How to Build a B2B Landing Page Testing Strategy That Actually Works

Louis

Welcome back! We’re getting ready to launch a new Labs service, so today felt like the perfect time to talk about how to build a proper, strategic split testing schedule for landing pages.

Labs isn’t a replacement for ads management. It’s a layer that sits on top. 

While Ads management improves performance month to month. Labs focuses on uncovering the optimal ads and landing page setup over the long term. We’ll call it uncovering the performance mix.

This type of process typically needs a much longer time horizon. As we know with split testing, some tests win, and some lose – so the results over the short and mid term go up and down – but in the long term, the compounding effect is a massive upshot.

The aim is to understand performance down to pipeline, not just lead volume or cost per lead. We want to learn how to attract the right people, increase qualification rates, and drive more profitable deals.

It’s also worth noting that this isn’t a design programme where cosmetic amends are applied based on feedback. It’s a performance programme built on hypotheses, evidence and iteration. And it’s not driven by personal taste or best practice – it’s about discovering what actually works in your market, even when it challenges assumptions.

Maelien

So how do you run that kind of testing in a way that’s structured, consistent and useful? In this episode we’ll break down what makes a strong landing page testing programme so you can run a Labs-style approach yourself. 

The first thing to cover is getting into the right mindset.

When you’re running a split testing schedule – as Louis said, you’re going to get winners, and you’re going to get losers. And truthfully, you’re probably going to get more losers than winners.

And actually the main point I want to make here is that we have to move away from thinking about split test results as winning and losing. It’s actually more about winning and learning.

Because for every failed test comes a big learning, that you can avoid or explore in the next one.

In sales coaching, aspiring sales people will often be coached to celebrate every “no” they get. Because each no takes you a step closer to a “yes”. And it’s exactly the same with landing page experimentation.

As long as you are starting with a strong hypothesis – So rather than “I think people will prefer blue buttons over green ones” – more like “the lack of price indication is stopping people from enquiring” – each test will move you closer to a big win.

Louis

So let’s talk about what you need to get in place to create a successful schedule.

First, you’ll need to learn as much as you can about your ideal customer with an audience discovery process. We’ve talked about the discovery process for ads before – but for landing pages, you have to go even deeper.

You’ll want to go through reviews, competitor pages, forums, social posts, live chat transcripts and session recordings – anything that shows what’s important to your ICP and how they buy.

Next, you need to audit your current landing page through the lens of everything you just learned and identify what needs to change – and why. 

Create a spreadsheet and comment on layout, messaging, headlines, visual content, page-length, pricing transparency, objection handling – everything you can think of.

By the end of this process you should have a really good list of all the things you believe could make an impact. 

Order the list in terms of priority – what you believe will make the biggest impact first, and now you have a ready made split testing schedule backed by rock solid hypotheses.

Maelien

It’s worth noting that this is just a starting point too. You will learn things with every experiment, and you’ll add new tests and update hypotheses as a result.

If that happens, it’s actually a good thing – its a sign that you’re really getting into the learning and chasing down what works and why. 

There’s two really important things to consider before you actually start testing though.

The first point is that, it isn’t about rolling out a new landing page from the start. It’s important to use your existing landing page as the control. 

A control is basically what you’re testing against. Every time a new variant wins – that becomes the new control that you take forward.

So when you start with your existing landing page as a control, you can clearly see the improvements from your split testing programme – from start to finish.

And that plays into the second point – which is that you should only test one thing at a time.

The pain of not doing that is a lived experience for me. Because if you change more than one thing at a time and the landing page underperforms – you won’t know why. 

Was it one of the changes? Was it all of them?

The downside is that you’ve essentially wasted a test because you’ll have to go back and test each element individually to figure out what went wrong.

Now I am going to contradict myself here and say there are a very small number of instances where you will need to take a big swing and make multiple changes at a time – but I can’t stress enough that you want to avoid doing this wherever possible.

Louis

So let’s talk a little bit about reporting. When you’re managing a split testing schedule, reporting isn’t just for logging performance. Reporting and insights are also a huge part of knowing what’s working and what to test next.

As we’ve been running through the alpha version of our labs approach with clients, we’ve been sure to log the spit test variant at the deal level in the CRM – so 1a, 1b, 2a, 2b – you get the picture. By doing this we can easily keep tabs on which tests generated the best fit enquiries and the most pipeline as we go.

