Opinions ship, evidence does not
Pages change because someone senior prefers the new version. Nobody can say afterwards whether it helped, so the site becomes a record of who was in the room rather than what worked.
Services
Conversion Rate Optimization
Most B2B sites lose the visit before the form. We find where, test the fix, and ship what wins. Every change is recorded with what it did, so the same idea never gets paid for twice.
Book a Discovery Call
Tested, Not Guessed
How CRO Should Work
The ProblemThe Redesign Cycle
Tests that cannot conclude
A B2B site rarely has the traffic for a clean A/B test on a low-intent page. Tests get called early, the wrong variant ships, and the team loses faith in testing altogether.
Form fills counted as wins
Conversion gets measured at the form and stops there. Nobody checks whether the extra leads became pipeline, so a change that doubled junk enquiries still reads as a success.
The SolutionThe Legency Approach
Research before any test
Analytics, session behaviour and the sales team point at where the site loses people. The backlog is ranked by the size of the leak, so the first test is on the page that costs the most.
Sized to the traffic you have
Where the traffic supports a test we run one and wait for significance. Where it does not, we say so and change the page on research instead, then track the before and after honestly.
Measured at pipeline, not the form
Conversions are tracked through to qualified opportunities wherever your CRM allows it. A change that lifts form fills and drops qualified leads is a loss, and it gets reported as one.
The Programme
What a CRO Engagement Runs
Conversion Audit
A full pass over the paths that matter: landing pages, the pricing page, the forms, the mobile experience and the speed budget. It ends in a ranked list of leaks with the traffic and the value behind each one.
Testing Programme
Hypotheses drawn from the audit, run in order of expected value. Each test states what it expects to move before it starts. Wins ship, losses get documented, and the log is yours to keep.
Landing Page Builds
Pages built as components in Webflow, so a new campaign page takes hours rather than a sprint. Built to be tested, which means the parts most likely to change are the parts easiest to swap.
Pipeline Tracking
Events, goals and CRM handoff wired so a conversion can be followed past the form. Without this the whole programme is guesswork dressed as data.
Before Anyone Sells You Testing
Most B2B Sites Cannot Run the Test They Are Being Sold
An A/B test can only prove a change bigger than the noise in your own traffic. That threshold is arithmetic, not opinion, and it is the first thing we work out.
On a 2% conversion rate, a site with a thousand visits a month can only prove a change that more than doubles conversion. Nobody makes that change on purpose. A site with fifty thousand can prove an 18% lift, which is a realistic target.
Where the traffic is not there, we say so and change the page on research instead, then track the before and after honestly rather than calling it a test.
Where B2B Sites Lose People
The Five Leaks We Find Most
These come up on nearly every audit. None of them is a design argument, and all of them are measurable before and after.
- 01
The form asks for things sales never uses
Company size, job title, phone number, "how did you hear about us". Each field costs completions. We check which of them the sales team has ever filtered on, and the ones nobody uses come off. This is usually the single largest recoverable loss on a B2B site, and it costs nothing to fix.
- 02
The page speaks to one person on a committee
B2B buying runs through several people who are not the champion. The page answers the practitioner and says nothing to the person who has to sign, so the champion has nothing to forward. Pricing posture, security posture and implementation effort are the three usually missing.
- 03
Mobile is treated as a smaller desktop
A third of B2B traffic reads on a phone even where nobody buys on one. Blueflame's mobile score went from 36 to 72 in a day, and that was a build problem rather than a design one.
- 04
The proof is generic
Four logos and a testimonial with no number in it. Proof has to be specific enough to be checkable, or the reader discounts all of it including the true parts.
- 05
Nothing is tracked past the form
Conversion is counted at submit and never followed. So a change that lifts form fills and fills the pipeline with junk reads as a win in every report, and the sales team quietly stops trusting marketing's numbers.
The Deliverable
You Keep the Log, Including the Failures
Every test is written down before it runs: the page, the hypothesis, what it expects to move, and how long it needs. Afterwards it gets a result, including the ones that lost.
The losses are the part that makes the log worth keeping. An agency that only records its wins hands the next agency a blank page, and your company pays to learn the same thing twice.
The log is yours. It stays useful after we are gone, which is the point.
TestExpectedResult
Cut form from 9 fields to 4+20% submitsShipped
Pricing posture above the fold+10% demo startsShipped
Video hero on the home page+15% engagementNo effect
Social proof bar under the CTA+8% submitsReverted
Shape of the log, not a client's results. Real test logs stay with the client.
Case Study
The ads and the pages, run by one team.
We manage Blueflame's paid search and the pages those clicks land on, and have for over eighteen months. Campaign and landing page get tuned together, against the same conversion data.
View Case Study
Common QuestionsFrequently Asked
Frequently Asked
Questions
Three things in order. It finds where a site loses people, using analytics, session recordings and the sales team rather than opinion. It tests changes against that evidence. Then it reports what each change did, including the ones that failed. Anything that skips the first step is redesigning, not optimising.
For testing specifically, more than most people are told. On a 2% conversion rate, a month of traffic split across two variants needs roughly 20,000 visits before a test can prove a 28% lift, and around 50,000 before it can prove an 18% one. At 1,000 visits a month the only change a test could prove is one that more than doubles conversion, which is not a change anyone makes deliberately. Below the threshold, CRO is still worth doing: the audit, the form work and the page rebuilds all pay, and we track them as before-and-after rather than pretending a test ran.
The audit produces fixable problems in the first fortnight, and the obvious ones ship straight away. A test needs two to six weeks to conclude depending on traffic. Compounding gains take a few months, because they come from a run of small wins rather than one redesign.
Only when the audit says the structure is the problem. A redesign resets everything you have learned and makes the result impossible to attribute, so it is the expensive answer. Most of the value sits in a handful of pages, and we would rather prove that first.
Landing page optimization is one part of it, and usually the part with the fastest return because campaign traffic is concentrated and high intent. CRO covers the whole path, including the pages someone visits after the landing page and before they convert.
Analytics access, CRM access if you want conversions tracked to pipeline, and about an hour a month. The reason we ask for sales input is that the sales team already knows which leads are junk, and that is the fastest route to the right test.