CEO at Quru Oy
23 June 2004 20:15pm
I couldn’t agree more with the headline of this article and it’s one I’m afraid I can’t take credit for. I found this line in Paco Underhill’s book, Why We Buy – The Science Of Shopping, and found myself comparing many of the things he has measured in the retail world to the tests I’ve done with online, visitor-based activity. The conversion rate on a website is easy to measure. Unfortunately, businesses too busy concentrating on their bottom line most often overlook it. The point of this article is to define what a conversion rate is and show you how you can begin to start improving your own website’s conversion rate and therefore your bottom line. At the same time, I will relate my observations to Paco’s on offline retailing.
In Cyberspace No-One Can Hear You Shop
According to Paco, the main problem with websites is that, owing to media attention and the love of technology, retailers went online without knowing why. It’s true that in the late 90’s businesses were going online because their competition had, or because they feared that they would be left behind by not embracing the new technology. Not great reasons to spend time, money and resources on a website. The painful thing is that, since going online, most of these websites have not changed much for the better. Yes, they look nicer now, but the number of glorified poster sites I still see never ceases to amaze me. In order to combat this lack of purpose, I propose you look at four goals and adapt them to your own business requirements. One of these goals should be the primary focus of your entire website design.
1) Prospect AcquisitionTo deliver qualified leads and prospects through the website.
2) Sales/E-commerceTo sell products and services online directly through an e-store.
3) In-House Cost SavingTo cut costs, usually resources such as printed material or time, by automating in-house processes online such as timekeeping systems and human resource procedures.
4) Customer ServiceTo improve customer service by providing answers to queries and complaints online automatically where possible.
With the goal clearly defined, it is easier to measure the effectiveness of your site because you know what to look for. Conversion is defined in relation to the goal you’ve chosen.
So measure prospect acquisition as the percentage of visitors who give you their details out of the total number of visitors to your website. Measure conversion on sales as the percentage of people buying a product against the total number of website visitors. Conversion on in-house cost saving is simply the number of people using the system as a percentage of the number of people supposed to be using the system. A good internal policy here will mean this is a 100% conversion rate. The number of people using the resources and systems you have put in place as a percentage of total visitors to the support web pages can give you your customer service conversion.
So why measure conversion? Because it allows you to accurately measure the impact of changes you make by measuring the performance of your website before and after the change. With that valuable information in hand, you can make adjustments accordingly.
The Butt Brush Factor
In many instances in his book, Paco refers to ‘The Butt Brush Factor’ — the way people, women in particular, don’t like enclosed spaces where other people constantly bump into them from behind. It usually led to the prospective shopper feeling frustrated or feeling uncomfortable and leaving the store or going somewhere else. You might be thinking, “well how does that relate to an online experience?” It is true that no-one usually bumps into you from behind while you’re sitting in front of a computer, but how many times are you made to feel irritated, uncomfortable or just downright frustrated by a website? How often do you leave one and look at another because the first one doesn’t have what you’re looking for? This ‘Butt Brush Factor’ is incredibly relevant to websites, more so I think than even in ordinary retail. Here are some examples of common online ‘Butt Brush Factors’ that you will see in many business websites.
1) Latest News.The landing page has the latest news about the company links. What exactly is the point of having a bunch of latest news links on your landing page? What good is that to a browser arriving at your landing page knowing and caring little about your company? A browser wants to know what you can do for him right there and then, not how your company stock is doing. An ‘About Us’ section is a much more reasonable place to put these links.
2) Awards.A landing page with awards screams, look at us, look at what we’ve achieved, aren’t we clever? It also completely wastes space on the most important page of your website. It can be compared to what Paco said when he talked about going into a car showroom and seeing manufacturer awards. That is unlikely to make much of an impression on the average shopper.
3) Poor Headlines.‘Welcome to Company Name’ is the most common waste of a headline I ever see. Probably the company is unknown to the visitor so you’re wasting his or her time. A headline, which communicates the need of the target audience and how you can solve that need, improves reading and click through by up to 35% in recent tests we made.
4) Submit Buttons.Why tell the visitor to ‘submit?’ Submit actually means “To yield or surrender (oneself) to the will or authority of another” according to dictionary.com, so why ask innocent web browsers to do that in order to read your monthly newsletter? Subscribe to our newsletter is much more friendly, I would say.
5) Bad Use Of Flash.This is a common problem with media companies in particular. I understand why they do these all singing all dancing interactive flash websites, which often are works of art and showcase their ability. However ‘skip intro’ is a common link on the majority of these websites. That is because some people find them a waste of time. Why have an intro at all? Why not just have a showcase of what you can do on a normal fast, efficient website which tells me what I need to know quickly? If I decide I have the time to look at flash animations I will.
