You have made your own web site, and you are satisfied with your work. Your site is up, so now it's time to make money! Well..., slow down a bit. Before spending big bucks on advertising, let's test your web site. You may be in love with your site, but that doesn't mean web surfers will love it, too.
Your Very First Test
Testing your web site is much easier than the final exams you took in school. First, you have to be able to track your visitors because how visitors browse your site can tell you its strengths and weaknesses. For your initial testing, your server's log file is sufficient, but you will eventually need some sort of visitor tracking software or service.
Once you have figured out how to track you visitors, buy some cheap ad space. Your first ad doesn't have to produce heavy traffic to your site. If it generates 30-50 visitors, that's fine. More than 90% of visitors (those who actually come to your site, not those who see your ad) should view pages on your site other than your home page, and 40-50% of them should subscribe to your newsletter. If these numbers are too low on your site, there is something wrong with your site. Before going to any farther, you must correct the problems. Once you get good numbers, you are finally ready to roll.
Keep Testing
That's not the end; that's just the beginning. You can test almost every step in running your business. For example, you can test your ad copy. You can run one ad in an inexpensive but targeted location, and see how it turns out. If it turns out good, use that ad copy. If it doesn't, revise the copy and test it again. You can even run more than one ad at a time using different copy for each ad. Testing may sound like too much work, but just think how many extra sales you will make with better ad copy.
A Tip
A "?" can be very handy for getting accurate tracking results. Suppose your URL is "www.yoururl.com" and you want to track ads to see how much traffic they are sending to you. Instead using "www.yoururl.com" for the link, use "www.yoururl.com/?123" or "www.yoururl.com/index.html?123". Anyone who clicks that link will still see your home page, www.yoururl.com, but your log counts that visit as a hit to "www.yoururl.com/?123". You can tell where visitors come from by the codes after "?". You can code any combination of any letters and numbers after the "?" for each ad. You just need to keep track of which code represents which ad campaign.
Testing takes time, but the test results help you pinpoint any problems. If you don't know you have a problem, you can't do anything with it. Instead of increasing, your profit will decrease.
Let testing be your habit. When you work with the problem, you can improve your business. You won't make the same mistake again and again.
Showing posts with label Analyzing and Testing. Show all posts
Showing posts with label Analyzing and Testing. Show all posts
Wednesday, May 2, 2007
Analyzing Log Files for Profit
You might already know the importance of analyzing server logs, but not all log analyzers are suitable for business use.
Problem
Once you own a business, you should figure out the profit, and make sure your business is always profitable. Let's say you bought two $100 advertisements at the same time, and you used the query string, "?ad-1" and "?ad-2." Then, you got the following statistics of viewed pages.
hits page
12,000 home.html?ad-1
9,000 home.html?ad-2
500 order_confirmation_page.html
From these stats, you can see that ad 1 generated more traffic to your site than ad 2, but you cannot see where the 500 sales came from. It is possible that ad 2 could generate more sales than ad 1. Or, maybe neither ad generated any sales at all; maybe all 500 orders were made by visitors via search engines.
Statistics of Click Trail
To find out who bought, you need the statistics of the click trail (or content path), which looks like this.
Visits Path
78 page1.htm page5.htm page3.htm page8.htm
45 page2.htm page3.htm page6.htm
It says 78 visitors browsed your site by following the path indicated above, 45 visitors took other paths, and so on. Here are a few vendors of log analyzers with click trails.
• www.clicktracks.com
• www.hitbox.com
• www.webtrends.com
Also, you can read reviews of log analyzers.
Accurate Profits
Going back to the above example, when you count entries of click trails, like
home.html?ad-1 (omit) order_confirmation_page.html and
home.html?ad-2 (omit) order_confirmation_page.html,
you can see how many sales or how much profit each ad made. Now, you can make decisions based on profitability. For example, when you analyzed the click trails, you could see that ad 1 generated only $80 in sales, while ad 2 made $3,000 in sales. It is clear that buying ad 2 was a good investment, but buying ad 1 was just wasting money, even though ad 1 sent you more people. If you were checking only statistics of viewed pages, you would buy ad 1 again, and lose money again, wouldn't you?
Log files don't tell you anything by themselves. But, once you analyze logs correctly, you will get lots of useful information. That would be the best advice you can get.
