agentleFS
reatlatreatlat/fullstory-claude-plugin/skills/funnel-doctor/SKILL.md

funnel-doctor

Skill36 starsChanged 16 days agoMITView on GitHub

Conversion funnel analysis — find where users drop off, quantify the loss, and diagnose why. Use when asking about conversion rates, funnel abandonment, checkout completion, signup flow performance, or "where are we losing users.

The file

We split this skill into the rules it gives an agent

10 rules. About 0 are common, 7 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈7

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If the user doesn't specify steps, ask for the start and end point.

    line 33
  2. Very specific:

    If a funnel metric already exists for this flow, search first: fullstory:get_metric(regex="checkout funnel").

    line 41
  3. Very specific:

    Call fullstory:compute_metric(metric_id) — or fullstory:compute_funnel(funnel_id) for funnel-specific options.

    line 45
  4. Distinctive:

    Call out the step with the biggest absolute drop AND the biggest percentage drop — they're often the same, but not always.

    line 78
  5. Distinctive:

    If the user asks about a specific cohort (mobile users, enterprise, new vs returning), build a segment and attach it:

    line 82
  6. Very specific:

    Call fullstory:get_sessions(metric_id=metric_id) filtered to sessions that reached but didn't complete the problem step

    line 96
  7. Distinctive:

    If applicable, estimated revenue impact (user count × average order value)

    line 108
  8. Distinctive:

    Always show both absolute drop (user count) and percentage drop at each step.

    line 112
  9. Distinctive:

    If a funnel step has a very high drop-off, prioritize investigating that step — don't spend equal time on every step.

    line 114
  10. Distinctive:

    If the user asks "did the new design help?", build two funnel metrics (before/after deploy date) and present side by side.

    line 116

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Funnel Doctor

Diagnose conversion funnels — find the leakiest step, quantify the loss, and show you the sessions that explain why users leave.

When to Use

  • "Where are users dropping off in our checkout flow?"
  • "What's the conversion rate from signup to first purchase?"
  • "Compare the old onboarding flow vs the new one — did we improve completion?"
  • "Show me sessions where users abandoned at the payment step"
  • "What's the biggest drop-off in our app? Quantify the lost revenue."

Mental Model

A funnel is a sequence of steps users must complete. The goal is to find the step where the most users leave — not just by absolute count, but by the percentage drop from the previous step.

Example: 10,000 land on /checkout → 8,000 enter shipping → 6,000 reach payment → 2,000 complete purchase. The biggest absolute drop is payment (6,000 → 2,000 = 4,000 lost). The biggest percentage drop is also payment (67% drop). Both matter — present both.

Workflow

Step 1: Define the funnel steps

Clarify what steps the user cares about. Typical funnels:

  • E-commerce: product page → add to cart → checkout → shipping → payment → confirmation
  • SaaS signup: landing → signup form → email verification → onboarding → first action
  • App flow: home → search → results → detail → booking

If the user doesn't specify steps, ask for the start and end point. You can infer the middle steps from page navigation data.

Step 2: Build a funnel metric

Call fullstory:build_metric with:

  • query describing the funnel steps in order (e.g., "users who viewed /checkout, then reached shipping, then reached payment, then completed purchase")
  • output_type="funnel" — this tells the builder to create a sequential funnel, not a simple count

If a funnel metric already exists for this flow, search first: fullstory:get_metric(regex="checkout funnel"). For funnels built with build_funnel, use fullstory:get_funnel(id) — it returns name, description, rendered definition, and correct num_steps (needed for the get_funnel_sessions workflow).

Step 3: Compute and analyze

Call fullstory:compute_metric(metric_id) — or fullstory:compute_funnel(funnel_id) for funnel-specific options.

Period comparison — add compare_to to see prior-period deltas:

fullstory:compute_funnel(funnel_id, compare_to="previous_period")

Each step gets a previous object with the prior-period count and conversion rate. Use this for "did the new design help?" questions instead of building two separate funnels.

Dimension breakdown — add dimension to split by a property:

fullstory:compute_funnel(funnel_id, dimension="device_type")

Each step gets a groups array broken down by that dimension (browser, country, device, custom var, etc.). Replaces manual segment-per-group workflows.

Present results as a step-by-step breakdown:

Checkout Funnel (last 30 days)
1. /checkout            10,000 users  (100%)
2. Shipping form         8,000 users  (80% — lost 2,000)
3. Payment page          6,000 users  (75% of remaining — lost 2,000)
4. Confirmation          2,000 users  (33% of remaining — lost 4,000) ← BIGGEST DROP
Overall conversion: 20%

With compare_to, append deltas:

4. Confirmation          2,000 users  (33% of remaining — lost 4,000)
   Previous period:      1,400 users  (28% of remaining)
   Delta:               +600 users    (+5pp conversion)  ↑

With dimension, show per-group conversion at each step.

Call out the step with the biggest absolute drop AND the biggest percentage drop — they're often the same, but not always.

Step 4: Scope with segments

If the user asks about a specific cohort (mobile users, enterprise, new vs returning), build a segment and attach it:

fullstory:build_segment("mobile users") → segment_id
fullstory:update_metric(metric_id, segment_id)
fullstory:compute_metric(metric_id)

For A/B comparisons of funnels (old flow vs new, mobile vs desktop), use the comparisons skill to pick the right mechanism (dimensionality vs segments).

Step 5: Investigate the worst step

Once you've identified the leakiest step, find out why:

  1. Call fullstory:get_sessions(metric_id=metric_id) filtered to sessions that reached but didn't complete the problem step
  2. Load 3-5 sessions through the session-context agent with focused tasks:
    • "What did the user do on the payment page before leaving? Did they encounter an error?"
    • "Was the form validation blocking them? What fields were highlighted?"
    • "Did they rage-click or dead-click anything before abandoning?"

Step 6: Report

Synthesize everything:

  • The funnel numbers (table format)
  • Which step is the biggest problem (with both absolute and percentage loss)
  • What the session evidence shows about why users leave at that step
  • If applicable, estimated revenue impact (user count × average order value)

Guidelines

  • Always show both absolute drop (user count) and percentage drop at each step.
  • Surface the metric_url so the user can verify in the Fullstory UI.
  • If a funnel step has a very high drop-off, prioritize investigating that step — don't spend equal time on every step.
  • For revenue impact, ask the user for average order value. Don't guess.
  • If the user asks "did the new design help?", build two funnel metrics (before/after deploy date) and present side by side.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.