loop - you cannot distinguish “people like our ad” from “people like our product.” Save the money.

Goal of launch: reach 10 to 30 paying customers where you can clearly articulate what brought them in. That clarity is the graduation criterion.

Grow stage: find the two loops that work

Grow stage is where most companies fail, because the temptation is to run everything.

The answer is the opposite. Pick two loops. Run them for 90 days. Measure. Kill one if it underperforms. Replace it. Run 90 more days. Graduate when one loop is reliably producing customers at a CAC you can live with.

Common loops to pick from:

• Content + SEO / GEO. Compounds slowly, produces warm inbound. Needs 6 to 12 months to show real lift. See the SEO & GEO playbook.
• Outbound to ICP. Fast signal, high effort, hits a ceiling. Works best for enterprise or mid-market with concentrated ICP.
• Community + events. High-touch, relationship-driven. Works when your ICP hangs out in identifiable spaces.
• Partner / integration. Ride another product's distribution. Works if you plug into a large existing user base.
• Product-led. Free tier, self-serve, virality. Works for broad TAM, low-friction products. Hard for enterprise.
• Paid search / social. Works once you know what positioning converts. Rarely works before that.

The key discipline: do not run a loop for 30 days and give up. Do not run 6 loops at once with $2k each of test budget. Commit. Measure. Cut. Repeat.

Scale stage: CAC, LTV, and the conversion obsession

At scale, the loops are known. The game is: can you pour more fuel into them without breaking unit economics?

The three numbers that matter:

• CAC per channel. Blended CAC is a vanity metric. Channel-level CAC is actionable.
• LTV per cohort. Gross margin LTV, not revenue LTV. A $20k/yr customer who takes $15k of support is not a $20k customer.
• Payback period. Months to recoup CAC at gross margin. Under 12 months for SMB, under 18 for mid-market, under 24 for enterprise. Much longer and the model is fragile.

At scale, the anti-patterns are different from grow stage. The risk is not “running the wrong loop.” It is “pouring too much into a loop that still works but at a lower margin than leadership is pretending.”

Attribution: do not overinvest, do not ignore

Attribution is a spectrum:

• None (launch stage). Ask every customer where they heard about you. Write it in a shared doc.
• Light (early grow). UTM parameters on every link, form field “how did you hear about us,” a spreadsheet to aggregate.
• Medium (late grow, early scale). Analytics tool (GA4, PostHog, Amplitude), self-reported attribution, simple last-click + self-reported hybrid.
• Heavy (scale). Multi-touch attribution, marketing mix modeling, incrementality testing. Worth it at $10M+ in revenue, not before.

Attribution is useful the moment you are making channel-spend decisions. Before that, it is a distraction.

The founder's role by stage

• Launch: founder is the growth engine. 70 percent of founder time is on customers, positioning, and content.
• Grow: founder makes the first hire (VP growth or head of marketing). Founder still pitches top 10 accounts. 30 to 50 percent of founder time is GTM.
• Scale: founder is advisor and strategist. 10 to 20 percent of founder time is GTM. Day-to-day is owned by a team.

Founders stuck in launch-stage behavior at scale starve their team of autonomy. Founders skipping launch-stage behavior too early under-invest in positioning and pay for it later.

AI in the growth stack: where it actually helps

We use Claude heavily inside the growth stack. Places it earns its keep:

• Research synthesis. Take 50 customer interview transcripts and extract recurring themes. Hours become minutes.
• ICP research. Given a company URL and LinkedIn profile, produce a 200-word brief on likely pain points. Not for spray-and-pray outbound - for a human reviewer.
• Positioning drafting. Claude is a strong sparring partner for positioning statements. Bad at deciding, good at generating options.
• Content drafts. Same discipline as SEO/GEO content - Claude drafts, humans edit.
• Competitive analysis. Scrape competitor sites and marketing materials; ask Claude to find the positioning gaps.

Places where AI-heavy growth stacks fail:

• Unedited bulk outbound. Detectable, gets flagged, burns the domain. See the outbound guide.
• Fully automated content machines. Google demotes them. Answer engines ignore them.
• Replacing the customer conversation. There is no AI substitute for 30 founder-led sales calls per week in launch stage.

A first-90-days plan by stage

Launch

• Weeks 1 to 4: positioning sprint. One sentence. Ten rewrites. Show to 20 ICP people. Iterate.
• Weeks 5 to 8: ship the launch essay, 3 case study drafts (even if fictional-but-representative placeholders), the home page, and the demo.
• Weeks 9 to 12: 30 founder-led conversations per week. Log every yes, every no, every reason.

Grow

• Weeks 1 to 4: pick two loops. Stand up lightweight attribution. Hire one growth operator if not already.
• Weeks 5 to 8: run both loops. Weekly cadence review. Weekly content shipping.
• Weeks 9 to 12: measure. Kill the underperformer. Double the investment in the survivor.

Scale

• Weeks 1 to 4: cohort analysis. Where is margin leaking? Which channel has quietly degraded?
• Weeks 5 to 8: concentrate spend. Kill marginal channels. Harden the operations for the top two loops.
• Weeks 9 to 12: run one clean experiment (new market, new ICP segment, new channel) with a clear stopping rule.

When to talk to us

Our marketing engagements match the stage. Positioning sprint for launch. Growth audit + two-loop build for grow. Conversion and cohort work for scale. Start a conversation if you want help picking.