Personas

You create personas to measure certain statistics for your website. To create a typical persona, profile a user who fits the demographic information of your target audience, but customize the profile to fit a real person.

Here are a couple of sample personas: Jill is a 20-something female from New York. She’s a professional with a fairly large disposable income, but she doesn't drive. She reads through your website, and because it’s about classic-car customization, she doesn't find anything of interest to her, so she clicks away. Doug is in his mid-thirties, works for a real-estate firm, and has three cars of his own already. He wants to stop and take a look at your site and quite possibly subscribe to your newsletter.

But here’s the thing: Neither Jill nor Doug is real. They’re made-up people, or personas, created by marketing or usability firms in order to go through your website to see whether your site is properly targeting its demographics. A persona can give you an idea of whether your website will work for your target demographic. A firm often designs seven to ten different personas that are then used as a preliminary test market for your website. These are people from different age groups, socioeconomic backgrounds, and ethnicities, and they go through your site and allow you to gather data on whether your pages are working the way you want them to or whether you're turning off the very people you want to entice. If your audience is the go-getter type like Jill, a long meandering trip to the conversion point is going to lose her early on. But rushing someone like Doug could make him uncomfortable and cause him to bail out, leaving a shopping cart full of unpurchased goodies behind.

If your site sells shoes, a persona can help you more effectively target your market because you can keep track of whether Jill is going through your site and actually making a purchase, as opposed to hitting your site and leaving immediately afterwards. We offer a lot more information about personas and creating them in Book V, Chapter 1.

A/B testing

One of the most commonly used tools for testing your website usability is A/B testing. It’s like doing a science experiment. You test your old version of your website (Version A) with the new version (Version B) to see which one measures up better. A/B testing and multivariate testing (discussed in the following section) are somewhat complex, but we explain what they mean and how they can help. Afterward, we describe options to implement testing.

The big advantage of A/B testing is that you can send half your traffic to the page(s) with the proposed changes while sending the other half to the current page. That way, you can compare your current conversion rate for at least part of your site traffic in case some of the proposed changes aren’t working. A/B testing is often the best choice for a page with lower traffic.

But you can’t run off and do a hack-and-slash job on the test page and expect to get any sort of meaningful data out of it. Here are some guidelines to help you get meaningful, measurable results if you plan to run A/B tests on a website change or an email campaign:

• Change only one variable at a time. Figuring out what exactly is working for you is harder if you’ve changed several variables on the site.
• Figure out the precise process for diverting traffic. One of the problems in A/B testing is that some marketers don't understand how to divert traffic and don't get accurate traffic numbers.
• Establish accurate measures of volume. Doing a comparison test is hard if you don’t know how many people you’re testing.
• Look for significant differences. If you see a difference in the conversion rate for the B test, you need to ensure that this difference is significant. A miniscule change to your goal probably won’t be worth the effort, whereas a significant change will.
• Take the time to do a null test. A null test is a test that you run on two A-version pages (pages you haven’t made any changes to) in order to establish a baseline and make sure the traffic isn’t coming in weird. This is to make sure that half your traffic is going to one page and the other half is going to the other page, and that you have enough people going into the test.
• Run your test long enough to ensure results are real. You won’t get an accurate amount of data if you run the test only for a day or a week. Make sure to run it long enough to get enough data to do an accurate comparison, typically a month or more. Remember, with web analytics, the more time you take to do something right, the better your results are.
• Run segmentation tests. A segmentation test tests the variables in your incoming traffic (such as the demographics of that traffic) by asking users to answer a few questions. Really, you can test any variable as long as you set it up right. The more information you have on your different variables, the better you can target specific changes to your site to drive up your conversions.

We cover much more on the ins and outs of A/B testing in Chapter 3 of this minibook.

Multivariate testing

A/B testing is about measuring big changes to your site. It’s comparing the old site with the completely new version. Multivariate testing is about testing all those smaller changes to your site, like the change to a certain font, or to a button instead of an arrow. Typically, you test many small changes to the same page at one time instead of two totally separate pages, as in A/B testing. Multivariate testing works better when a page has a large volume of traffic. If you are testing a medium- or low-volume page, use A/B tests instead.

Most of the testing tools involve copying and pasting a piece of JavaScript into the code of the pages that you are testing. The control code on the top of the HTML page tells you that people are trying to load the page. The tracking script at the bottom of the code tells you that the visitors saw the page, and then you have another code on the conversion page (whatever page the users view after they have completed a conversion) that tells you they converted and what version of the page they were looking at. If you do a test, each version of the landing page has a unique sticker for you to identify it by. If you're doing the test with Google Analytics, after the test runs for a while, Google populates reports for you. Other programs work similarly.

