- A pipeline plateau with growing traffic is usually a conversion problem, and the website owns most of the conversion between a first click and a booked call.
- A/B tests change the surface of a page, so they can't fix a funnel whose structure sends the wrong buyer down the wrong path.
- Detecting a 10% relative lift on a 2% conversion rate takes roughly 80,000 visitors per variant, which is more traffic than most B2B sites get in a quarter.
- Rising CAC on flat conversion means you are paying more to push traffic through the same leaky funnel; the channel itself may be working fine.
- You can diagnose a conversion plateau in an afternoon by putting four quarters of sessions next to four quarters of booked demos.
Here's the quickest diagnosis I know for a stalled pipeline. Put sessions and booked demos for the last four quarters side by side. If sessions climb and demos don't, you have a conversion problem, and I'd look at the website before I'd touch the ad account.
That pattern hides well, because every number marketing reports still looks good. Sessions grow, so marketing has a good quarter. Pipeline is flat, so sales has a hard one. The two teams describe different companies, and the website sits between them with nobody looking at it.
Why has our pipeline plateaued when traffic is still growing?
A pipeline plateau with rising traffic means your funnel is converting a shrinking share of a growing crowd, and the website decides most of that share. Ads and content bring more visitors in; the site is what turns a visitor into a booked call.
It's a conversion problem. Your website decides what happens between someone landing on a page and booking a call, and that's the part that has stopped working. When acquisition keeps improving and pipeline doesn't, the constraint has moved out of the channel and onto the page. More traffic just means more people walking past the same locked door. Of the signs a B2B company has outgrown its website, this is the one that hides longest.
Where exactly is the funnel leaking?
The funnel leaks in the stretch the website controls: the path from landing on the site to booking a call. A page has to qualify the visitor, convince them and route them to a next step, and a page built for an earlier buyer does none of those well.
You can see it in the shape of the demos, not only the count. Sales cycles get longer because prospects arrive at calls under-informed: the site didn't do the pre-selling. More demos turn out to be "just browsing" because nothing on the page filtered out poor-fit visitors. Higher-intent visitors bounce because the site answers the questions last year's buyer had. The funnel behaves as one system, and when one part of it is built wrong the whole thing under-converts.
Why can't A/B testing fix a conversion plateau?
A/B testing optimises the surface of a page, so it cannot change the structure underneath it, and structure is usually what causes a B2B conversion plateau. Headline wording, button colour and hero image are surface variables; which buyer the page speaks to, the order of the argument and where it asks for the booking are structural.
If the structure sends a VP down a path built for a practitioner, testing the button copy optimises a route to the wrong place. There's a maths problem too. Detecting a 10% relative lift on a 2% conversion rate, at the usual 95% significance and 80% power, needs roughly 80,000 visitors per variant (you can check it in Evan Miller's sample size calculator). Most B2B sites don't get that in a quarter, so teams run tests for months and read noise as results.
If you've run A/B tests for two quarters and conversion hasn't moved, that isn't bad luck. Tests can only move small things, and your problem is a big one. Rebuild the path first, then test at the edges.
How does a conversion plateau show up in CAC?
A conversion plateau shows up as rising customer acquisition cost, because pushing more spend through a funnel that converts at the same rate means paying more for every customer you win.
The board sees CAC going up and asks why. In my experience the honest answer is often the website, and it's the last place anyone checks. Teams respond by buying more traffic, which raises CAC again. Fixing the leak usually improves CAC faster than tuning ad spend, because it lifts conversion on all the traffic you already pay for. And if every change to the funnel needs an engineering ticket, the plateau hardens while you wait in the queue, which is the real cost of a dev-dependent website.
How do I diagnose a conversion plateau in an afternoon?
You can diagnose a website conversion plateau in an afternoon with three things: your analytics, your CRM and a fresh browser window. Run these four checks in order.
- Compare the two lines. Export sessions and booked demos for the last four quarters. Sessions up and demos flat means conversion; both falling means traffic.
- Check demo quality. Pull the last 20 demos from your CRM and mark how many were a real fit. A rising share of poor-fit calls means the site isn't filtering.
- Walk the path as a buyer. Land on your best-performing paid page and count the steps, and the unanswered questions, between you and a booked call.
- Read the page for the buyer you sell to now. Ask whether the first screen speaks to the person who signs today's contracts or the one who signed two years ago.
If the path is longer or vaguer than it should be, that gap is your rising CAC in slow motion.
What does fixing the conversion path involve?
Fixing a conversion plateau means rebuilding the path from high-intent landing to booked call for the buyer you sell to now: the messaging, the order of the argument, the proof and the placement of the booking.
Start with the leak you found, not with a new visual design. Map each step and ask whether the page qualifies the current buyer and answers their current objection. Then build it so marketing can keep tuning the funnel without a developer, because plateaus come back when teams learn something and can't act on it for two sprints. When the leak runs across the whole site, it becomes a focused rebuild; a typical one takes 8–12 weeks, and the B2B website rebuild process sets out each phase. The Opsi Systems case study shows how I approach a site built for an enterprise buyer, and my Webflow agency page covers how I build them.
