Skip to content

The method

Hypothesis → Experiment → Proof → Scale. The same loop runs every channel we manage. Here's exactly how it works — including the rules we hold ourselves to.
Stages per loop
04
Deciding metric per test
01
Possible verdicts
03
Days in the weekly rhythm
05
01 HYPOTHESIS02 EXPERIMENT03 PROOF04 SCALEREPEAT↻
  • 01Hypothesis
  • 02Write down what we believe, and why
  • 03Experiment
  • 04Test small, test fast, test fairly
  • 05Proof
  • 06Keep what works. Kill what doesn't
  • 07Scale
  • 08Put budget behind evidence

[01]  The method

Four stages. No guesswork.

Every channel, offer and landing page starts as a hypothesis. We test it on a budget you can afford to lose, agree in advance what counts as proof, and only then put real money behind it.

  1. 01Hypothesis

    Write down what we believe, and why.

    We audit your funnel, data and market, then turn the biggest opportunities into specific, falsifiable statements — each with the metric that will prove or disprove it.

    • Acquisition audit
    • Ranked hypothesis backlog
    • Agreed success thresholds
    Fig. 03.1Hypothesis card · Example
    If
    we lead with a pricing calculator instead of a demo request
    Then
    qualified enquiries rise at the same cost per lead
    Because
    buyers want a price signal before they talk to sales
    Proof
    result clears the threshold agreed before launch

  2. 02Experiment

    Test small, test fast, test fairly.

    Each hypothesis gets a controlled test with a capped budget and a fixed run time. Nothing is judged on gut feel or a good week.

    • Test design and budget cap
    • Tracking verified before launch
    • Weekly test log
    Fig. 03.2Test log · Example
    Example test log
    IDTestStatus
    H-07Calculator offerRunning
    H-04Proof-led ad copyComplete
    H-09Two-step formQueued

  3. 03Proof

    Keep what works. Kill what doesn't.

    Results are read against the threshold we set before launch. Winners are documented with the evidence; losers are cut quickly so they stop costing you.

    • Plain-English readout
    • Decision log: scale, iterate or kill
    • Updated forecast
    Fig. 03.3Readout · Example
    THRESHOLDCONTROLVARIANT

    DecisionKeep → Scale

  4. 04Scale

    Put budget behind evidence.

    Proven plays get budget in staged steps, with guardrails on cost per lead so performance is protected as spend grows. Then the loop starts again.

    • Staged budget plan
    • Cost guardrails and alerts
    • Next hypothesis cycle
    Fig. 03.4Staged budget · Example
    CPL GUARDRAILSTEP 1STEP 2STEP 3

[02]  Anatomy of a test

What every test is made of.

Before a test spends anything, it's written down. Scroll through the six parts of a test card.

  1. 01ID

    A name and a number

    Every test gets an ID, an owner and a date, so results can be traced and nothing is quietly forgotten.

  2. 02Belief

    The hypothesis

    If we change X, then Y will happen, because Z. Written specifically enough to be proven wrong.

  3. 03Metric

    One deciding metric

    The single number that decides the test, chosen before launch — usually tied to qualified leads, not clicks.

  4. 04Pass mark

    The threshold

    The result that counts as a win, agreed with you in writing. No moving the goalposts afterwards.

  5. 05Budget

    A capped spend and run time

    The most the test can spend and how long it runs, so a wrong idea costs little and a result isn't cut short.

  6. 06Decision

    Scale, iterate or kill

    The outcome is logged with the evidence and the next step — including the tests that fail.

[03]  Decision rules

Three outcomes. Agreed in advance.

  • Scale

    When

    The result clears the pass mark over the full run time.

    Then

    Budget increases in staged steps with cost guardrails.

  • Iterate

    When

    Promising but below the pass mark, with a clear reason why.

    Then

    A sharper follow-up hypothesis goes into the backlog.

  • Kill

    When

    No meaningful effect, or it moves the wrong way.

    Then

    Switched off quickly. What we learned is written down.

[04]  The rhythm

A week inside the loop.

  1. Mon

    Plan

    Review the backlog and confirm which tests launch or continue this week.

  2. Tue

    Launch

    New tests go live once tracking is verified end to end.

  3. Wed

    Monitor

    Guardrails checked. Nothing is judged early — we only watch for breakage.

  4. Thu

    Learn

    Finished tests are read against their pass marks and logged.

  5. Fri

    Report

    You get the weekly test log: what ran, what we learned, what's next.

[05]  What we don't do

Things you won't hear from us.

  • Guaranteed results

    Nobody can promise a number before seeing your data. We promise the process.

  • Calling a test early

    A good first week isn't proof. Tests run their full length.

  • Vanity metrics

    Impressions and clicks are inputs. We report on qualified leads and revenue.

  • Hiding the losers

    Failed tests are reported as clearly as winners — they're how we learn.

[06]  Questions

The method: questions.

01How long does one loop take?

It depends on your traffic and budget. A single test usually needs a few weeks to reach a reliable answer, and several tests run in parallel. [PLACEHOLDER] Add your typical test length.

02Who decides what counts as proof?

You do, with us. We propose a pass mark for each test and you sign it off before any money is spent.

03What if most tests fail?

Many will — that's normal and cheap, because budgets are capped. A loss you understand narrows the search; one proven winner can be scaled for a long time.

Next step

Bring us your growth problem. Leave with a testable plan.

One focused call. We'll show you where your acquisition is leaking and the first experiments we'd run.

Book a Strategy Call

No obligation, no hard sell. If we're not the right fit, we'll say so on the call.

[PLACEHOLDER] One verified proof point — a short client quote or an approved result.
— [PLACEHOLDER] Name, CompanyPlaceholder