Problem
Once you own a business, you should figure out the profit, and make sure your business is always profitable. Let's say you bought two $100 advertisements at the same time, and you used the query string, "?ad-1" and "?ad-2." Then, you got the following statistics of viewed pages.
hits page
12,000 home.html?ad-1
9,000 home.html?ad-2
500 order_confirmation_page.html
From these stats, you can see that ad 1 generated more traffic to your site than ad 2, but you cannot see where the 500 sales came from. It is possible that ad 2 could generate more sales than ad 1. Or, maybe neither ad generated any sales at all; maybe all 500 orders were made by visitors via search engines.
Statistics of Click Trail
To find out who bought, you need the statistics of the click trail (or content path), which looks like this.
Visits Path
78 page1.htm page5.htm page3.htm page8.htm
45 page2.htm page3.htm page6.htm
It says 78 visitors browsed your site by following the path indicated above, 45 visitors took other paths, and so on. Here are a few vendors of log analyzers with click trails.
• www.clicktracks.com
• www.hitbox.com
• www.webtrends.com
Also, you can read reviews of log analyzers.
Accurate Profits
Going back to the above example, when you count entries of click trails, like
home.html?ad-1 (omit) order_confirmation_page.html and
home.html?ad-2 (omit) order_confirmation_page.html,
you can see how many sales or how much profit each ad made. Now, you can make decisions based on profitability. For example, when you analyzed the click trails, you could see that ad 1 generated only $80 in sales, while ad 2 made $3,000 in sales. It is clear that buying ad 2 was a good investment, but buying ad 1 was just wasting money, even though ad 1 sent you more people. If you were checking only statistics of viewed pages, you would buy ad 1 again, and lose money again, wouldn't you?
Log files don't tell you anything by themselves. But, once you analyze logs correctly, you will get lots of useful information. That would be the best advice you can get.
Log Analyzer Types and Accuracy
There are 2 types of log analytic tools, the software version and the hosted version. Many internet marketing gurus insist on using the hosted version to get perfectly accurate stats. But are they more accurate?
The Difference
The difference between these 2 methods is how to identify individual PCs (or visitors). The software version analyzer uses IP addresses to identify individual visitors. The hosted version analyzer writes a cookie on every visitor's PC and uses it to identify visitors.
Accuracy
Because new IP addresses are assigned every time modems are turned on, many internet marketing gurus think cookie-based analysts are more accurate. But increasing numbers of always-on broadband users make IP address-based analysts more accurate. Furthermore, just writing cookies won't give you accurate results anymore since today's security software blocks cookies by default, or people have access to more than one PC, or family members share one PC, and so on. Identifying individual users has become a real headache for researchers. (You may use this story to get some discount if you are outsourcing analytic work.)
Fortunately, although you don't get 100% accurate stats, both methods are accurate enough to evaluate the performance of your marketing campaigns. You can still reliably determine which campaign made a profit. Forget about getting accurate results; focus on interpreting the results.
Risk
However, using a cookie-based analyst has one big drawback. The code you must add on every page of your site may slow down your page's loading speed. Sometimes it takes an extra 6 seconds to load a page. Only the entry page would be affected, but the loading time of the entry page is the most important one. As broadband internet connection becomes more common, visitors get less patient. An extra few seconds of loading time could turn visitors away.
The Difference
The difference between these 2 methods is how to identify individual PCs (or visitors). The software version analyzer uses IP addresses to identify individual visitors. The hosted version analyzer writes a cookie on every visitor's PC and uses it to identify visitors.
Accuracy
Because new IP addresses are assigned every time modems are turned on, many internet marketing gurus think cookie-based analysts are more accurate. But increasing numbers of always-on broadband users make IP address-based analysts more accurate. Furthermore, just writing cookies won't give you accurate results anymore since today's security software blocks cookies by default, or people have access to more than one PC, or family members share one PC, and so on. Identifying individual users has become a real headache for researchers. (You may use this story to get some discount if you are outsourcing analytic work.)
Fortunately, although you don't get 100% accurate stats, both methods are accurate enough to evaluate the performance of your marketing campaigns. You can still reliably determine which campaign made a profit. Forget about getting accurate results; focus on interpreting the results.
Risk
However, using a cookie-based analyst has one big drawback. The code you must add on every page of your site may slow down your page's loading speed. Sometimes it takes an extra 6 seconds to load a page. Only the entry page would be affected, but the loading time of the entry page is the most important one. As broadband internet connection becomes more common, visitors get less patient. An extra few seconds of loading time could turn visitors away.
