Conversion funnel calculator: TOFU/MOFU/BOFU, drop-off and the ROI of CRO
In serious digital marketing — CMOs, heads of growth, performance agencies, in-house e-commerce and SaaS teams — the conversion funnel stopped being a conceptual diagram and became a quantitative model with measurable stages, drop-off percentages, session-to-lead and statistical MDE. Unbounce, HubSpot, CXL Institute and WordStream publish industry benchmarks that define what a healthy funnel looks like versus a broken one.
A correct calculator solves four equations:
Per-stage conversion rate = Users in stage N+1 / Users in stage N × 100
Drop-off % = 1 − Conversion rate
Session-to-lead % = Leads / Sessions × 100
ROI of improving stage X = (ΔConv × AOV × Sessions) / Experiment cost
TOFU / MOFU / BOFU: the segmentation that decides CRO ROI
TOFU (Top of Funnel): visitor-to-lead. Landing pages, ad clicks, form fills. Typical B2B rates 2-8%, B2C 1-4%. MOFU (Middle of Funnel): lead-to-qualified. Email nurture, demo requests, trial signups. B2B SaaS rates 15-35%. BOFU (Bottom of Funnel): qualified-to-customer. Demo-to-close, trial-to-paid. SaaS rates 15-30%, e-commerce cart-to-purchase 5-15%.
The classic mistake is measuring only global conversion (visitor-to-customer). A funnel with 0.8% global may have healthy TOFU (5%) and broken BOFU (16%), or broken TOFU (1.5%) with healthy BOFU (55%). CRO experimentation must prioritize the stage with the largest monetary impact (drop-off × AOV × volume), not the stage with the worst isolated number.
Critical drop-off: where the lead dies
Multi-step forms lose 10-18% per additional field above 4 (Baymard Institute). E-commerce checkouts lose a global average of 69.8% (Baymard 2024), distributed across: 48% from unexpected extra costs (shipping, tax), 24% from mandatory account creation, 17% from complicated processes, 11% from errors or slow pages. The calculator models each cause with its expected lift to prioritize the test queue.
CRO and MDE: when a test is statistically conclusive
An A/B test requires a Minimum Detectable Effect (MDE) computed with 80% power and 95% significance. To detect a 5% lift on a 3% baseline with a 50/50 split, you need roughly 15,500 visitors per variant (per CXL calculator). Tests with insufficient sample size produce false positives — the most expensive CRO mistake, because decisions end up driven by noise. The calculator integrates a sample-size module to close this loop.
Multi-step form optimization: the underrated B2B lever
For B2B lead generation, splitting a 9-field form into three steps (3+3+3) lifts conversion 28-42% per Unbounce Conversion Benchmark Report 2024. Mechanism: reduces cognitive load, leverages commitment bias (started → finish), and enables progressive disclosure of sensitive fields (phone, company size) at the end. The calculator multiplies this lift by volume and AOV to quantify incremental revenue.
Session-to-lead by industry: real benchmarks
Unbounce Conversion Benchmark Report 2024 publishes medians by industry: SaaS 3.0%, Finance 5.1%, Legal 7.4%, Real Estate 2.6%, E-commerce 1.8%, Travel 2.4%, Health 3.3%. A B2B SaaS funnel at 0.9% session-to-lead carries enormous structural margin; an e-commerce at 2.5% is already above median and marginal CRO ROI decreases. The first strategic decision is to locate yourself in the competitive quartile: bottom quartile leaves clear headroom; top quartile forces the team to seek lift further down the funnel or in lead quality, not in aggregate session-to-lead.
Mobile vs desktop: the gap no one can ignore anymore
Baymard and HubSpot document a persistent gap: mobile conversion is typically 60-70% of desktop in e-commerce and 45-60% on B2B forms. Mobile today represents 58-72% of traffic depending on sector, but only 38-52% of revenue. Identifying whether the problem is responsive design, page speed (LCP above 2.5 seconds destroys conversion), input friction (keyboards, autocomplete) or digital payment methods (Apple Pay, Google Pay, Shop Pay, Venmo) shifts the test queue priority. The calculator decomposes session-to-lead by device to expose where revenue drop-off truly sits.
