Analysis of performance requires you have the information in the first place. The purpose of this article is to describe the framework you should have in place if you want to assess the performance of your online activities. The aim is to provide you with a simple checklist you can use to ensure you are covering all the bases.
Your online market is made of individuals. Your relationship with these individuals goes through a series of stages. The first stage is “discovery,” in which they learn of your site’s existence. Online advertising, listings in search engines, and being linked to in a friend’s email are all examples of ways in which people can discover and access your website. When people first arrive they quickly scan the site to decide whether to stay. This is the “scanning stage.” If they decide to stay, they enter the “interaction stage,” interacting with the site, reading pages, putting items in shopping carts, and so on. At some point you’ll have a target action you want them to do in order for the visit to count as successfully complete. This stage is referred to as “acquisition.”
You should have processes in place to measure each stage. Between each stage is a decision point, in which the individual must decide whether to proceed to the next stage or leave. You should be measuring these decision points. Together, the stages and decision points constitute the “conversion funnel.”
You want to get more people into the funnel, and increase the percentage of those who complete it.
Assessing the discovery stage involves counting impressions (how many people were exposed to information about your site) and click-throughs (the number of people each item of information sends to your site). Click-throughs are best evaluated as a percentage (of impressions). Where a source of traffic is costing you money, you should divide the cost by the number of visitors to determine the “cost per visitor.”
You will need reports for every possible traffic source, plus aggregation data. There are different ways people can get to you, and the mechanisms for assessment depend on these. Systems that get paid for each visitor they send you will do their own counting. These include affiliate programs and PPC ads in search engines. You should gather your own numbers in order to check theirs. Their numbers will be higher than yours. There are many technical reasons why absolute precision is impossible, but any variation that exceeds 5 percent should be questioned. Some major outlets can over count by up to 25 percent until questioned.
You can’t count incoming ad traffic simply according to the site it came from. You need to know which ad was responsible. This requires that each ad has a unique identifier of some form. The commonest form is a “URL parameter” added to the link to the site. For example, if we were running an ad against the phrase “used cars” in Google and Overture for Dan’s Autos, we might use “www.dansautos.com?src=usedCarsG” for the link in Google and “www.dansautos.com?src=usedCarsO” for Overture. That way when http://www.automuseum.com/ sends a visitor to Dan’s site with the parameter “src=usedCarsG” I know which ad and which outlet was responsible.
By comparing the performance of the ads across multiple sites you can assess their quality. By comparing the performance of sites across multiple ads you can assess the value of the different sites.
PPC advertising and affiliate programs will involve ads in other people’s sites which you won’t have heard of. You need to know that a visitor came from an ad, not ordinary copy in the site, and you need to know which ad.
Similar logic applies in email marketing, including newsletters and direct email campaigns. The link to your site in an email should contain a unique parameter that identifies which email sent you the visitor. Most email marketing software will do this, but you won’t always get the stats if you don’t ask for them.
Search engines will also send you free visitors via their listings. You should be getting stats regarding which search engines have done so. Be sure to separate native listings from ads. Native listings won’t have additional ad-specific parameters associated with them.
It should be possible to get lists of the phrases people searched for to find you. Try to get this information across all search engines. You need to run search positioning software which will report how your listings stand in these search engines against those phrases. If there are other phrases you think you should be getting traffic for, but are not, you may find you have low ranks in the search results for them.
With these processes in place you should know how many people are coming to your site, where from, at what cost, and why. If your spend is below budget, improve your ads or put more out. If you’re pushing budget, focus on reducing cost-per-visitor. If you’re not getting satisfactory results from native listings, ramp up your search engine optimization program.
Scanning is easy to overlook, yet it is the most important process in your site. It costs money to get someone to your site. That money is wasted if they leave without looking at it. Designers talk about the “eight-second” rule, which reflects the fact that most visitors will spend about eight seconds scanning the first page they see before deciding to leave or stay. Only after that decision is made will they actually read what’s on the page.
Scanning analysis is about “entry pages.” You should get a list of your entry pages regularly. Look for surprises in the list. The new Microsoft search engine loves pages that are text-heavy. You’ll be surprised how many people are now entering your site from MSN through your privacy page.
Your primary concern is each entry page’s “bounce rate” -- the percentage of readers who left without entering the site. Experience teaches that the bounce rate for ads is lowest if each ad has a dedicated “landing page” assigned to it. It reinforces the message in the ad, and demonstrates how the site delivers what the ad promised. Its sole task is to persuade someone to enter the site. You can often improve the ROI for any given ad by tweaking its landing page.
Once people are in the site and using it, measurement involves assessing what pages they read, and for how long. In general you want to know the average duration and number of pages in a visit. You also want to know what your “exit pages” are. These are the pages they were looking at when you lost them. Your aim is to establish why and take steps to stop it. You need to be able to review individual visits in order to do this. You can then take a sample of exited visits to look for patterns.
The things you look at here depend on your site and what you’re trying to achieve. The critical thing is to have expectations. Does what you’re seeing match what you’d expect? If not something is wrong.
Acquisition usually means completing a form of some type, such as a subscription form, or a contact form. In online shopping sites it’s the credit card details form.
You need to know how many people saw the form, and how many of those filled it in. The percentage of people who view a form but don’t complete it is the “abandonment rate.” Acceptable rates vary. The average abandonment rate for shopping is 40 percent. A dedicated contact form should have a rate below 30 percent. The usual reason forms get abandoned is they ask too many questions or are confusing.
You also want to know how many people are getting to the form in the first place. The percentage of visitors who see the form is the “prospect rate.” There’s not much point having a great form if no one can find it.
Many sites require completing a sequence of forms in order. Online quotes for insurance and loans are the most obvious, but even a simple shopping system usually involves a form for name and address followed by another for credit card information. In such cases you need to monitor each step in the sequence because you will lose people along the way. You need to ensure you don’t lose too many, and that no single page stands out as a problem point.Comparing Stats
All activities can be put on a comparable basis by looking at their overall conversion rate -- how many people who hit that point eventually completed the required form. The conversion rate of a particular ad can be compared to the conversion rate for a search listing or a link in an email. You will find that certain outlets have a better conversion rate than others. Analyzing these will help you understand what it takes to attract the best visitors.
Here’s a summary of what needs to be measured to assess your online activity properly:
1. Being found:
a. Individually identify each source of traffic. If a referring site has more than one link to you, each link should, as much as possible, be identifiable. This is essential for PPC and affiliates.
b. Search listings require knowing what the listings are and how the search engines are indexing you.
2. Getting them through the door:
a. What are your entry pages? What is each one’s abandonment rate?3. Moving them through the site:
a. Average duration, average number of pages?4. Acquisition:
b. Visit polling -- look at individual visits to assess behavior.
a. What percentage of visitors view the target pages (prospect rate), what percentage complete them (abandonment rate).
b. What are your exit pages? Is there a pattern to visits which fail?
Brandt Dainow is CEO of Think Metrics, creator of the InSite Web reporting system. Read full bio.