6) Poor Use Of Imagery.I’m guilty of this myself. We used to have a picture of a squirrel flying through the air with ‘what’s your objective’ on our landing page. It might have worked had we been selling nuts or seed, but a company improving website conversion? Not really relevant! It was more a result of my ego, pride and photographic luck in capturing said squirrel with my digital camera, and then thinking of a way I could use the picture, than thinking of a good picture which was relevant to what we were trying to say and using that. This kind of thing is repeated on many websites — people with briefcases, bridges, animals and other general graphics, which can be turned with words into anything you want the image to say. But on first glance, they don’t really show any relevance. All communication should be relevant and, ideally, persuade the user to do something.
Again, conversion is an important measurement here. It can be applied to all of the changes you make to your site as you eliminate these ‘Butt Brush Factors’. Later in this article, I’ll explain how.
Attention All Shoppers…
“For the next fifteen minutes, in the frozen food section, free passion fruit sorbet for everyone” is a perfect way to instill urgency in shoppers to go to that section of the store and get the freebie. They know they only have 15 minutes, and they know that after that time they won’t get the lovely sorbet. This was Paco’s way of showing how stores could be more imaginative. The store knows that that section of the store is going to be jammed with people for that 15 minutes and can capitalize on impulse sales. That’s how it works in the retailing world, but what about online? Instilling urgency online is a major factor overlooked by many business websites. Some examples of how you might want to start employing this technique online are listed below.
1) Time Expiry Offer.Just as in the above example, you could let your readers know they will miss out if they haven’t subscribed or bought your product by a certain time.
2) The First Number.Your website could offer the first 50 subscribers a free e-book or could advertise that the first 50 items sold will be at a 30% discount. This could be combined with a counter showing the number of places/items left, so that the browser thinks “I have to subscribe before those places are taken up”.
3) The Nth Number Competition.The website states that if you are subscriber number 1000, you get a free website makeover, again combined with a visible counter of the current number of subscriptions. This could be tied into a referral deal so that if the subscriber is not the lucky number and does not get the deal, at least he could be offered something for making the referral while his friend might still end up being the lucky number and win the prize.
So how does conversion relate to all these changes? The conversion rate should and can be measured in every instance.
The Science Of Online Marketing
There are two incredibly significant lines in Why We Buy:
“Science is by and large the study of very small differences” and “When you change one thing, everything changes”.
The first ‘very small difference’ and ‘changing one thing’ situation I came across in my online marketing career was a complete mistake. I was working for a large press organization and one day I had to change some HTML code on a sales form. By mistake, I removed a voucher entry field from the form. As a result, people could no longer enter their voucher number to get a cheaper deal. Conversion improved by three times. I told our editor who was amazed but instructed me to put the voucher field back on the form while they figured out what to do. There was a good reason for the voucher; in fact, it was the entire reason the page was there. However, putting the voucher entry field back resulted in a drop in conversion to almost the identical sales that we had been getting before my mistake. The voucher idea was eventually scrapped on that page and sales sky rocketed again. The reason, we ascertained, was that visitors figured that they could get a cheaper deal with a voucher. The voucher could only be gotten by physically buying a newspaper and that limited us to around 10% of the audience. Nine out of ten people visiting the website did so from a place where they couldn’t buy the newspaper at that time, so it was obvious that the voucher idea could only be good for the local readers. This experience was a catalyst for me personally, and from then on, I began to understand the importance of measurement online. In particular, the measurement of conversion.
So in order to turn the online changes you make into a science, follow three simple rules.
1) Measure Conversion.Conversion is a percentage, a calculation of the number of people who take the action you desire as a percentage of the total number of visitors to the page. Using percentages makes the actual number of people arriving at a page irrelevant. It becomes possible to compare a busy week with a quiet week.
2) Change one thing at a time.An average page has lots of variables: graphics, headlines, paragraphs, sentences, links, testimonials and probably a lot more. By only changing one thing and always measuring for the same period of time (30 days is good), you will get a fair result. So for instance, if you change a headline, look at the page click-through and if possible the length of time an average visitor stayed on the page for 30 days before the change. Make the change and measure the results for the next 30 days. Then if conversion is higher (more people reading or more people clicking through), keep the change. If it’s lower, revert to what you had before.
3) Experiment.Don’t limit yourself to headlines. Copy, content, graphics, adding competitions, etc. — try them all. But remember the rule: change only one variable at any one time.
I’ve desperately been trying to keep this article short; I think I could have written an epic on this subject. If I were in the same room as Paco Underhill, we would have an awful lot to talk about. However what I’m trying to say is that businesses should start waking up to the fact that online marketing is as much a science as Paco demonstrates in the retailing world. Measuring conversion rates online is the beginning of making it scientific.
