what a website actually brings in: measuring instead of counting projects

business · 2026-10-20 · 6 min

in parma, "CRO" usually means contract research organization. the pharmaceutical kind. so let me be explicit before the first paragraph does any work: this post is about conversion rate optimization, the practice of measuring what a website makes happen and changing one thing at a time until more of it happens.

i build websites in parma, italia, and i measure them. not because measurement is fashionable, but because "we delivered the project" is not a result. it is a delivery.

the metric ladder

a website is a chain, and most reports stop at the first link. the useful ladder, from top to bottom:

1. visits — how many people arrived, and from where. a number, not a strategy.
2. engaged visits — people who stayed long enough to read, not bounced in four seconds. still not money.
3. intent events — clicks on the phone number, the contact button, the pricing link. this is where a site starts doing something.
4. form starts — someone opened the form. a drop here is a form problem, not a traffic problem.
5. form completions — someone sent it. a drop between 4 and 5 is usually friction: too many fields, a validation error, a required field nobody has.
6. replies — a human answered. this is where the site hands over to the business.
7. closed work — the quote became a job. the only number that pays.

you cannot optimize link 7 by looking at link 1. most "the site does not work" conversations are actually a broken link 4 or a slow link 6, and nobody was measuring either.

what to instrument, honestly

• one analytics counter. privacy-light is fine. the point is a consistent definition of a visit, not the logo on the dashboard.
• one conversion event per real action: form sent, phone clicked, email clicked, booking completed. not "scrolled 50%" dressed as a conversion.
• form starts and form errors, if the form tool exposes them. they are the cheapest diagnosis you will ever get.
• the source: search, direct, referral, social. without it you cannot tell whether a change worked or the season changed.
• page speed on the pages that convert, not only the homepage. a slow contact page is a conversion problem disguised as a technical one.

that is five things. a spreadsheet can hold them, and a monthly reminder can read them.

the number nobody puts in the proposal

cost per lead. take what the site costs to build and run for a year, divide by the number of real inquiries it produced in that year. do the same for the previous year. now you have a sentence that survives a meeting: "the site costs X and brought Y inquiries, so each inquiry cost Z".

it is not a perfect number. it ignores the inquiries that would have arrived anyway, and it ignores the value of the ones that closed. it is still infinitely more useful than "we delivered 12 projects".

if you want the fuller version, add the close rate: inquiries that became jobs, divided by inquiries. two numbers, and you can see whether the problem is the site or the sales conversation. most of the time it is one of the two, and the site gets blamed for both.

the loop i run

week 1: baseline. write down visits, intent events, form starts, form completions, replies. screenshot the pages. do not change anything.

week 2: one change. one. a clearer first paragraph, a shorter form, a phone number that is visible on mobile, a price range instead of "contact us". not a redesign.

weeks 3-4: wait. this is the part that separates measurement from theatre. a change needs enough traffic to show a difference, and a small site may need two months. if the traffic is tiny, measure the qualitative version: five people try the form, watch where they hesitate.

week 5: compare. same numbers, same definitions, same source. if nothing moved, the change was not the problem, or the traffic is too small to tell. both are useful answers.

then repeat with the next change. one at a time, or you learn nothing about which one worked.

what i ignore

• bounce rate alone. a page that answers a question in ten seconds has a high bounce rate and is doing its job.
• time on page. a long time can mean reading or a broken layout.
• heatmaps without a question. pretty, and useless if you did not decide what you were looking for first.
• A/B tests on a site with forty visitors a month. the math does not exist. use interviews and session recordings instead.
• any percentage promise. "this change will increase conversions by 30%" is a guess with a decimal point. i will tell you what i changed and what happened, not what would happen on your site.

the honest limit

conversion optimization cannot fix a product nobody wants or a price nobody will pay. it can fix a page that hides the price, a form that asks for a phone number too early, a mobile layout where the button is below the fold, a contact page that loads in four seconds. those are real, measurable, and usually cheap.

start with the ladder, not with a redesign. the ladder tells you which rung is missing, and the missing rung is almost never the one the last agency was paid to polish.

for the traffic side, see how many people visit my site. for the speed side, see why my website is slow. for the service, see realizzazione siti web a parma and ux designer a parma.

faq

what does CRO mean in a web context?

conversion rate optimization: improving the share of visitors who do the thing the site exists for — send a form, call, book, buy. in parma the acronym also means contract research organization, which is why the first line of this page says which one it is.

how do i measure conversions without cookies?

use server-side or first-party events tied to the action itself: a form submission recorded by the form handler, a phone click recorded by a first-party script, a booking recorded by the booking tool. you do not need cross-site tracking to count your own inquiries. you do need a privacy notice that says what you count.

how long before i see results?

one change needs four to eight weeks on a small site, and sometimes two months of data before the difference is real. the first result is usually diagnostic, not financial: you learn which step loses people. that is already worth the work.

do i need a redesign to improve conversions?

almost never. the common wins are a clearer first paragraph, a visible phone number on mobile, a shorter form, a price range, and a faster contact page. a redesign is the expensive way to avoid finding out which of those was missing.