A client needs to diagnose where prospects or users are dropping out of a conversion funnel, compare performance against benchmarks, and prioritize optimization efforts by expected impact.

---

How it works

1. You provide the client name, funnel stages with volume data, benchmarks, and context
2. The skill maps the funnel with conversion rates at each stage, diagnoses drop-off patterns, compares against benchmarks, and identifies friction points
3. It returns a funnel diagnostic with optimization recommendations ranked by expected impact and a test backlog Kate can present to the client

---

Prompt

You are building a conversion funnel analysis for a Kate Makrigiannis consulting engagement. Kate uses this to help clients see where they are losing prospects, quantify the cost of each drop-off, and focus optimization on the highest-leverage stages. Before writing, read knowledge/voice-tone-guide.md -- use the client-facing voice.

Inputs I will provide:

• Client: {{CLIENT}} (company name, product, business model)
• Funnel stages: {{FUNNEL_STAGES}} (ordered list of stages with volume at each stage -- e.g., "Visitors: 50,000 → Signups: 2,500 → Activated: 800 → Paid: 200 → Retained 90d: 120")
• Time period: {{TIME_PERIOD}} (the window these numbers cover -- e.g., "Q4 2025" or "last 30 days")
• Benchmarks (optional): {{BENCHMARKS}} (industry benchmarks, historical data, or competitor estimates for comparison)
• Context (optional): {{CONTEXT}} (known friction points, recent changes, business goals, team capacity for experiments)

Step 1: Map the funnel

Build a complete stage-by-stage funnel table. If the client provides raw volume, compute all conversion rates yourself. Show the math.

Funnel Map

Stage	Volume	Stage Conversion Rate	Cumulative Conversion	Drop-off Volume
[Stage 1]	X	--	100% (top of funnel)	--
[Stage 2]	X	X% (= Stage 2 / Stage 1 x 100)	X%	X lost
[Stage 3]	X	X%	X%	X lost
[Stage 4]	X	X%	X%	X lost
[Stage 5]	X	X%	X%	X lost

Show the math for every conversion rate. For example: "Signup → Activated: 800 / 2,500 = 32.0%"

Funnel Visualization

[Stage 1]  ████████████████████████████████████████  X (100%)
[Stage 2]  ██████████████████                        X (X%)
[Stage 3]  ████████                                  X (X%)
[Stage 4]  ████                                      X (X%)
[Stage 5]  ███                                       X (X%)

Step 2: Benchmark comparison

Compare each stage conversion rate against benchmarks. If the client provided benchmarks, use those. If not, use industry medians from the client's business model (SaaS, e-commerce, marketplace, etc.) and flag: [ASSUMPTION: using {{industry}} median benchmarks -- validate with client's historical data or published research]

Benchmark Comparison

Stage Transition	Client Rate	Benchmark	Gap	Verdict
[Stage 1 → 2]	X%	X%	+/-X pts	[Above / At / Below / Critical]
[Stage 2 → 3]	X%	X%	+/-X pts	[Above / At / Below / Critical]
[Stage 3 → 4]	X%	X%	+/-X pts	[Above / At / Below / Critical]
[Stage 4 → 5]	X%	X%	+/-X pts	[Above / At / Below / Critical]

Verdict criteria:

• Above: Client rate exceeds benchmark by 20%+. Low optimization priority.
• At: Within 20% of benchmark. Maintain, minor improvements possible.
• Below: More than 20% below benchmark. Active optimization opportunity.
• Critical: More than 50% below benchmark or absolute rate below viability threshold (e.g., <1% visitor-to-signup in SaaS). Urgent fix needed.

Step 3: Drop-off diagnosis

For each stage transition rated Below or Critical, diagnose likely friction causes:

Drop-off Diagnosis

For each underperforming transition:

[Stage X] → [Stage Y]: X% conversion (benchmark: X%)

• Volume lost: X prospects per {{TIME_PERIOD}}
• Revenue impact: If these lost prospects converted at benchmark rate, the client would gain approximately X additional customers, worth ~$X in revenue. Show the math: "(Benchmark rate - Current rate) x Stage X volume x revenue per customer = $X"
• Likely friction points:
  1. [Friction cause with explanation -- e.g., "Registration requires 6 form fields. Industry best practice for SaaS trial signup is 3-4 fields."]
  2. [Friction cause]
  3. [Friction cause]
• Evidence strength: [Data-backed / Hypothesis based on patterns / Needs investigation]

Step 4: Optimization recommendations

Rank recommendations by expected impact. For each recommendation, estimate the conversion lift and downstream revenue impact.

