First, what are landing pages?
A landing page is a standalone web page separate from the main website. It’s designed to fulfil a particular business objective. As such you don’t need to include the same navigation as your main site – the global navigation. The idea is to restrict people’s movements in a targeted way so they act the way you want them to, without distractions.
As such a landing page is a classic piece of direct marketing collateral, written to drive people down a specific route and carry out specified action. You might, for example, create a landing page designed to encourage people to give you some information, accept an offer or buy a product.
- Click through landing pages persuade a visitor to click through to another web page, for example performing a ‘warm up’ function to make them feel more inclined to buy. It’s no good just sending people direct from an inbound ad to your shopping cart or registration page. You want them to be in possession of all the facts first, so they end up farther down the sales funnel and more likely to buy.
- Lead generation landing pages are all about gathering data about people so you know more about them. Why do you care? Because the more you know about your target audience as individuals, the better you’ll be at fulfilling their needs. If you know someone prefers blue to orange, for example, you can offer them a blue thing and stand a better change of making a sale.
What is testing in a marketing context?
You might have two creative approaches for a landing page, and you want to find out which performs best. One might be short, the other long, where one gives more information and the other less. Or one design might be full of empty white space, the other colourful and dense.
Whatever you want to test you can assume all you like, but you won’t know for sure which works best until you test them head-to-head, directly against one another. It’s called AB testing, which simply means testing version A against version B, although there’s no reason why you can’t test multiple different approaches against each other simultaneously.
An example of a DM test
Here’s a real life, old-school example of AB testing for direct mail.
You don’t want, at this stage, to test whether different types of people respond differently to each approach, just which approach works best in general. So you hook out two segments of 10,000 random contact records from your database of 100,000 prospects on a 1 in n basis, as good a way as any to create genuine randomness. Then you send 10,000 people the offer in a plain white envelope, the other 10,000 in a bright yellow envelope.
As it turns out, you get a perfectly respectable 0.3% response to the white envelope and a much better 0.5% response to the yellow envelope. Because your segments were statistically relevant you can draw reliable conclusions from the test, and you conclude it’s worth rolling out the campaign to the remainder of the database in a yellow envelope.
About the yellow envelope thing
This is a real example from a financial services direct mail campaign I worked on in another lifetime, in the early 1990s when I worked as a direct marketing executive. But the principle works as well today as it did then, and it translates beautifully to digital marketing.
The yellow envelope thing generated a huge amount of profit because, back then, it signified urgency, being the type of envelope old fashioned telegrams came in. People subconsciously felt the envelope contained important information, so they were more likely to open it and, having opened it, respond positively.
To show you how apparently arbitrary the results of testing can be, we also tested a white envelope against one with a red ‘urgent’ flag on one corner, and contrary to our expectations the plain envelope out-performed the flagged one.
AB testing on steroids
I remember the most complex campaign we carried out, an offer we initially sent to a million high street bank customers. We segmented the data and AB tested fifty different creative approaches, each to a 1 in n ‘random’ segment of 20,000 people. Once we’d established the most successful approach we rolled it out to all seven million remaining customers and made – if I say so myself – a killing of legendary proportions. All very satisfying.
Segmenting people depending on their behaviour
You can do the same with types of prospect / customer. Your database might be made up of people who’ve already bought from you once, twice and three times, a chunk who’ve asked questions but not bought yet, and those who’ve got as far as the shopping cart then changed their minds. To maximise conversion you’d send a different offer to each – in digital marketing terms sending each segment to a different, specially tailored landing page created to suit their circumstances.
Patience is a virtue…
If you’re only driving a small amount of traffic to each of the landing pages you’re testing, you might have to exercise patience before you draw conclusions. Analysing the behaviour of fewer than 1000 visitors per landing page is risky because it just isn’t statistically significant. As a rule, the more people take part in your test, the more reliable the results.
What can you test via landing pages?
You can test all sorts of exciting things with landing pages, including different:
- Sales messages
- Communication styles
- Products and combinations of products
How to write a high-conversion landing page?
Because a landing page is the perfect direct marketing medium, you need someone who knows how to write DM copy. All that means is writing with the sales funnel front-of-mind, driving readers towards a specific course of action, whether it’s handing over their contact details or actually buying something. And doing all that in an enjoyable and satisfying way so they don’t know – or don’t mind – that they’re being manipulated.
That’s me, that is. With 25 years’ experience in the DM industry, ten of which have been in the freelance copywriting and content creation sector, I’m perfectly qualified to write hard-working landing pages. Just ask.
(Thanks to sxc for the free image)