You’ll also want to keep a log of all of the test variants and their rationale in your testing schedule. So variant 1a would be the control and 1b might be a headline rewrite, or a new conversion mechanism. 

We also save and label screengrabs of every split test version for future reference. Definitely do this – your future self will thank you for it when you try and remember what variant 2b looked like after test 8 finishes running.

In terms of data, we measure qualified deals and pipeline value in the CRM and then also spend, conversion volume, conversion rate and cost per conversion on the ads side. 

Then on the qualitative side – we’re looking at scroll maps, click maps, attention maps and session recordings. We use Microsoft Clarity a lot for this. These will give you further insight into how your landing pages are being used and will uncover any blockages.

It’s worth noting that using session recording alone helped unlock an extra £250,000 revenue per year for one of our past clients. We saw that website visitors were trying to use their service configurator at the bottom of the page, giving up and then scrolling back up and hanging at the top of the page where the telephone number was. Some potential clients were getting stuck and phoning through whereas others were giving up and abandoning. So after some UX work to make the process more user friendly, we managed to fix it and the lost sales started converting.

Maelien

So the next big question is How long should you actually run a split test for? And the honest answer is… it depends. It’s a bit of a “how long is a piece of string” situation, because as we know B2B is very different to B2C.

In B2C Ecommerce – you can run a split test for 30 days, look at the sales numbers and total revenue, and that’s usually enough to call it. But in B2B, you don’t have that luxury. Sales cycles are longer, audiences are smaller, CPCs are higher, and you’re not going to get a flood of purchases in a week to validate a variant.

So instead of trying to optimise around full sales, we optimise around pipeline. For most B2B tests, we work to a maximum test flight time of 60 days. That’s the max, but shorter is better. The real goal is reaching 10 to 15 deals per variant. 

So: on a test with two variants 

• 20 total deals to make sure we’re heading in the right direction and that the test is worthwhile

And then

• 30 total deals or more to be able to determine a winner. 

To calculate how long it will take to get there, you just need your monthly budget, average CPC and conversion rate to lead. 

We’re going to run through how to do it now – and if you’re listening, it might be a good idea to pause and come back to this section when you can take notes, or if you’re watching the video version on YouTube, we’ll put the formula up on the screen as we go.

Start with your monthly budget and work out how many clicks you get per day on average.

So take your budget, divide it by your CPC, and then divide again by 30.4.

Take the answer and multiply it by your conversion rate to lead. That gives you your total daily lead flow. 

If you’re testing 2 landing page variants divide that number by two. This will give you the leads each landing page will get per day.

Now If you need around fifteen deals per variant, just divide fifteen by the number you just calculated – and that gives you your expected test duration.

Louis

And then to end, let’s talk about how to determine a winner.

When we talk about calling a winner in Labs, we’re not looking at “Which page got more leads?” We’re asking “Which page generated more valuable pipeline?”. And as a side note – when we say pipeline, we actually mean qualified deals – including closed lost. 

We need a way of knowing when the signal is strong enough to trust, and a strong signal really comes down to two things. 

First, have we collected enough data? So the 30 overall deals, or 15 per variant, we just talked about.

And then the second part is the gap. We need a difference that actually means something.

As a rule of thumb, if one variant is generating at least 20 percent more qualified deals and pipeline – that’s usually meaningful enough to call a winner.

You can think of it in simple scenarios.

If one page drives loads of cheap leads but they break down during qualification, that’s not a strong signal.

If one page brings in fewer leads but much better deals – so higher value and more reliable progression – that’s a much stronger signal.

If there’s a real, consistent difference in volume and value – not just a blip – and we’ve hit that deal volume target of 30, then that’s when we’re happy to end the test and roll the winning version forward.

Maelien

So to wrap this up, running a solid split testing schedule really comes down to a few simple things: 

• Do the upfront discovery so your tests actually mean something.
• Test one hypothesis at a time.
• Track everything at deal level, not just form fills.
• Give the test enough runway to collect real pipeline.
• And only call a winner when there’s a strong enough signal

Do that consistently and you’ll get two bits of massive value from it.

The first is really figuring out what’s driving pipeline – and what isn’t. 

And the second is ending up with a landing page that generates you more pipeline more efficiently longer term.

Louis

That’s it for today, if you found this episode useful, hit subscribe and share it with someone who’s working on their landing pages right now. 

Thanks so much for listening, and we’ll catch you on the next one.