Conversion rate is the web's queen metric: it links your traffic to your results. It is also the most misused — compared against out-of-context benchmarks, computed on fuzzy denominators, optimised without diagnosis. Let's put things in order.
Quick Answer: what is a good conversion rate?
There is no universal "good" rate — only orders of magnitude per context:
- 1–3% for an e-commerce sale;
- 2–5% for a B2B quote;
- 20–40% for a trial sign-up from a dedicated landing page.
The calculation: conversions ÷ visitors × 100, on bot-free traffic. The three real levers, by impact:
- Traffic quality — analyse the rate per channel, never globally.
- Journey friction — located step by step through the funnel.
- Trust — proof, price transparency.
One measurement trap: if your analytics requires consent, your rates are computed on a biased population — the 50–70% who accept banners do not behave like the rest.
Definition and calculation
Conversion rate = conversions ÷ visitors × 100, over a given period.
Three clarifications that change everything:
- A conversion is what YOU define: sale, quote, sign-up, download, booked meeting. A non-merchant site has conversions just like a store.
- The denominator must be explicit: per unique visitor or per session? Both are defensible; consistency is what counts.
- Clean traffic: bots inflate the denominator and artificially crush the rate. Make sure your tool filters bots before comparing anything.
Benchmarks by industry: orders of magnitude
Handle with care — the definition of a conversion varies between studies:
| Context | Typical rate |
|---|---|
| General e-commerce (sale) | 1 – 3% |
| Niche / loyal e-commerce | 3 – 6% |
| B2B: quote/contact request | 2 – 5% |
| SaaS: free-trial sign-up (dedicated landing) | 20 – 40% |
| SaaS: trial → paid | 10 – 25% |
| Newsletter (popup or inline) | 1 – 5% |
The real question is not "am I average?" but "is my rate improving, and where does it leak?". A site at 1.2% that understands its funnel beats a site at 2.5% that doesn't know why.
The four factors that make the rate
- Traffic quality. The most powerful and most forgotten lever. 1,000 visitors from a transactional query beat 10,000 curious readers of a viral post. Analyse the rate per channel: that's often where the problem — and the opportunity — hides. AI traffic (ChatGPT, Perplexity) is a textbook case: low volume, often very high intent.
- The promise and its consistency. The ad, search snippet or post that earns the click must match the landing page. Any break (message, price, visual) is paid in bounces.
- Journey friction. Overlong forms, late-revealed fees, forced account creation, slow mobile pages. Every needless step takes its cut.
- Trust. Reviews, guarantees, clear legal pages, a transparent pricing page. A doubting visitor doesn't convert.
The 4-step diagnosis method
1. Build the funnel
Split the journey into measurable steps: arrival → product page → basket → details → payment → confirmation. The overall rate hides everything; step-through rates reveal the leak. One step at 70% abandonment when the others sit at 25%: there's your project.
2. Segment
By device (mobile often converts 2–3× less — but not everywhere), by acquisition channel, by country, by new/returning. A stable overall rate can hide a collapsing mobile offset by an improving desktop.
3. Understand the leak
On the leaking step: heatmaps to see what attracts or blocks, session replay to watch the abandonment in real conditions. It's the difference between "the basket loses 70%" and "visitors look for shipping costs, can't find them, open the FAQ and never come back".
4. Test the fix
Hypothesis → A/B test → decision on data. Then move to the next step. CRO is a cycle, not a project.
Improving the rate: the five highest-yield fixes
Once the diagnosis is set, the corrections that pay most often, across industries:
- Announce full costs as early as possible. Shipping, tax, options: every late-discovered cost is a documented abandonment driver. Early transparency loses a few lukewarm clicks and saves many baskets.
- Shorten the forms. Every removed field is measurable. Ask the minimum for the step; the rest after conversion — the optional registration-number field in our replay example was worth −34% abandonment.
- Allow purchase/enquiry without an account. Forced account creation is an emotional toll booth; offer it after conversion, once trust is established.
- Fix mobile speed. Beyond 3 seconds of loading, abandonment climbs steeply — and mobile is often the traffic majority but the conversion minority. The gap between the two is your headroom.
- Align page and acquisition promise. A click from a "−20%" ad must land on a page where the −20% is visible at a glance. Every continuity break gets paid for.
Tracking the rate over time: reading traps
A conversion rate reads as a trend, not a snapshot. Three precautions:
- Compare comparable periods: week vs week, not week vs sales weekend. Seasonality dwarfs everything else.
- Beware mix shifts. A campaign doubling cold traffic mechanically lowers the global rate without any page regressing — hence systematic per-channel reading.
- Tie the rate to value. A rising rate with a collapsing basket value (aggressive promo) can destroy margin. Track conversions × value, not the percentage alone.
Micro conversions: instrumenting weak signals
Between arrival and the macro conversion, instrument the milestones: pricing views, add to basket, form starts, contact clicks. These micro conversions serve three uses: locating the leak more finely than page views, detecting regressions early (form starts drop before the macro moves), and qualifying channels — a channel generating many pricing views but zero quotes tells a precise story: the audience is right, the offer or price isn't following.
In practice with Mirage — Conversions and funnels are defined in a few clicks (commercial goals or not), with step-through rates, segmentation by source/device/country, and the revenue generated when there is some. From a leaking step, open the matching sessions and launch the test — all in one tool, measured on 100% of traffic since there is no banner. Free 30-day trial.
The blind spot of banner-gated sites
Rarely mentioned: if your analytics requires consent, your conversion rates are computed on the 50–70% of visitors who accept. Refusers do not behave like accepters. Your internal benchmarks are biased in a way you cannot quantify. Consent-exempt measurement removes the bias: 100% of traffic, 100% of conversions.
FAQ
What is a good conversion rate for a website?
It depends on the conversion measured and the industry. Usual orders of magnitude: 1–3% for an e-commerce sale, 2–5% for a B2B quote request, 20–40% for a free-trial sign-up from a dedicated landing page. Your best benchmark remains your own history: aiming for +20% on your current rate is more useful than chasing an industry average.
How do you calculate a conversion rate?
Conversions divided by visitors (or sessions), times 100, over the same period. Always state the denominator: a per-session rate is mechanically lower than a per-visitor rate. What matters is consistency over time.
Why is my conversion rate falling while my traffic grows?
The classic diluted-traffic scenario: the new audience (campaign, viral article, off-target keywords) is less qualified. Segment the rate by acquisition channel: if every channel stays stable but the mix changes, the overall rate drops without any page having regressed.
What is the difference between micro and macro conversions?
The macro conversion is the final goal (purchase, quote, subscription). Micro conversions are the intermediate steps leading there: add to basket, account creation, pricing page view. Tracking micro conversions pinpoints exactly where the funnel leaks.