A single site-wide figure hides the thing you need to act on. Splitting by device, by source, by new versus returning, and by landing page turns one uninformative number into four that point somewhere.
Why the overall number misleads
A site-wide conversion rate is an average across groups that behave completely differently.
Somebody arriving from a search for your business name is nearly ready to call. Somebody arriving from a general article is not. Averaging them produces a figure describing neither, and a change in that figure can mean the mix shifted rather than anything improving.
The same applies to speed, bounce, and time on page. Every site-wide metric is a blend, and the blend is what moves.
The four splits that pay
Mobile against desktop
The most revealing split for most small businesses, and frequently the most uncomfortable.
Typically most visitors are on phones and the conversion rate there is lower, sometimes considerably. That gap is a design and speed problem rather than a fact of life, and it is invisible in the combined figure.
By traffic source
Search, direct, referral, social, and paid behave differently enough that comparing them against each other is more useful than watching the total.
The finding that matters: which source produces enquiries rather than which produces visitors. Those are frequently not the same source.
New against returning
Returning visitors convert at a much higher rate almost everywhere, which is expected and still worth seeing.
The useful reading is the proportion. A site where nearly everybody is new is not building any relationship, and a site with a healthy returning share has content that brings people back.
By landing page
Where somebody arrived, rather than which pages they viewed. This tells you which pages are doing the acquiring, which is usually a small number and rarely the homepage.
It also finds the pages receiving traffic and converting nobody, which are the candidates for attention.
The splits worth adding for a local business
Two more that matter here and less elsewhere.
By location. Whether visitors are inside your service area at all. A site with substantial traffic from outside the region has a targeting problem, and the conversion rate is being dragged down by people who could never buy.
By time and day. When enquiries actually arrive, which tells you when somebody needs to be available to answer and when advertising should run.
The location split in particular changes how the whole account reads. A business appearing to convert poorly is frequently converting fine among people it can serve.
Combining two at once
Where the actual insight usually is.
Mobile visitors from paid search. New visitors landing on a specific service page. Returning visitors from the local area. Each of those is a specific group whose behaviour you can act on.
The caution is that combining splits divides the data, and small businesses run out of numbers quickly. Two dimensions is usually the practical limit before the counts become too small to mean anything.
What to do with what you find
Each split points at a different kind of fix.
A mobile gap points at speed, layout, or a form that is awkward on a phone. A source producing visits and no enquiries points at either the targeting or the page they land on. A landing page with traffic and no conversions points at that page. Traffic from outside the area points at targeting.
That specificity is the whole reason to segment. An overall figure tells you something is wrong and a segment tells you where.
Setting it up once
Most of this exists by default and needs finding rather than building.
Device, source, new against returning, landing page, and location are standard reports in any analytics tool. The work is knowing to look at them rather than at the summary.
Saving the four or five views you actually use, so they are one click rather than a rebuild each time, is what makes this a monthly habit rather than an occasional project.
The one to check first
Mobile conversion against desktop, this month.
For most small businesses that single comparison identifies the largest available improvement, because the majority of visitors are on phones and the site was almost certainly designed and tested on a desktop.
It also has to be read carefully when the numbers are small, since a gap between two small figures can be noise rather than a finding, as covered in reading data when the numbers are small.
Frequently asked questions
Why is a site-wide conversion rate misleading?
It averages groups that behave completely differently. A change in the figure can mean the mix of visitors shifted rather than anything actually improving.
Which split matters most?
Mobile against desktop. Most visitors are on phones, the rate there is usually lower, and that gap is a design and speed problem invisible in the combined number.
What should I look for by traffic source?
Which source produces enquiries rather than which produces visitors. Those are frequently not the same source.
Why segment by landing page?
It shows which pages are doing the acquiring, usually a small number and rarely the homepage, and finds pages receiving traffic while converting nobody.
What extra split helps a local business?
Location. Traffic from outside your service area drags the conversion rate down with people who could never buy, and it makes a fine site look like it is failing.
Can I combine splits?
Two dimensions is usually the practical limit for a small business, since each combination divides the data and the counts become too small to mean anything.
West Coast Media Solutions Inc. provides web design, web development, hosting, digital marketing, and business consulting to organisations across Canada, drawing on more than twenty-five years in the field.
Watching one conversion number that never explains anything?
We set up the four views that point somewhere, starting with mobile against desktop, which is usually where the largest gap is.
Start a Conversation