Fix a leaky funnel

A funnel is a chain: visit → click → signup → activation → habit → purchase → retention → expansion. Money leaks at exactly one or two joints, and every joint has a different owner. The costliest mistake is to redesign the landing page when the real leak is activation, or to build a referral loop when the real leak is that trials never see value. This playbook makes you locate the leak with evidence, then hand off to the one discipline that plugs it.

Governing rule: measure the break point before you touch anything. The stage with the worst conversion relative to its benchmark is the leak. Fix that stage, re-measure, then move to the next-worst. Never fix two stages at once, or you will not know which change worked.

When to use this

• Traffic or spend is steady, but signups, activations, or revenue are flat.
• The user describes a gap between two stages ("sign up but don't stick", "trials don't convert").
• Growth plateaued and nobody can point to the stage that broke.
• CAC looks acceptable but the business is not growing, so the leak is downstream of acquisition.
• The user is about to "fix the funnel" by rebuilding one page on a hunch.

When NOT to use this (reach for instead)

• The problem is upstream: wrong market, wrong channel, no product-market fit. Then the funnel

is not leaking, it was never a funnel. Use strategy-fundamentals first.

• The user already knows the exact leak and just wants the fix. Skip the diagnosis and go

straight to the matching discipline below.

• There is no traffic yet. You cannot diagnose a leak with no water in the pipe. Use

launch-gtm or the acquisition disciplines (paid-acquisition, seo-content, organic-social) to get first users, then return here.

Step 1: Find the break point (always first)

Pull analytics-data. Do not skip to a fix.

Get from it:

• The one metric that matters for this business and the stage-by-stage conversion map

(visit → click → signup → activation → retention → paid).

• Each stage's conversion rate versus a sane benchmark, so the worst relative performer stands out.
• Whether the drop is real or a measurement artifact (broken event, attribution mismatch,

reported ROAS that does not match real growth). Fix instrumentation before diagnosing behavior.

• If it is a multi-step on-site flow, a belief-and-energy read of where intent dies, not just where

clicks stop.

Output of Step 1: one named leak stage (or at most two), stated as a conversion gap with a number. That single finding decides which branch below you take.

Step 2: Route to the owner of the leak

Match the leak stage to its discipline. Take exactly one branch; return to Step 1 to re-measure after the fix.

Leak: visit or click → signup (the page does not convert)

Route to conversion-cro. Get from it: hero and above-the-fold fixes, the 7-section page skeleton, trust and friction levers, checkout or signup field surgery, and an A/B test that proves the lift. If the leak is specifically the words on the page, pair with copywriting-messaging; if it is why visitors resist, pair with marketing-psychology. For deeper page-level work, see the sibling improve-conversion playbook, which sequences all three.

Leak: signup → activation (people sign up but never reach value)

Route to product-led-growth. Get from it: the activation metric and aha moment definition, time-to-value compression, onboarding and empty-state design, and setup friction removal. This is the most common misdiagnosed leak: teams blame the landing page when users are signing up fine and dying in onboarding.

Leak: activation → retention / habit (users try it once, then fade)

Route to retention-lifecycle. Get from it: onboarding-to-habit loops, streaks and daily-usage design, churn and failed-payment recovery, win-back and post-purchase flows, and lifecycle tooling. If the retention leak is about users not inviting others or the loop not compounding, pull product-led-growth for referral and viral-loop mechanics too.

Leak: visit or trial → buy (people engage but do not pay)

Route to pricing-monetization. Get from it: the value metric, tier structure, anchoring and display psychology, trial-vs-freemium model choice, and discount/add-on levers. A "low conversion to paid" leak is usually a pricing-model problem wearing a conversion-page costume. Pair with marketing-psychology for the display and anchoring psychology, and with conversion-cro for the pricing-page layout itself.

Leak spans several stages / you are unsure it is a funnel problem at all

Escalate to strategy-fundamentals. A funnel that leaks everywhere is usually a product, market, or positioning problem, not a conversion problem. Diagnose the upstream cause before optimizing joints.

Step 3: Prove the fix, then move to the next leak

Return to analytics-data. Run a real experiment (adequate sample, holdout where possible) to confirm the fixed stage actually improved and that the gain is causal, not seasonal. Make the winner the new baseline. Then re-run Step 1: the next-worst stage is your next target. One leak at a time.

How to adapt

• B2B SaaS: the leak is usually signup → activation or trial → paid. Weight Steps toward

product-led-growth and pricing-monetization.

• Ecommerce / DTC: the leak is usually add-to-cart → purchase or first order → repeat. Add

ecommerce alongside conversion-cro (cart recovery, AOV, shipping) and retention-lifecycle (reorder, win-back).

• Content / newsletter / community: the leak is usually visit → subscribe or subscribe → active.

Pair conversion-cro with email-marketing and retention-lifecycle.

• Data too thin to diagnose: if analytics-data cannot see the stages, the first fix is

instrumentation. A leak you cannot measure, you cannot plug. Do not guess a branch; fix tracking.

• Two leaks look equally bad: fix the earlier-stage one first. Downstream stages inherit the

volume you fix upstream, so an early fix can resize the later problem.

From God of Skills: a curated, hand-tested directory of AI skills, prompts, templates and image style guides. Source: https://godofskills.com/skills/fix-a-leaky-funnel?ref=claude-skill