I85PREP
⌘ K
OMCP study materials
OMCP study resources

OMCP: Conversion Rate Optimization

OMCP · Work through your exam, one topic at a time.

All guide sections

Conversion Rate Optimization

Conversion Rate Optimization on the OMCP exam tests experimentation and funnel improvement: how to diagnose friction, form testable hypotheses, run valid A/B tests, and decide when a result is trustworthy. Expect questions on funnel analysis, landing pages, forms, sample size, statistical significance, segmentation, personalization, and QA.

Core rules and concepts

  • Define the primary conversion before testing; distinguish macro conversions (purchase, lead) from micro conversions (scroll, add-to-cart).
  • Write hypotheses in a testable form: because [evidence], changing [element] will [effect] for [segment], measured by [metric].
  • Use a control, random assignment, one primary variable, and a pre-registered metric. Calculate sample size from baseline rate, minimum detectable effect, alpha (often 0.05), and power (often 0.8).
  • Check validity: sample ratio mismatch, instrumentation QA, seasonality, novelty effects, and pre-specified segments only.
  • Avoid early peeking and stopping at the first significant result; use a fixed horizon or a valid sequential method.
  • Prioritize high-traffic friction first; optimize the full funnel, not just the button color.

Worked example

A mobile checkout form has five fields. The team tests removing the company field. Baseline conversion is 2.1%, and a power analysis shows the test needs far more traffic than the two-week window produced. After one week the variant is up 0.3 percentage points, and someone wants to ship it. Reasoning: the test is underpowered and was peeked early, so the result is inconclusive. Run to the planned sample or use a sequential design, then judge practical significance.

Traps that cost marks

Stopping at first significance, changing multiple elements, ignoring sample ratio mismatch, optimizing micro conversions that do not affect revenue, segment fishing, and skipping QA.

Key points

  • A valid test needs a control, randomization, and a pre-registered metric.
  • Sample size depends on baseline, MDE, alpha, and power.
  • Early peeking inflates false positives.
  • Segment only as pre-specified.
  • Fix high-traffic friction before cosmetic tests.