Optimization Recommendations

Priority	Recommendation	Target Stage	Expected Lift	Revenue Impact	Effort	Confidence
1	[Specific action]	[Stage X → Y]	+X pts (X% → X%)	~$X/{{TIME_PERIOD}}	[Low / Med / High]	[High / Medium / Low]
2	[Specific action]	[Stage X → Y]	+X pts	~$X/{{TIME_PERIOD}}	[Low / Med / High]	[High / Medium / Low]
3	[Specific action]	[Stage X → Y]	+X pts	~$X/{{TIME_PERIOD}}	[Low / Med / High]	[High / Medium / Low]

Show the revenue impact math: "If Stage 3 → 4 conversion improves from 25% to 30%, that produces X additional customers x $X average revenue = $X/{{TIME_PERIOD}}"

Prioritization logic:

• Priority 1: Highest revenue impact per effort. Usually the stage with the largest absolute drop-off and a clear, low-effort fix.
• Priority 2-3: Strong revenue impact but may require more effort or have lower confidence.
• Lower priority: Stages already at or above benchmark, or fixes that require significant engineering or design investment with uncertain payoff.

Step 4b: Statistical validation of conversion differences

Before prioritizing optimizations, validate whether observed conversion differences are statistically meaningful:

Statistical Confidence on Funnel Metrics

Stage transition	Conversion rate	Volume	95% CI	Is the gap vs. benchmark significant?
[Stage X → Y]	X%	X	[lower%, upper%]	(Yes -- CI doesn't overlap benchmark / No -- could be noise)

Confidence interval formula for a conversion rate: CI = p +/- 1.96 x sqrt(p x (1-p) / n), where p = conversion rate and n = sample size.

Example: 32% conversion on 2,500 users = 32% +/- 1.96 x sqrt(0.32 x 0.68 / 2500) = [30.2%, 33.8%]. If the benchmark is 35%, this gap is real.

When comparing segments or time periods: Use a Z-test for two proportions to determine whether a conversion rate difference is statistically significant before attributing it to a cause. A stage showing 31% this month vs. 33% last month on 1,000 users per period is likely noise (p > 0.05). The same difference on 50,000 users per period is likely real.

Multi-path funnel consideration: If users can skip stages or take alternate routes (e.g., direct signup bypassing the landing page), note the alternate paths and their volume. A "drop-off" at Stage 2 may actually be users entering at Stage 3 through a different channel. Account for this before diagnosing friction.

Related skills: For choosing the right statistical test for your data type, use /statistical-test-selector. For deeper experiment design on recommended tests, use /experiment-design.

Step 5: Test backlog

Convert the top recommendations into testable experiments.

Test Backlog

For each top-priority recommendation:

Test #	Hypothesis	Metric	Current Baseline	Target	Test Design	Duration	Traffic Needed
1	"If we [change], then [stage] conversion will improve by X pts because [reason]"	[Stage conversion rate]	X%	X%	[A/B test / Before-after / Cohort comparison]	[X weeks]	[X visitors/users needed for statistical significance]
2	"If we [change]..."	...	...	...	...	...	...

Sample size note: For A/B tests, estimate minimum sample per variant using: n = 16 x p x (1 - p) / (MDE)^2, where p = current conversion rate and MDE = minimum detectable effect (the lift you want to detect). Show the calculation.

Step 6: Compound impact model

Show what happens if the top 3 optimizations all succeed:

Compound Impact Projection

Stage	Current Volume	Projected Volume	Change
[Stage 1]	X	X	-- (top of funnel, unchanged)
[Stage 2]	X	X	+X
[Stage 3]	X	X	+X
[Stage 4]	X	X	+X
[Stage 5]	X	X	+X

Bottom-line impact: "If these three optimizations achieve their expected lift, the client moves from X to X customers per {{TIME_PERIOD}}, an increase of X% and approximately $X in additional revenue."

Kate's Talking Points

• The headline finding: which stage is the biggest leak and what it is costing
• The quick win: the lowest-effort, highest-impact recommendation
• The strategic question: whether the client should optimize the existing funnel or rethink the funnel structure entirely (e.g., skip a stage, change the entry point)

For defining the North Star Metric this funnel should drive, use /north-star-metric. For post-campaign funnel analysis, use /campaign-post-mortem. For building the dashboard to track these metrics ongoing, use /analytics-dashboard-design.

---

Landing Page Visitors  ████████████████████████████████████████  18,400 (100%)
Trial Signups          ███                                         1,290 (7.0%)
Activated              █▍                                            387 (2.1%)
Converted to Paid      ▍                                              89 (0.5%)
Retained 90d           ▎                                              61 (0.3%)