A dashboard is built around outputs: signups, ARR, CAC, pipeline, shown as observed numbers over time. It's for monitoring. When a number moves, your job is to notice it and work out why. It doesn't need a cause-and-effect model, because the data is simply what happened. Dashboards can be interactive, but their filters only slice data that already exists.
A lever board is built around inputs you control, such as channel budgets, pricing or sales headcount. You move one, and the board recalculates the expected effect on the outputs. That makes it a decision tool rather than a reporting tool. It only works if there's a causal model behind it, for example how many signups an extra 1 000 EUR spent on creators actually brings in.
This page is the second kind. The company is invented and so are its numbers; the method is not. Every lever shows how its effect was measured, how wide the range is, and which of the fourteen studies stands behind it. The model runs here in your browser; nothing is sent anywhere until you ask Claude to read a move.
Built-in numbers, as of 15 October 2026.
Nine inputs the company controls. Each one carries a grade: tested means the effect was measured in a holdout, attributed means a platform reported it and the board counts a share, assumed means history and judgement. The red tick is the plan.
| # | Lever | ARR added, EUR | Per EUR of cost | Payback | Why |
|---|
Self-serve: a signup becomes an active user when the first workflow runs, and some active users pay for Starter or Pro within 90 days. Sales-led: opportunities come from the field, from events and from the product itself, and sellers can only work so many.
| Step | Plan | This board |
|---|
| Step | Plan | This board |
|---|
The board gives the number and the range. It does not say whether to believe them. Claude reads the moves and the model behind them and writes what would have to be true, the biggest risk and the cheapest test that would narrow the range. If the move holds up, propose it: the finance partner gets an email with one button, and an approved move is written to the budget sheet.
One n8n run and one Claude call. The sliders cost nothing.
Move a lever first, then ask.
One n8n run, one email to the finance partner, and a row in the sheet once approved. In production the budget owner does this; here anyone can, once per visit.
A lever board is only as good as its causal model. This is the model, lever by lever: how the effect was measured, the parameters with their ranges, and the published research behind the rule. Every number is invented for this page; the shape of the evidence is what a real one would need.
How it was measured: Geo holdouts on brand and non-brand terms, Q1 2026.
| Spend in the plan | 125 000 EUR a month |
| Result at that spend | 4 000 signups a month |
| Response to more spend | elasticity 0.35 (range 0.15 to 0.5): ten percent more spend brings about 3.5 percent more |
| Activation | 34 percent of these signups run a first workflow |
| Lag | results land in the month of the spend |
Studies: Consumer Heterogeneity and Paid Search Effectiveness: A Large-Scale Field Experiment, The Unfavorable Economics of Measuring the Returns to Advertising.
In production this lever reads from the ad platform and product analytics.
How it was measured: Platform attribution, cut to 40 percent until a holdout runs.
| Spend in the plan | 155 000 EUR a month |
| Result at that spend | 5 000 signups a month |
| Response to more spend | elasticity 0.5 (range 0.3 to 0.7): ten percent more spend brings about 5.0 percent more |
| Activation | 28 percent of these signups run a first workflow |
| Lag | results land in the month of the spend |
Studies: Close Enough? A Large-Scale Exploration of Non-Experimental Approaches to Advertising Measurement, The Unfavorable Economics of Measuring the Returns to Advertising.
In production this lever reads from the ad platform, product analytics and the holdout results.
How it was measured: Promo codes and a dedicated link per creator.
| Spend in the plan | 105 000 EUR a month |
| Result at that spend | 6 000 signups a month |
| Response to more spend | elasticity 0.75 (range 0.6 to 0.9): ten percent more spend brings about 7.5 percent more |
| Activation | 42 percent of these signups run a first workflow |
| Lag | results land in the month of the spend |
Studies: Revenue Generation Through Influencer Marketing.
In production this lever reads from the creator codes in billing and product analytics.
How it was measured: Registration lists and ambassador referrals; no holdout is possible.
| Spend in the plan | 140 000 EUR a month |
| Result at that spend | 3 000 signups a month, 25 qualified opportunities a month, of which 2 250 signups and 19 opportunities land inside the year |
| Response to more spend | elasticity 0.55 (range 0.3 to 0.8): ten percent more spend brings about 5.5 percent more |
| Activation | 45 percent of these signups run a first workflow |
| Lag | 3 months before the results start |
Studies: The 5 Principles of Growth in B2B Marketing, Meta-Analysis of Advertising Effectiveness: New Insights from Improved Bias Corrections.
In production this lever reads from the events platform, the community platform and the sales system.
How it was measured: Traffic to pages and templates built this year.
| Spend in the plan | 85 000 EUR a month |
| Result at that spend | 3 000 signups a month, of which 2 000 signups and 0 opportunities land inside the year |
| Response to more spend | elasticity 0.4 (range 0.2 to 0.6): ten percent more spend brings about 4.0 percent more |
| Activation | 40 percent of these signups run a first workflow |
| Lag | 4 months before the results start |
Studies: Meta-Analysis of Advertising Effectiveness: New Insights from Improved Bias Corrections.
