The standard reaction to weak sales is to increase the advertising budget. Sometimes that is the right move. More often the problem is not the number of visitors but what happens to them once they reach the site.
Simple arithmetic shows why. A site with 5000 monthly visitors converting at 1 percent produces 50 enquiries. To double that through traffic you need 5000 more visitors, which means doubling the ad spend. To double it through conversion you need to reach 2 percent. The second route is almost always cheaper.
This article is about finding the places where customers are lost, and about verifying whether a change genuinely helps.
What counts as a conversion
The first mistake is a vague definition. Conversion means something different for a store, a consultancy and a repair shop.
For an online store the primary conversion is a completed order, but the intermediate ones are worth tracking too: add to cart, checkout started, account created.
For services the primary conversion is a submitted form or a phone call. A common blind spot appears here: calls are not tracked and the whole channel looks weaker than it is.
For B2B with a long cycle, the site conversion is a document download, a demo request or a booked consultation. The sale happens later and off site.
Define one primary conversion and two or three intermediate ones. More than that dilutes focus. If you do not yet have correct tracking, start with Google Tag Manager and GA4.
Where visitors are lost
Before testing anything you need to know where the leak is. Four information sources each give a different angle.
Quantitative data shows where the drop off happens. The funnel report in GA4 shows how many people pass each step. If 800 people add to cart and 200 reach checkout, the problem sits between those two steps.
Session recordings and heatmaps show how people behave. Tools like Microsoft Clarity are free and reveal things no report shows: rage clicks on an element that is not a button, scrolling back and forth hunting for the price, filling in and clearing the same field.
Form analytics show which field people abandon. Often it is the phone number or the company registration field.
Conversations with customers give the reason. Five calls to people who enquired and did not order deliver more clarity than a month of analysis.
The most common leak points
The same issues repeat. The price is invisible or appears too late. Shipping terms are missing from the product page. The form asks for too many fields. On mobile the action button sits below the fold and requires scrolling. There is no clear indication of what happens after submitting.
How to run an A/B test properly
An A/B test splits traffic between two variants and measures which performs better. It sounds simple, yet most tests in practice prove nothing, because they run on too small a sample or get stopped the moment the result looks good.
The process has five steps.
First, a hypothesis. Not "let us try a green button" but "if we move the price above the fold, more visitors will reach the cart, because the data shows scrolling up and down in that zone".
Second, one variable. Change the headline, the image and the button at once and you will know variant B is better, but not why.
Third, a sample size decided in advance. Calculate how many conversions are needed for a difference of the expected size to be meaningful. For a low traffic site this often means A/B testing is simply the wrong tool.
Fourth, a fixed duration. At minimum two full weeks, so every weekday is covered. Behaviour on Monday morning and Saturday evening is not the same.
Fifth, a decision and documentation. Record the failed tests too. They are just as useful, because they prevent repetition.
When A/B testing is not appropriate
This is the section most often skipped.
Below roughly 1000 conversions a month, a test will rarely reach statistical significance in a reasonable time. With 50 enquiries a month, detecting a 10 percent improvement would take years.
When testing minor changes with a small expected effect. A different button shade will almost never produce a measurable difference at average traffic.
When making major structural changes. If you are rebuilding an entire product page, an A/B test is possible, but it is usually more sensible to ship the change and compare periods while accounting for seasonality.
For low traffic sites the better tools are a heuristic audit, user testing with five people, and consistent application of well established practices. Five observed sessions with real people uncover more problems than three months of an underpowered test.
Changes that usually pay off
These are based on recurring observations, but should always be verified in your context.
| Change | Typical effect | Applies to |
|---|---|---|
| Removing unnecessary form fields | Noticeable lift in submissions | Any site with enquiries |
| Showing shipping cost on the product page | Fewer abandoned carts | Online stores |
| Clear call to action above the fold on mobile | Higher conversion on mobile traffic | All |
| Adding reviews near the button | More trust on a first visit | Services and higher priced goods |
| Guest checkout without forced registration | Substantially less checkout drop off | Online stores |
| Specific headline instead of a generic message | Better match with the advert | Landing pages |
Note the last row. A mismatch between ad copy and page headline is among the most expensive mistakes in paid campaigns. The user clicks for something specific and lands on a generic page. There is more in our article on landing pages.
Mobile conversion is a separate problem
On most sites mobile traffic sits between 60 and 75 percent, yet mobile conversion is roughly half the desktop rate. Part of that gap is natural, because people browse on a phone and buy on a computer. But much of it is technical.
Check specifically on a phone: how many scrolls it takes to reach the action button, whether the keyboard covers the field being filled, whether numeric fields open a numeric keypad, whether buttons are large enough for a thumb, whether the popup can be closed.
If you have never done this review, the fastest win is probably sitting there. See also our article on mobile optimisation.
Organising the process over time
Conversion optimisation is not a project with an end but a cycle. A sensible rhythm is monthly: one week collecting data and observations, one week prioritising and preparing, two to four weeks executing and measuring.
Prioritise on three criteria: expected impact, confidence in the hypothesis and effort to implement. A change with high expected impact and low effort goes first, even at medium confidence.
Keep a log of what you learn. After a year that log is worth more than any external playbook, because it is about your audience.
WEBPROGRESS quotes conversion audits and optimisation individually, because scope depends on the type of site and the traffic available. You get an answer within 24 hours.
Frequently asked questions
What is a normal conversion rate
There is no universal figure, and comparisons with foreign benchmarks often mislead. For online stores typical values sit between 0.8 and 2.5 percent; for service sites with an enquiry form, between 2 and 6 percent. It is more useful to compare your site against itself over time than against an unfamiliar average.
How much traffic does a meaningful A/B test need
As a rough guide, detecting a 20 percent improvement on a 2 percent baseline needs around 3000 to 4000 visitors per variant. For smaller expected differences the numbers rise quickly. If you do not have that volume, use other methods.
Which tools are required
For measurement, GA4 plus Google Tag Manager is enough. For behaviour observation, Microsoft Clarity is free and entirely sufficient for most sites. For the testing itself there are paid platforms, but at low traffic they rarely pay for themselves.
Can optimisation harm SEO
Not if the test is implemented correctly. Use server side or client side splitting with a canonical pointing to the main variant, and never show different content to bots than to users. Problems arise from heavy scripts that slow loading, so keep an eye on speed.
Where should a company with no data start
With correct tracking and with Clarity. Spend the first two weeks just observing. Then list the obvious problems and fix them without testing, because most of them need no proof. Tests come after the obvious has been handled.
Conclusion
Raising conversion is not a set of tricks but systematic work: a clear definition of success, correct measurement, observation of real behaviour, prioritised changes and disciplined verification. On most sites the first 20 percent of improvement comes from fixing obvious things rather than from tests.
If you have traffic that does not turn into customers, write to us through contact. We audit the funnel, deliver a concrete prioritised list and quote individually within 24 hours. See also our services for web development, online stores and advertising.