Multi-touch attribution and the real ROI per stage
The average B2B buyer has 6-8 brand interactions before converting (Gartner). Attributing the sale to last-click overestimates BOFU and underestimates educational TOFU content. Multi-touch attribution models — linear, time-decay, position-based, GA4 data-driven — redistribute conversion credit. The calculator allows running each model against the same dataset to expose how perceived channel ROI shifts with attribution method.
Micro-conversions: the metric that leads the macro
Scroll depth at 75%, video watch above 50%, clicks on pricing comparator, TOFU ebook download — micro-conversions correlate strongly with final conversion and enable experimentation with much smaller samples. CXL Institute recommends identifying 3-5 predictive micro-conversions and optimizing against them when final-conversion volume does not reach statistical MDE. The calculator maps historical micro → macro correlation to calibrate which micro-conversion is worth optimizing.
Attribution models: why the same data tells four different stories
The same marketing data produces radically different channel ROI depending on the attribution model applied:
- Last-click: 100% credit to the last touchpoint before conversion. Overweights BOFU channels (branded search, retargeting). Used as the default in Google Ads.
- First-click: 100% credit to the first touchpoint. Overweights TOFU discovery channels (organic, display). Misrepresents the role of closers.
- Linear: equal credit across all touchpoints. Directionally fair but mathematically blunt.
- Time-decay: more credit to touchpoints closer to conversion. A middle ground; reflects the urgency recency creates.
- Data-driven (GA4 Conversion Modelling): machine-learning attribution based on real conversion path data. Requires minimum 600 conversions and 3,000 clicks over 30 days. The most accurate but the least portable — it is locked to the platform.
Implication for the funnel calculator: running the same MQL or SQL data under last-click vs data-driven typically moves perceived channel ROI by 20-45%. A team optimizing to last-click without knowing data-driven attribution is cutting TOFU channels that are actually the demand-generation engine.
AARRR framework: the startup lens on funnel
Dave McClure's AARRR (Acquisition, Activation, Retention, Referral, Revenue) framework organizes the funnel beyond the purchase event:
- Acquisition: how do users find you? Channel, campaign, organic. Measured by sessions, new users, CPL.
- Activation: do they have a 'happy first experience'? For SaaS: reach the Aha moment (first project created, first report run, first connection). For e-commerce: first purchase. Benchmark: 30-day activation rate 15-35% for PLG SaaS.
- Retention: do they come back? For SaaS: DAU/WAU/MAU ratios, churn rate. For e-commerce: repeat purchase rate (healthy: 25-40% repeat within 90 days). The leakiest buckets in most digital businesses.
- Referral: do they tell others? NPS, referral program participation, viral coefficient. Dropbox's referral program is the canonical case: 2-way referral (giver + receiver both get storage) drove 3,900% user growth without paid acquisition.
- Revenue: do they pay? ARPU, LTV, expansion revenue.
The AARRR frame prevents the common error of optimizing Acquisition (traffic) while ignoring Retention (churn) — which produces growth that leaks like a bucket.
2026 SaaS funnel benchmarks
Based on OpenView SaaS Benchmarks 2025 and Klipfolio benchmarks:
| Stage | Median | Top Quartile |
|---|---|---|
| Website visitor → MQL | 2.5% | 5.0% |
| MQL → SQL | 35% | 55% |
| SQL → Demo/Trial | 60% | 75% |
| Demo → Closed-Won | 20% | 35% |
| Overall visitor → Customer | 0.3-0.6% | 1.0-1.5% |
For a SaaS with 50,000 monthly website visitors at median rates: 1,250 MQLs → 437 SQLs → 262 demos → 52 new customers/month. At $5,000 ACV, this yields $260K new ARR/month. Lifting SQL-to-demo from 60% to 75% adds 13 demos/month → ~3 customers → $15K ARR/month. The simulator shows which stage improvement generates the highest marginal ARR per hour of work.
Conclusion
The conversion funnel is not decorative; it is the growth team's financial model. Segmenting TOFU/MOFU/BOFU, computing per-stage drop-off with opportunity cost, prioritizing experiments by monetary impact, closing each test with statistically valid MDE, and calibrating multi-touch attribution with predictive micro-conversions are the five practices that separate a data-driven growth team from one running A/B tests on intuition. The calculator is the layer where those practices become explicit and defensible to finance.