The essentials
- Week 1: four readings that locate the loss, by device, by source, by landing page and by funnel step.
- Weeks 2 to 6: the high-return fixes, mobile and speed first, form and proof next.
- Weeks 7 to 12: tests on high-traffic pages, one change at a time.
- Expected gain: 10 to 30% more conversions in three months on a site that has never been optimised, without a single extra euro of acquisition.
Improving the conversion rate is the marketing action with the best return for a small business: it benefits all existing traffic, whatever its source, and its effect stays acquired. The principles of the discipline are set out in CRO: definition ; this article offers a concrete ninety-day plan, applicable without a large testing budget and without a redesign.
Week 1: the four diagnostic readings
- By device: compare the mobile and desktop conversion rates. A gap of more than 50% makes mobile the first job, which is the case on most small business sites.
- By source: the conversion rate of organic, paid, direct and social traffic. A source well below the others signals a mismatch between the promise and the landing page.
- By landing page: the ten pages that receive the most traffic, with their conversion rate. That is where most of it plays out.
- By funnel step: the pass-through rate between each step, from product page to order or from form to submission; the method is in analysing a conversion funnel with GA4.
These four readings take half a day in GA4 and almost always point to an obvious job that intuition would not have identified. Complete them with a full run through the funnel on your own phone, in real conditions: the obstacles you meet are worth every audit.
Weeks 2 to 6: the high-return fixes
| Area | Action | Observed gain | Effort |
|---|---|---|---|
| Speed | Compress images, defer third-party scripts, check the server with PageSpeed Insights | 5 to 15% per second saved | 1 to 3 days |
| Mobile | Accessible buttons, adapted form, text readable without zooming | 10 to 30% on mobile traffic | 2 to 5 days |
| Form | Cut down to four fields, remove the compulsory account creation | 10 to 25% | 1 day |
| Clarity of the offer | A headline that says what you sell and to whom, from the first screen | 5 to 20% | 1 day |
| Proof | Reviews, references and guarantees placed at the moment of doubt | 5 to 15% | 1 to 2 days |
| Objections | Fees, lead times, commitment and security addressed explicitly | 5 to 15% | 1 day |
| Consistency with the ad | A dedicated page for each advertising promise | Up to 100% on paid traffic | 2 to 4 days |
These fixes require no prior testing: they are defects, not hypotheses. The format of a dedicated page is detailed in landing page.
Weeks 7 to 12: the tests
Once the defects are fixed come the open questions: which headline, which offer to lead with, which order of blocks, which price to display. These questions are settled by a test, provided you have the volume needed, that is several hundred conversions per variant to conclude. Below that, it is better to put the version judged best into production and compare periods, taking the season into account. Three rules: one change at a time, a duration of at least two weeks to cover a full cycle, and a hypothesis written before launch, naming the supposed cause and the expected effect.
Measuring the gains
- The overall conversion rate: read at a comparable season, never from one month to the next in a seasonal business.
- The rate by device: to check that the mobile work has produced its effect.
- The cost per acquisition: it falls mechanically when conversion rises, at constant budget; that is the argument that speaks to a management team.
- The number of conversions: the final indicator, because a rate can rise simply because traffic has fallen.
- Lead quality: a shortened form can increase volume and reduce qualification; the check is done in the CRM, as noted in MQL and SQL.
What does not work
- Universal tricks: button colour, countdown timers, aggressive pop-ups; no effect or a negative one in the medium term.
- The complete redesign: everything changes at once, nothing can be attributed any more, and organic traffic often suffers.
- Optimising poorly targeted traffic: no page converts a visitor who came for something else; the problem is then in the campaign.
- Testing without volume: concluding on thirty conversions amounts to tossing a coin.
- Copying a competitor: their page is suited to their offer, their price and their reputation, not yours.
How GreenRed helps
Rather than juggling several tools, GreenRed's Site health module brings these metrics together in a single dashboard, compares them over time and tells you which actions come first. You can try it free, with no card, from the Pricing.
Frequently asked questions
How do you improve a site's conversion rate?
In ninety days: a week of diagnosis by device, source, landing page and funnel step; five weeks of high-return fixes (speed, mobile, form, clarity, proof, objections); six weeks of testing on high-traffic pages, one change at a time.
What gain should you expect?
From 10 to 30% more conversions in three months on a site that has never been optimised, with no extra acquisition spend. The mobile work generally produces the first gain, followed by shortening the form and by consistency between the ad and the landing page.
Where should you start?
With the comparison between the mobile and desktop conversion rates, then a full run through the funnel on your own phone in real conditions. The obstacles you meet point to the first fixes, which need no prior testing since they are defects.
Do you need A/B tests?
Only once the defects are fixed, and provided you have several hundred conversions per variant. Below that volume, it is better to put the version judged best into production and compare the periods before and after, taking the season into account.