More articles like this at:
AM at IDF Technologies, LLC
25 June 2004 10:27am
Great article, but I disagree with one statement:
>Using percentages makes the actual number of people arriving at a page irrelevant.
>It becomes possible to compare a busy week with a quiet week.
Conversion rates alone do not provide enough information. You can't be sure that the results of your test represent the reality with any degree of certainty; your sample size might not be large enough for the difference in conversion rates.
For example, "group A has 2.5% conversion rate and group B has 1.7% conversion rate" is not a sufficient answer.
Since your sample size has to be limited (for obvious reasons) you have to factor in the margin of error, which will decrease as your sample size (for each group independently) increases.
There are many ways to represent such data.
You can compare the standard error of the mean for each group as check how much overlap you get for these groups. You could have something like: with 95% confidence level, mean conversion rate for group A is 2.0-3.0%, for group B 1.6-1.8%. With such numbers, you could be sure that group A indeed performed better. But you could've had group B with 0.7%-2.7% in which case, the distributions of means overlap too much to consider that data conculsive. In any case, you would not be able to calculate that without taking sample sizes (for each group individually) into account.
Another way to judge the "reliability" of your data is by using contingency tables. With either Fisher's or Chi-square tests you can calculate the probability of your data being a result of random occurance (null hypothesis). For that, you don't need conversion rates at all, but the number of actions and totals for each group.
To make things even more interesting, there is one more way to look at your data. This method gives you the best guess as to the reliability of the data for small sample sizes, but is rarely used.
What if you looked at your data not as the number of actions vs total number visitors, but as a sequence of visitors up to each action? In other words, how many visitors it takes to make a subsequent sale (or whatever action you want them to perform).
You start a test and count visitors until you get a sale. Then you reset the counter and count till the next order, and so on.
Let's say you ran two tests for two groups: 100 visitors and got 4 orders each. Of course, this same size is way too small, but I'll use it just to explain the concept.
For the first group, you got the following sequence: 7,45,4,44
For the second group: 23,28,24,25
Both groups have 4% conversion rate, but if you look at the numbers, you can intuitively say that the second group had a constant conversion of around 4% throughout the test, while the first group has not. Based on those numbers and the sample size, I can bet money that the mean conversion rate for the second group will be 3.52-4.63% in 95% of future trials. But the first group does not have enough data to calculate a reliable estimate. Even though, we had the same conversion rates and the same sample sizes.
So it's not all that simple :)
25 June 2004 11:55am
>You can compare the standard error of the mean for each
>group as check how much overlap you get for these groups.
>You could have something like: with 95% confidence level,
>mean conversion rate for group A is 2.0-3.0%, for group B
>1.6-1.8%. With such numbers, you could be sure that group
>A indeed performed better. But you could've had group B
>with 0.7%-2.7% in which case, the distributions of means
>overlap too much to consider that data conculsive. In any
>case, you would not be able to calculate that without
>taking sample sizes (for each group individually) into
You are of course right. This is what we do with clients and for our own purposes. We measure conversion rates on mean averages per sampled 1000 visitors. However my article was 2000 words plus and I wanted to explain what conversion was and illustrate some of the things which effect it. You're so far ahead of the average reader of my articles that it's untrue. To give you an idea, one of the responses I had asked how he should go about measuring conversion with webalizer and told me his most visited page was style.css.
What I am saying is I was aiming at a less educated reader ;o). That said your observations are spot on.
>To make things even more interesting, there is one more
>way to look at your data. This method gives you the best
>guess as to the reliability of the data for small sample
>sizes, but is rarely used.
This is excellent. A method I've never used and something which might work for some of the B2B websites we work with which have smaller visitor counts. I'm pleased you posted this here because I learned something today.
26 June 2004 01:20am
I understand. Sometimes, it's better to keep things simple so everyone can grasp the concept before moving on to the details.
I know a lot of people who tried spilt run testing and never got good results, mainly because they did not do things properly.
You might want to show this piece to your readers, who already got the basics straight:
I am surprised at the amount of people who do just what I described there.
Also, I'm glad that I stumbled across this foum last night, and I got a question.
I'm in the process of writing an e-book about testing and tracking that I've done in the past 3 years or so. It's basically click-stream tracking meets split-run testing meets visitor segmentation - how to use those contepts together.
If you think that the tip I posted last night (sequences instead of totals) would be interesting to your audience, we might do a jv once I'm done with the e-book. Let me know what you think.
Free market research on digital marketing
Daily Pulse: award winning newsletter
It takes 30 seconds to register