The Chatterbox

Where to start testing?

A question I get asked quite often regarding Conversion Testing is which page exactly one should start testing once they've determined that some conversion testing is in order. It's a good question, and honestly one that doesn't have any stock answer.

Without knowing any of the details of a given site the best general advice I can give you is to start where the testing is likely to give you the biggest bang for your buck. This could be your most popular landing page(s) or your home page or whichever page the visitor needs to make a buying decision.

The best way to sort out where you should start is to already have some sort of stats program installed so that you can get at least some details on what's currently happening with your site's traffic. There are lots of free or low cost stats programs out there, as well as some that can get quite expensive.

But even if you have one of these installed already, many seem to get confused by the data. There is so much data it's hard to narrow your focus down on what's most important. So let's take an example of a site I was just looking at this morning. I'm going to be conducting some conversion testing/analysis on in the near future so it's a good case study. (I'm obfuscating some of the identifiable details so as to not put the exact site out there, but it's a pretty typical situation.)

This site has a Subscription business model and delivers a service via the web. It does offer some stuff for free, but the goal is to get people to sign up for the service. The stats program it currently uses is AWStats, which is one of the pretty good freebie stats tools out there.

After talking to the webmaster I logged in to look at the stats. I saw a couple of things immediately.

1. The site is averaging around 750 Unique Visits per day by AWS' definition of Uniques. That's not a bad number for the market this site inhabits. Could be better, but could also be a lot worse. One note here is that existing members are going to be included in the Unique Visits total, but still there is a lot of new traffic coming in each day.
2. According to the webmaster the site is really having trouble converting visitors into customers. It's dismal, but not anything all that much outside the norm from what I see with sites on which no Conversion testing has ever been conducted.

The bad news is the site is averaging right around 2.8 sales per day, which equates to a 0.037% Conversion Rate based upon the AWS Unique numbers. Like I said, it's dismal. But a lot of sites convert at a pace of well less than 1%, so it's not all that unusual.

So, we're back to the main question. Where do we begin?

Well, in the AWS stats I scroll down the page a bit and come to the Pages - URL (Top 10) area. For your reference I've made a screen capture of this section. It looks like this: Screen Capture

The first two lines there are not important in the case of this site. They're simply files that are being included via PHP to construct the site navigation.

The third listing is actually a series of Samples pages. There are literally a hundred of more of these that are dyncamically generated for samples in each category the site offers, but are all being reported as a single page. These pages do get a fair amount of long tail traffic since their content gets pretty specific.

The fourth page is where they keep the Free stuff they offer. We'll gloss over this one for the moment, but will probably come back to it down the road since it would be a good place to move people from being Free users to Subscribers.

The next is the Pricing page, which is towards the end of the sales process. This and the Subscribe page (number seven) basically provide the same function, though in a bit of a different way visually. This is the point where visitors need to choose either to:
A. Continue surfing the site
B. Buy
C. Leave

These are the pages where the rubber meets the road, so to speak. They're where whatever anxiety or friction is going to come out naturally. And based upon the data AWS is giving us it's pretty apparent that right now the anxiety and friction are winning!

Do you see those Entry and Exit numbers?

Relatively few people are Entering the site on either of these pages, with 130 and 108 Entries respectively. From this we know that most people are hitting other pages of the site first. We also know they must have been interested enough in the service to reach these end-of-sales-process pages. So they already like the service, or they wouldn't have made it to these pages. Forget those few per day that are going on to purchase, who also had to have come through these pages. Look at the number of people who are Exiting from the site from these pages. Especially as compared to the number of Entries.

1,721 have exited the site from the Pricing page in the first half of January. An additional 1,219 have left from the Subscribe page so far this month.

Doing a quick bit of math, that's 2,940 prospective customers who were interested enough to make it all the way to the point they wanted to know the costs. This is just through the first 15 days of January.

Compare this figure to the 42 sales the site has had through the same time period and it quickly becomes evident where we should start. Where we stand to get the biggest bang for our buck.

If we totally discount the overall traffic to the site and concentrate simply on those visitors who make it to these two Call To Action pages we should be able to make some pretty decent improvements. This traffic is already interested or they wouldn't be showing up at these two pages. Realistically, we should be able to really crank up the sales ratio for visitors hitting these pages.

We just have to get out of our own way!!!

Seriously, even at a moderate 5% conversion rate for these visitors, who have already been qualified to a large extent, would equate to 147 sales in the first half of the month. Or just shy of 10 per day. (FTR, I think it can do much better than 5% of this already-qualified traffic.)

Having talked to the webmaster, I can tell you he'd be estatic to have 10 sales per day. It's a small web site and basically a one-person show, but with their price point 10 sales per day would amount to around $8,500-$10,000 per month as opposed to the $2,500 or so per month the site is bringing in now. And since it's a web-deliverable service, there are no extemporaneous costs. It's almost all profit.

So there ya go. There's a (pretty common) example of how one figures out where to start when they've made the decision to start some conversion testing. Sure we could (and probably will) optimize the sales funnel, but by starting off with a realitively simple and easy series of tests on just these two pages we're likely to see a drastic positive change as compared against the current sales and profits. So that's where we want to start.

What will we test? A few different of things come to mind immediately.

We'll change the copy on the page in order to improve it. We'll clean up some things I think are probably either contributing to or at the very least not helping to reduce the anxiety and friction of prospective customers. These simple text edits will be something we'll want to test to see how the changes affect conversions.

We may also deconstruct the page and start over with a radical redesign so that it's 100% evident to visitors where to click to make a purchase. As well as making sure Why one should purchase gains a more prominent mention. We'll probably test several of these benefits to see which produces best.

We may even do some price point testing. The webmaster is afraid to raise his pricing because others that offer something similar are his pricing model. I personally think he's charging too little because his service is better than any of the others out there that I've seen. IMHO he's devaluing his service, which is a mistake.

So there you have it. That's where and what we'll be testing. I'll be tracking it all of course. At some point in the future I hope to bring you an update to let you know exactly what worked and what didn't. And how much a tiny bit of conversion testing on just a couple of pages have improved things.