Skip to content

Marketing & Growth Systems

Conversion optimization: find the reason before you change the page

When a page doesn't bring in requests, the usual answer is a redesign. Conversion optimization starts with evidence instead: how far people scroll, what callers still ask, where forms are abandoned. Each piece becomes a hypothesis you can state in one sentence, and the most promising change goes to a test before anyone calls it a win.
Discuss your systemSee three hypotheses ↓
C-114 · water heater page · EastsideHeading and photoWhat's includedPrice rangeQuote form11 · 68% leave before the price rangePrice range in the first screen→ more quote requests22 · 4 in 10 calls ask: same day?Same-day answer under the heading→ more requests, fewer calls33 · mobile: photo step abandonedPhoto after the request is sent→ more mobile requestsalready in a testchosen changePrice range + same-day answer in the first screentest · 50/50 · 4 weeks · guardrail: job value

Evidence, and what it can't tell you

Each source answers one question well and another not at all. The useful hypotheses come from two sources agreeing.
Scroll depthHow far people getWhy they stopped
Calls and emailsWhat people still need to askHow many didn't bother to call
Form step drop-offWhich step loses peopleWhether they called instead
Session recordingsHow people actually use the pageWhether a few sessions are typical
Asking visitorsWhat they say they wantWhat they'd actually do

Three hypotheses from one page

Written before any change: what we saw, what we'll change, what should happen, how it's measured, and what mustn't get worse.

Price range

Observation
68% of visitors leave before the price range
Change
show the range in the first screen
Expect
more quote requests per visitor
Measured by
requests ÷ visitors, 4 weeks, split 50/50
Guardrail
average job value doesn't fall

What we won't build

Some changes lift a number and cost you the customer's trust. They're out, whatever a test says.

  1. 01

    False urgency

    “Only 2 slots left!”

    No countdowns or scarcity that isn't true. If slots really are short, say so plainly.

  2. 02

    Hidden costs

    Fees at the last step

    The price range shown is the price range. Anything extra appears beside it.

  3. 03

    Pre-ticked boxes

    Consent by default

    Permission to send messages is asked for, never assumed.

  4. 04

    Guilt buttons

    “No, I like cold showers”

    Saying no is a plain choice in plain words.

  5. 05

    Hard exits

    Easy in, impossible out

    Unsubscribing or cancelling takes no more steps than signing up.

What conversion optimization hands off

In

Behaviour
scroll, clicks, form steps
What customers ask
calls, emails, questions
The page
structure and copy
Conversion optimization
evidence → hypothesis → change

Out

Hypotheses
one sentence each, ranked
Changes
handed to a test
Learnings
kept, losers included

Worth doing when visits don't become requests

Good fit

  • Pages get visits but few requests.
  • You can see form steps and record calls.
  • Enough people visit to test a change in weeks, not years.

Not the right fit

  • A page gets a few dozen visits a month. Fix the obvious and move on.
  • Nothing is measured yet. Start with instrumentation.

Where conversion optimization is used

Next to conversion optimization

Questions about conversion optimization

Isn't this just A/B testing?

Testing is the last step. Most of the work is finding a good reason to change something, so the test is worth running.

How much traffic do we need?

Enough that a test finishes in weeks. With less, fix clear problems directly and save tests for the big questions.

Will you redesign the page?

Only if the evidence says the structure is the problem. Usually a few specific changes come first.

What if the test loses?

That's a result. The hypothesis was wrong, it's written down, and nobody suggests the same change next year.