Saturday, April 28, 2007
How to Test
You know how important it is to test everything you do. But, you might wonder how to do it right.
Know Why
The biggest mistake people make is that they only figure out what worked better, but not why it worked better. For example, you can buy an A/B split test script and use it with different sales pages. After the test, everyone can see which page performed better, but few people think about why one page performed better than the other.
Preparation
To find out why, you need to carefully prepare each test. Conversion rates of sales pages will change by different factors (such as headline, colour scheme, composition of entire page, etc). When you are making different sales pages for the test, you can change only one factor. If you change the headline, the rest of the sales page must remain completely the same. This allows you to make sure that it is the headline that made the difference. If you change the headline and the background colour, you'll never know what made the sales pages performance different - headline or background colour. Moreover, if you change the headline phrase, you shouldn't change size or font colour. If you do, you can't tell what made the difference - phrase or font.
Learn
This is what professional copywriters have done. Because they learned what made the difference, they are able to provide their clients with good sales copies. At the beginning, you might not understand whether a particular change will make the sales copy better. But, if you keep testing, soon you will be able to tell what will make the copy better and what won't. And, you will be able to write sales copy like a real professional.
Know Why
The biggest mistake people make is that they only figure out what worked better, but not why it worked better. For example, you can buy an A/B split test script and use it with different sales pages. After the test, everyone can see which page performed better, but few people think about why one page performed better than the other.
Preparation
To find out why, you need to carefully prepare each test. Conversion rates of sales pages will change by different factors (such as headline, colour scheme, composition of entire page, etc). When you are making different sales pages for the test, you can change only one factor. If you change the headline, the rest of the sales page must remain completely the same. This allows you to make sure that it is the headline that made the difference. If you change the headline and the background colour, you'll never know what made the sales pages performance different - headline or background colour. Moreover, if you change the headline phrase, you shouldn't change size or font colour. If you do, you can't tell what made the difference - phrase or font.
Learn
This is what professional copywriters have done. Because they learned what made the difference, they are able to provide their clients with good sales copies. At the beginning, you might not understand whether a particular change will make the sales copy better. But, if you keep testing, soon you will be able to tell what will make the copy better and what won't. And, you will be able to write sales copy like a real professional.
Test by Yourself
Many internet marketing gurus talk about test results. Can you trust those results?
No Way
One day, I read an article about the performance of different ad formats on Google AsSense. Basically, it says the borderless (border colour is the same as the background colour) ad format performed better than any other ad formats. Before changing the border colour of all the ad units on my site, I decided to test the findings myself. Below is how I did it.
I picked my 8 most read articles.
I assigned a unique channel to ad units on those pages. (Ad format: 336x280 text only)
I waited for a month and checked the click though rate. (I did this for each border colour: blue, gray, and borderless)
Surprisingly, the results showed that the borderless ad format didn't perform well on my site. The ad format with a gray border had the best performance.
What's Going On?
Why didn't my result confirm the other guy's result? It's simple. Dfferent ad formats suit different webpage designs. The design of my webpages and the other guy's webpages differed. So, the same ad formats shouldn't work equally well on both websites.
Get Your Own Results
Don't take another person's test results. Test things yourself to find out what works best on YOUR site. Yes, testing takes time. But remember this: If I had taken the other guy's words and started using the borderless ad format, I would be making less money today.
No Way
One day, I read an article about the performance of different ad formats on Google AsSense. Basically, it says the borderless (border colour is the same as the background colour) ad format performed better than any other ad formats. Before changing the border colour of all the ad units on my site, I decided to test the findings myself. Below is how I did it.
I picked my 8 most read articles.
I assigned a unique channel to ad units on those pages. (Ad format: 336x280 text only)
I waited for a month and checked the click though rate. (I did this for each border colour: blue, gray, and borderless)
Surprisingly, the results showed that the borderless ad format didn't perform well on my site. The ad format with a gray border had the best performance.
What's Going On?
Why didn't my result confirm the other guy's result? It's simple. Dfferent ad formats suit different webpage designs. The design of my webpages and the other guy's webpages differed. So, the same ad formats shouldn't work equally well on both websites.
Get Your Own Results
Don't take another person's test results. Test things yourself to find out what works best on YOUR site. Yes, testing takes time. But remember this: If I had taken the other guy's words and started using the borderless ad format, I would be making less money today.
Subscribe to:
Posts (Atom)