Run · Chapter 09

Growth loops

Acquisition experiments that turn measured outcomes into business learning.

A growth experiment should resolve an uncertainty the business can act on. Who needs the offer? What helps them understand it? Which channel can reach them at a cost the business can sustain? Choose one question and connect the experiment to the buyer evidence gathered in Discovery.

Set an audience, an approved offer and a meaningful conversion event. A click, a reply and a purchase describe different outcomes. Decide what result would change your next move, how much you can spend learning it and who owns the decision to stop.

Build one experiment you can interpret

Start with outbound, paid creative or content. Changing audience, offer and creative together makes the cause of a result difficult to understand. Corey Haines’s marketing skills can help structure the work, but installing a method does not connect a campaign or establish performance.

For a web experiment, a tool such as PostHog can record the chosen conversion. Verify that event in the intended project before launch. For outbound, a tool such as Instantly can prepare a sequence; review the recipients, claims, personalization and unsubscribe handling, then send a test to an owned inbox. Keep drafts separate from live actions.

Bring the result back into the business

Launch only within the agreed audience, approvals and limits. Check exclusions again before retries. With small samples, an honest experiment ledger may tell you more than a weak claim of statistical significance. Follow the result through to customer response, cost and contribution.

Use the evidence to propose the next change. Expand a campaign or enter a new niche when its results justify that decision, checking the audience and claims again. A growth loop becomes useful when acquisition teaches the business something that changes what it does next.

Tools and setup

Practitioner: Corey Haines's Marketing Skills provides reusable marketing procedures. It is a method library, not connected execution or evidence of campaign performance.

  1. Establish product/audience context, approved claims and one experiment metric. Review and install only the relevant marketing skill in Skills and playbooks; check current names before selecting it. Produce draft messages or creative briefs; use the chosen media tool only after rights and required inputs are clear.
  2. For a web conversion test, install PostHog using the guide for the actual platform; configure consent/data collection appropriately and verify a test conversion in the intended project. Follow its experiment guide: set hypothesis, variants, allocation and metrics; save a draft, exercise variants in testing, then approve launch separately. For small volumes, a simple experiment ledger can be more useful than an underpowered A/B claim.
  3. For outbound, Instantly's setup guide covers creating a campaign, importing leads, writing sequences, mapping variables, adding unsubscribe links, previewing and testing. Use only authorized contacts and a permitted sending account; check suppression before launch and retries. Keep the campaign unlaunched until recipients, copy, limits and tracking are approved. Verify test mail in an owned inbox before any prospect send.
  4. If choosing paid ads, start in the chosen platform's draft campaign UI. This manuscript does not verify Meta API app review, permissions or publishing endpoints; re-research those before writing an API setup. Do not claim paid-ad execution is connected by installing a marketing skill.

Setup paths checked against linked documentation on 15 November 2026. Verify current account support during setup; reader integrations have not been tested here.

Work through this with your agent

Paste the request into your assistant. It will interview you and help produce the working artifact.

Interview me to build ONE acquisition experiment, not every channel. Ask at most three questions per turn and wait. Ask about buyer, approved offer/claims, channel, consent and outreach requirements, brand voice, asset rights, conversion event, attribution, budget and owner. Choose outbound, paid creative or content reuse based on evidence. Draft the relevant prospect criteria/message, creative variants or case study using authorized sources; protect customer confidentiality. Define review, suppression/opt-out handling, tracking, spend/rate limits, stop criteria and who may approve changes. Rehearse on a small permitted sample. Save GROWTH-RUNBOOK.md, reviewed assets and an experiment record. Obtain approval of actual recipients/assets and spending limits before sending or launching. Verify action results; distinguish drafts from live campaigns. Do not scale or expand into a new niche until measured results justify it and context/claims are revalidated. Return proposed learning, never silently rewrite company facts.

Use these documented setup paths as references; check current account support before acting:
https://github.com/coreyhaines31/marketingskills
https://posthog.com/docs/getting-started/install
https://posthog.com/docs/experiments/start-here
https://help.instantly.ai/en/articles/6386134-how-to-create-campaigns-complete-setup-guide

The Hypothesis Diff · weekly

Know when a hypothesis is stressed or changes.

One email a week: which hypotheses hold, which are stressed, which changed, and the tools worth trying. Nobody else publishes what stopped being true.