Conversion rate is the first number anyone blames and the last one worth changing.
01 · The question as it arrives
“Our conversion rate is bad.” It usually arrives with a benchmark attached — a figure from an industry report, or from whoever produced the last audit.
The proposal is already written by the time the question is asked. Redesign the site.
It is rarely the marketing team’s own question. It usually arrives after someone senior saw a figure and asked what is being done about it, which means the answer has to be visible as well as correct.
A redesign has one quality that makes it hard to argue against. It is obviously an action.
02 · Why the obvious answer fails
Conversion rate is a ratio. A redesign changes one side of it. It changes the page and leaves untouched the question of who arrives at the page.
The benchmark makes it worse. An average conversion rate for an industry averages businesses with different traffic mixes, different price points and different buying cycles. A company selling a two-thousand-dollar service and a company selling a two-million-dollar one appear inside the same figure.
A redesign also removes the only baseline there was. After it ships you cannot say which of forty changes did anything, so the next argument about conversion starts with no evidence at all.
The cost is months, and the risk is rebuilding the part that was working.
Underneath all of it is a definition problem. Conversion often counts several different events at once — a contact form, a download, a newsletter signup, a phone tap on mobile — and a change in the mix moves the rate while nothing about the business changes. A rate that includes downloads improves when you publish more downloads. Nobody bought anything.
03 · What we look at instead
Split the traffic before judging the rate. Three groups is usually enough: people who searched for the company by name, people who searched for the problem, and people who arrived from a campaign.
If branded search converts and the rest does not, the site works. The traffic is wrong, and the money belongs upstream in who is being reached rather than in the page.
If branded search does not convert, the site is the problem. Someone arrived with the highest intent available and still left. That is worth months; the other case is not.
Then find where they leave. A drop between the page and the form is a page problem. A drop inside the form is a form problem. They do not have the same fix and they are routinely treated as one.
And check what is being counted as a conversion. Separate the events. One number for enquiries that a person has to answer, everything else reported apart from it. Most conversion arguments end here, because the two lines move in opposite directions and only one of them was ever the business.
04 · Where to start
Begin with a split rather than a redesign. Look at the last ninety days of conversion for branded search on its own, and for everything else.
If branded search converts well and the rest does not, the site is doing its job, and the effort belongs in who is being reached.
If both are low, the page itself needs work, and you now have a clear baseline to measure the change against.
Either way, record today’s numbers before anything changes. Without a baseline, no redesign can be shown to have worked, and the same argument returns a year later.