In production this lever reads from web analytics and the template library.
How it was measured: Opportunity source in the sales system.
| Spend in the plan | 110 000 EUR a month |
| Result at that spend | 40 qualified opportunities a month, of which 0 signups and 37 opportunities land inside the year |
| Response to more spend | elasticity 0.6 (range 0.4 to 0.8): ten percent more spend brings about 6.0 percent more |
| Activation | not a signup channel |
| Lag | 1 months before the results start |
Studies: The State of Go-to-Market in 2026, 2026 B2B Marketing Budget and Performance Benchmark Report.
In production this lever reads from the sales system.
How it was measured: no price test has run; the elasticity is a judgement with a wide range, which is why the board shows ranges that can cross zero.
| Price in the plan | 24 EUR a month, plus 12 percent from executions above the plan |
| Who takes it | 65 percent of the activated users who pay, at the plan price |
| Price response | elasticity 0.9 (range 0.6 to 1.2): ten percent more on the price loses about 9 percent of the new accounts |
| How long an account stays | 14 months on average; a cohort keeps exp(minus age over 14) of its accounts |
Studies: The SaaS Conversion Report, 2026 SaaS and AI Metrics Benchmarks.
In production this lever reads from billing and product analytics.
How it was measured: no price test has run; the elasticity is a judgement with a wide range, which is why the board shows ranges that can cross zero.
| Price in the plan | 60 EUR a month, plus 12 percent from executions above the plan |
| Who takes it | 35 percent of the activated users who pay, at the plan price |
| Price response | elasticity 0.5 (range 0.3 to 0.8): ten percent more on the price loses about 5 percent of the new accounts |
| How long an account stays | 24 months on average; a cohort keeps exp(minus age over 24) of its accounts |
Studies: The SaaS Conversion Report, 2026 SaaS and AI Metrics Benchmarks.
In production this lever reads from billing and product analytics.
How it was measured: the sales system, two years of closed deals.
| Sellers in the plan | 10, each working 10 opportunities a month |
| Cost | 200 000 EUR a year per seller, fully loaded |
| Ramp | a new seller reaches full capacity after 4 months, so the year counts 8 of 12 |
| Win rate | 25 (range 20 to 30) percent of worked opportunities |
| Deal size | 36 000 EUR of ARR, closing 3 months after the opportunity |
| Where opportunities come from | 40 a month from field programmes, 25 from events, and 0.4 percent (range 0.3 to 0.5) of activated users at large companies |
Studies: The State of Go-to-Market in 2026, 2026 SaaS and AI Metrics Benchmarks.
In production this lever reads from the sales system and the HR system.
| Assumption | Value | In production, read from |
|---|---|---|
| Organic signups | 39 000 a month, not moved by any lever: word of mouth, templates, the free edition | product analytics |
| Organic activation | 36 percent run a first workflow | product analytics |
| Paid within 90 days | 10 (range 8 to 12) percent of activated users, about 3.6 percent of signups | billing |
| Plan mix | 65 percent Starter, 35 percent Pro, at the plan prices | billing |
| Gross margin | 80 percent, used for payback | the accounts |
| Horizon | 12 months from 15 October 2026; results that land later are not counted, which understates lagged channels | |
| Company today | 60M EUR of ARR, 400 employees, 1.2M active users | the accounts and product analytics |
| Active users | today plus every user activated in the twelve months; churn of existing users is not modelled | product analytics |
| Revenue per employee | ARR today plus ARR added, over employees today plus sellers added | the accounts and the HR system |
Benchmarks next to the outputs: payback median 16 months and best quarter 6 months (2026 SaaS and AI Metrics Benchmarks); free to paid 4 to 6 percent of signups normal, 10 to 15 strong (The SaaS Conversion Report).
One page, one sheet, three n8n workflows and Claude. The sliders never leave your browser, so the board itself costs nothing to run. The sheet holds every parameter above with its range, so finance changes a number without touching the page.
A webhook reads the two sheet tabs, LeverBoard (every parameter with its range) and LeverEvidence (the method, the evidence lines, the studies), and answers with the model as JSON. The page asks for it on load; the edge keeps the answer for an hour, so this costs at most one run an hour. If it fails, the page uses its built-in copy.
A webhook takes the lever positions, the plan and the board outputs, the bridge and the ranked next moves. Claude writes what has to be true, the biggest risk and the first test, as JSON, and the page shows it. One run and one Claude call per click.
A webhook takes the move and a name. Claude writes a short memo, the page gets its answer at once, and the workflow then waits: it emails the finance partner the memo with Approve and Decline buttons. The decision is written to the Moves tab of the sheet with the move, the expected effect and the range. One run, one email.
The same method, applied to one marketing action at a time: Profit or Burn. How that site is built: how it is built. The research: the studies.