In most cases, people are guessing. There's no guarantee to say that 3 or 4 or lots is the right answer. Given the same powerful call to action, different sets of people act differently.

The only way you can test whether this theory works, is to have each Buy Now button go to a completely different payment link, and then measure all the links over the period of a campaign.

It's still a very inconclusive theory. I'd say three is good, only because it's putting the customer in the frame of mind to buy AND that if I'm ready to buy, then I don't want to keep reading or scrolling to the bottom.

However, here you run a risk. You may not have built up enough value in the customer's brain. So when the customer clicks on the link, he may go through to a payment page, but may not purchase, because there isn't enough value in his brain. The only factor is price. So then you have the complexity of measuring clicks vs. customer's purchases vs. the price/value factor in the customer's brain.

So while Link 1 or 2 or 3 or 4 may get lots of clicks to the payment page, the value may not be built up. And therefore the sale may not go through.

It's a tricky question alright. And testing would be complex, if not reasonably impossible.

The verdict: We're really playing safe here with peppering links through the site. We don't really know for sure that three or four or twenty work.

Sean http://www.psychotactics.com Why Customers Say No-And What Causes Them to Say Yes.

Two quick comments:

Whenever a webmaster tells me he/she has tested some technical aspect of sales conversion, I am always anxious to dig a little deeper and find out exactly how the test was performed.

In most cases, it comes out that the "test" was simply something like "I tried it this way with one of my products, and that way with another."

IMO, "testing" by most webmasters is rarely the controlled, scientific, statistically valid exercise that it could be.

Even if you are testing with only one web site variable (i.e. order button placement), are you accounting for different testing time frames, segments of the audience, days of the week, months of the year, etc.

Are you comparing Saturday's results with Wednesday's? Results in July with results in November?

And what about the many external factors? Testing, even under the very best of conditions, is never an absolute guarantee. It's not surprising that people's opinions are all over the map with this subject.

My second comment is this:

I've seen reports that suggest up to 80% of the shopping carts with items in them are abandoned prior to the final sale. If that's the case, shouldn't the "buy now" links be tracked all the way through to the conclusion of the sale?

Could it be that some link placements lead to more abandonment than others? I don't know.

When any of you have all the answers, please call me right away!

Testing

Hi All:

For the record, when I say test, I mean A/B split testing. In this case, splitting traffic between two identical pages where the only difference is the link configurations. As pointed out, any other type of "test" result may be influenced by some other set of independent variables.

We also design all our tests to make sure we get at least 150 orders for both the A and B split to make sure any difference is not just random variation and is a statistically repeatable difference. Over the past 10 years, I’ve been able to run more than 3,000 tests on landing pages using this strict methodology – all for information products like magazines, newsletters, books and events.

You are correct to be wary of any test not run on an A/B split that takes the order all the way thru net including cancels and returns.

There is a lot of unfounded mythology floating around our industry.

Always ask about the test methodology.

Thanks for the feedback and discussion.

Don