What is conversion rate optimization?
Conversion rate optimization is the process of improving how often eligible people complete a defined action, such as creating an account or reaching a useful result. It includes finding obstacles, understanding why they occur, proposing changes, and evaluating what happens.
Optimizely's overview covers conversion goals, analysis, hypotheses, and testing as parts of the process. For a SaaS, improving a form is useful when it helps the right people complete the workflow. A higher signup count can still be a poor result if users misunderstand the offer or encounter more failures afterward.
This guide uses an original fictional reporting app, ClearReport. The example examines a journey from an appropriate first visit to signup, email verification, a delivered report, and a paid subscription. The counts are synthetic and are not ShowMeYourSaaS results or conversion benchmarks.
Define the conversion and its denominator
Write the unit and eligibility rule before calculating a percentage. A visitor-to-signup rate uses visitors as the denominator. A form-completion rate uses people who started the form. These answer different questions and should carry different labels.
For the fictional workbook, one eligible visitor maps to at most one new account. It excludes staff and test traffic, assigns device group at the first visit, and follows the September 1–7 first-visit cohort for thirty days per visitor. Signup through first delivery is measured within fourteen days; paid adoption is measured within thirty.
A real product may involve several devices, shared accounts, blocked tracking, and unmatched anonymous visits. Document how those cases affect measurement. If the numerator is account-based and the denominator is browser-based, explain that limitation instead of implying an exact person-level rate.
Our sales funnel guide explains stage definitions. The product analytics guide helps verify the underlying events and identities before you interpret the ratios.
Read the synthetic signup funnel
ClearReport's fictional cohort contains 1,000 eligible visitors. Of these, 120 start signup, 80 create an account, 60 verify an email, and 30 deliver a first report within fourteen days. Twelve start a paid subscription within thirty days.
Visitor-to-account creation is 80 / 1,000 = 8%. Signup form completion is 80 / 120, approximately 66.7%. Verification among created accounts is 60 / 80 = 75%. First delivery among verified accounts is 30 / 60 = 50%. Paid adoption among all visitors is 12 / 1,000 = 1.2%.
These rates locate questions, but they do not tell you why someone left. A visitor might be ineligible, prefer another approach, encounter an error, or intend to return later. Preserve the cohort boundaries when splitting the data so a later-stage numerator remains a subset of the relevant earlier group.
| Stage | Desktop | Mobile | Total |
|---|---|---|---|
| Eligible visitors | 600 | 400 | 1,000 |
| Started signup within 14 days | 72 | 48 | 120 |
| Created account within 14 days | 60 | 20 | 80 |
| Verified email within 14 days | 48 | 12 | 60 |
| Delivered first report within 14 days | 24 | 6 | 30 |
| Started paid subscription within 30 days | 9 | 3 | 12 |
Investigate a specific mobile failure
The device split gives the founder a useful question. Desktop form completion is 60 / 72, approximately 83.3%; mobile completion is 20 / 48, approximately 41.7%. Device group is associated with a difference here, but that alone does not establish that the layout caused it.
In the fictional investigation, the founder attempts signup with synthetic values on a small screen and encounters a password validation message hidden behind the on-screen keyboard. The problem can be reproduced without accessing customer passwords or private form contents.
The proposed change keeps the field requirements visible and makes the validation message available beside the affected input. The founder also checks focus behavior and whether the message is announced to assistive technology. This directly addresses a reproducible problem, while the effect on completed accounts remains a separate measurement question.
Review the entire path after the fix. A person who can create an account must still understand email verification and deliver a useful report. Improving one local rate while increasing later failure would leave the customer's task unresolved.
Prioritize changes using evidence and guardrails
Keep each proposal tied to an observed problem, the affected group, and an expected mechanism. A reproducible broken control deserves a different response from an untested preference for a new button color. Estimate effort and uncertainty without pretending that a scoring formula removes judgment.
Choose guardrails that represent outcomes you do not want to damage. For ClearReport, these include verification success, first report delivery, support requests about signup, and account quality. If a shorter form admits people who cannot use the service, the initial conversion increase needs a fuller review.
Check offer clarity as well as interface behavior. The landing page guide helps communicate audience, workflow, price, and commitment. A clear limit may reduce unsuitable signups while improving the experience for appropriate buyers.
Keep the review aligned with informed choice. Hidden charges, unclear consent, or difficulty leaving a process can inflate a local metric while harming the customer's experience. Prefer a path that explains the commitment and lets people make a sensible decision.
Choose how to evaluate the change
Use direct functional verification for a reproducible defect: the message should be visible, understandable, and available with the relevant input on supported devices. Observe permitted participants attempting the task when comprehension remains uncertain.
A randomized experiment can help estimate the causal effect of a change when the design, traffic, outcome rate, and observation time support it. Define the unit, primary metric, guardrails, and analysis rule in advance. Our A/B testing guide covers sample-size planning and the risks of repeated peeking.
With low traffic, a small before-and-after comparison is usually weak evidence about causation. Keep the denominator and counts visible, inspect whether traffic sources or eligibility changed, and avoid announcing a winner from a few extra signups.
For the fictional mobile issue, the immediate decision can be to repair the reproducible validation problem, verify the task, and monitor the defined cohort metrics. The workbook deliberately includes no invented post-fix uplift.
Maintain a conversion decision log
Record the baseline, evidence, hypothesis, change, verification, outcome window, and review date. Include what you learned when a proposal did not help. This prevents the same weak idea from returning every time a team member sees a short-term decline.
Revisit a small number of meaningful journeys rather than continually changing every page. New acquisition channels, pricing terms, authentication options, or product workflows can change who arrives and what counts as a successful conversion.
Link the website and product view. A person who starts with an article, evaluates the offer, creates an account, and completes a report has progressed through several distinct decisions. Useful CRO makes that path easier to understand and complete, while keeping the measurements honest.
Download the conversion audit and workbook
The original audit includes an eligibility checklist, a completed mobile investigation, hypotheses, guardrails, and a blank decision log. The JSON workbook contains the synthetic device cohorts, stage definitions, and expected rates for checking the calculations. All example figures are fictional. Source checked October 27, 2026.