What is conversion optimization?
Conversion optimization (also called conversion rate optimization, or CRO for short) is the systematic process of getting more visitors to a website or app to take a desired action. That action is called a conversion. It can be a purchase, an inquiry, a sign-up, or a download. The foundation is understanding why people hesitate. Whether a change actually works is shown by an A/B test.
There are two kinds of conversions. A macro conversion is the actual goal, for example the order. Micro conversions are the steps before it: a click on a filter, adding an item to the cart, or starting a form. Any action that has value for your business can be measured and improved.
Conversion optimization is not a project with an end date. Customers, competitors, and offers change. That's why successful companies work in a permanent cycle of analysis, testing, and implementation.
Calculating the conversion rate
The conversion rate is the ratio of conversions to visitors. It shows how well a website turns its visitors into customers. The data comes from your web analytics tool, for example Google Analytics 4 or Adobe Analytics. The prerequisite is that conversions are tracked there as events and, ideally, linked to revenue.
Conversion rate (%) = conversions ÷ visitors × 100
Example: 9,000 orders from 300,000 visitors give you a conversion rate of 3%.
Today | After optimization | |
|---|---|---|
Visitors per month | 300,000 | 300,000 |
Conversion rate | 3% | 4% |
Orders | 9,000 | 12,000 |
Average order value | €100 | €100 |
Revenue | €900,000 | €1,200,000 |
Margin | 20% | 20% |
Profit (contribution margin 1) | €180,000 | €240,000 |
Worked example: the conversion rate rises by one percentage point, everything else stays the same.
In this example, one percentage point more means a third more orders. Profit rises by €60,000 a month, or €720,000 a year. Not a single additional visitor was needed for that. That is exactly what makes conversion optimization attractive when advertising gets more expensive every year.
What is a good conversion rate?
The 2025 Conversion Rate Report from uptain puts the average for small and midsize German online shops at 2.01%. It is based on more than 3,000 shops from early 2023 to early 2025. On smartphones, the rate is 0.61 percentage points below desktop.
Averages like this are only a rough guide. Assortment, price, brand, and traffic source shift the value strongly. A furniture shop at 1% can be doing better than a drugstore at 4%. The more honest comparison is your own conversion rate, broken down by device, channel, and new versus existing customers.
The conversion rate isn't everything
A higher conversion rate can still cost money. An aggressive discount raises orders and lowers margin. A change that creates more bad-fit purchases drives up returns. So measure contribution margin, return rate, and customer lifetime value alongside the conversion rate. Only then do you know whether a test really won.
Where conversion optimization is used
You can optimize anywhere people make a decision online. The most common fields of use are these:
Online shops: product pages, cart, checkout, cross-selling and upselling
Lead generation: inquiry forms, appointment booking, calculators, and configurators
Closing funnels: plans, loans, insurance, and travel bookings
Registration and retention: customer account, newsletter, app onboarding, and repeat purchases
The conversion optimization process in six steps
Every company works a little differently. But you'll find the following six steps in every program that works.
1. Analysis: understanding where and why customers drop off
Web analytics shows where visitors leave. It doesn't show why. For that, you need qualitative methods like user interviews, usability tests, surveys, or an analysis of customer service inquiries. The combination of both sides is called mixed methods. A framework like our 7 Levels of Conversion helps organize the causes systematically: from the relevance of the offer through trust to usability.
2. Forming and prioritizing hypotheses
Every problem you find becomes a hypothesis. A good hypothesis states the observation, the planned change, and the expected effect. An example: Many visitors abandon checkout because shipping costs only appear at the end. If we show them in the cart already, the number of completed orders goes up. Then you prioritize by expected value contribution, effort, and required test duration.
3. Concept and design
For every hypothesis, a rough concept comes first, for example as a wireframe. At our company, a testing board of several specialists reviews every concept. The most important question is: Why doesn't the current page solve the customers' problem, and what does the new variant do differently? Only then is the variant designed in the corporate design.
4. Implementation and quality assurance
The variant is programmed in the testing tool, client-side or server-side. Before the start, you check it on all important devices and browsers. A bug in the variant distorts the result and, in the worst case, costs real revenue.
5. Testing and evaluating
In an A/B test, one part of the visitors sees the current page, the other part sees the variant. Set the duration beforehand, and don't look at the result every day so that you could stop the test at the first green number. Otherwise you declare chance hits winners. A test should run for full weeks at the least, so that weekdays and weekends are equally represented. At the end, you check statistical significance and effect size and document what you learned.
6. Rollout and learnings
Winners are built in permanently and become the new baseline for the next test. Losers are just as valuable. They show which assumption about your customers was wrong. New hypotheses come out of the learnings, and the cycle starts over.
Tactics that raise the conversion rate
Tactics fall into two groups. Hygiene factors are only noticed when they're missing: load time, error-free forms, a working search. Consumer psychology tactics work on motivation. They make an offer more relevant, more credible, or easier to decide on. For that, we use proven Behavior Patterns such as social proof or loss aversion.
Forms: fewer required fields, fitting input types, clear error messages
Landing pages: a concrete value proposition matching the ad
Product pages: copy that answers questions, good images, visible reviews
Cart and checkout: early cost transparency, guest checkout, familiar payment methods
Trust: seals, guarantees, real customer voices, and an accessible service
Personalization: content by customer type, more at Personalization
None of these tactics works everywhere. What wins in one shop loses in the next. That's why analysis belongs before every tactic and a test after every tactic.
Examples from practice
Four programs from our work show how different conversion optimization can look. Every number is measured against a control.
The Female Company (D2C): Three A/B tests per week, the first after 13 days. After three months, the conversion rate was 54% higher cumulatively. 59% of the tests won.
Haufe (B2B software): Personas and jobs to be done led to 33 tests with a 52% success rate. They brought in €5.2 million in additional revenue.
Pferdesporthaus Loesdau (equestrian sports): As an external growth team, we quadrupled the number of tests. After two years, the ROI was 1,493%.
hessnatur (fashion): Personalization by customer type. All three of the first experiences beat the control, with more than 10% higher revenue per visitor after four weeks.
AI in conversion optimization
Since 2023, AI has noticeably sped up conversion optimization. It clusters thousands of customer comments in minutes. It drafts hypotheses, writes copy variants, and helps program tests. That lets the same team implement significantly more tests.
AI doesn't replace the test. A language model doesn't know how your customers react to a change. It only knows patterns from other people's data. The bottleneck therefore shifts from implementation to the question of what is worth testing. Customer understanding counts more than ever here.
Common mistakes
Copying best practices instead of understanding your own customers
Stopping tests too early, as soon as the result looks good
Testing changes that are too small, whose effect your traffic can never detect
Extrapolating uplifts to twelve months even though effects fade over time
Measuring only clicks instead of revenue, margin, and returns
Do it yourself or with an agency?
Many companies start with an internal team and a testing tool. That works as long as there is enough research, design, development, and statistics knowledge in-house. If one of them is missing, the hit rate drops after the first quick winners. An agency can take over the program or build up your team so it can carry on itself. What to look for in a good agency and what it costs is covered in the guide on finding a CRO agency. How we do it is on our page on conversion optimization as a service.
Books for further reading
André Morys: Die digitale Wachstumsstrategie. 10 Prinzipien für ein profitables Online-Geschäft (in German)
Stefan Thomke: Experimentation Works. The Surprising Power of Business Experiments
Ron Kohavi, Diane Tang, Ya Xu: Trustworthy Online Controlled Experiments. A Practical Guide to A/B Testing
Steve Krug: Don't Make Me Think. A Common Sense Approach to Web Usability