Here are some quick guidelines to keep in mind when running your test:

• Test a small number of variations. The rule of thumb is less than 100 variables per combination of tested pages.
• Test big changes. If you can’t see any difference between two variations in eight seconds, your visitors probably won’t either and their reactions won't tell you anything. They can't react to what they don't notice.
• If conversions are relatively rare in your business, consider testing for early indicators. If you’re selling a $100,000 software package, for example, you won't have a high number of sales to test. Instead, optimize for conversion indicators such as request info, view product details, and so on.
• Don’t jump to conclusions. A two-week test is not enough time to gather your data. Run each test for at least one month, if not two.

Cookies

When we talk about cookies, we don’t mean a tasty sugary snack. Cookies are little files that get saved in your browser to keep track of information on a particular site. A cookie is what enables you to automatically log on to your Facebook account regardless of whether you’ve closed your browser session or even logged off and powered down your computer.

Once upon a time, a server would send out web pages when they were requested without recording any data on who requested the page, where it went, or any other associated user behavior. Cookies were created to save this information. Cookies are used to enhance the browser experience, improve usability for customer interactions, increase purchase behavior, and improve commercial website performance by keeping track of what the user’s doing.

Cookies are either first party or third party, depending on the type of website that sets them. A first-party cookie is set by the site that the user is visiting, such as www.classiccarcustomization.com. A third-party cookie is set by a third-party site, such as a web analytics vendor, that provides a service to the main website.

A first-party cookie can contain personal information such as username and a login ID so that users can be recognized when they visit a site. If cookies didn’t store this data, websites would have to request it every time the users returned to the site.

A third-party cookie tracks a visitor's path through www.classiccarcustomization.com so it can identify which pages work and which don't, helping optimize for better site performance. The ad network cookies track user behavior across multiple sites, helping them classify user behavior. This helps in the targeting of ads to user segments. For instance, frequent visitors of sports sites are given sports-relevant ads. Although anonymous, this multi-site gathering of visitor information has also caused some controversy regarding privacy violations.

Session IDs

Instead of using a cookie, you might be tempted to use a session ID. A session ID is a way of tracking users when they come to your website. Generally, we recommend that you don’t use session IDs because they are assigned no matter who the visitor is, including a search engine robot. This means that every time a session ID is used, it is possible that the search engines will treat it as a new page, and you'll wind up with duplicate content that mucks with your rankings in the search engine. Additionally, a session ID is not very useful when it comes to measuring your website usability because a session ID tracks that user only for the duration of his visit to the site. A cookie remembers that user when he returns, whereas a session ID doesn't.

Overall, using cookies to track your visitors is much better, if they have cookies enabled in their browsers.

Measuring marketing campaign effectiveness

The first thing you should look at are your marketing campaigns. It’s important to measure the effect of marketing campaigns on your website traffic. The following metrics are specific to marketing campaigns aimed at driving traffic to your site:

• Campaign conversion rate: Determines the effect of conversions from specific campaigns. Did the conversions rise due to the ads you placed on other websites or due to a grassroots viral marketing campaign, like These Come from Trees? (An environmental group asked its members to place These Come from Trees stickers in public restroom stalls in order to curb overuse of paper products — towels, toilet paper, and so on. These stickers included the URL of the organization's website, where it provided more information and stickers.)
• Cost per conversion: Figures the cost effectiveness for specific campaigns. You have a great idea for a marketing campaign, but giving away $20 bills stamped with your web address might cost more than the actual conversion you’re aiming for. Make sure that you can afford the campaign before you start it.
• Campaign ROI (return on investment): Determines the cost effectiveness for specific campaigns. Is your campaign bringing in the conversions you need, or are you losing money?
• Segment conversion rates: Tracks conversion progress over time. Your conversions most likely won't change overnight. Watch them over a long period of time to make sure that your campaign is effectively working.
• Percent of orders from new and repeat customers: Determines the effectiveness of marketing or customer-retention programs. You want to attract new customers, yes, but you also want them to turn into repeat customers.
• New and repeat customer conversion rates: Helps understand barriers to online purchases. One repeat customer is worth more than a new customer because not only do repeat customers mean future conversions, they also cost less than new customers because you don't have to spend a whole lot to keep them.
• Sales per visitor: Measures marketing efficiency. How much is someone likely to buy? How little? Get an average so that you can figure out how to budget your campaign effectively.

Here are some key metrics you should track, regardless of whether your site is e-commerce, research, or any other kind of website:

• The conversion rates for any process that makes or saves money or that is critical to the customer experience
• The campaign conversion rate for current campaigns or the most expensive campaigns, if you have a lot of them
• The cost per conversion for the campaigns you decide to monitor
• The segment conversion rates for key or critical group conversions

Here are some specific metrics that e-commerce sites should be tracking:

• The site-wide conversion rate (all purchases to all visits or visitors)
• New and repeat site-wide customer conversion rates
• The percents of orders from new and returning customers
• The average order value, site-wide and for new and returning customers
• Sales per visitor (compare to site-wide conversion rate)

After you decide which site-wide processes you want to measure and how to measure them, the following metrics can help you understand visitor success or failure. These metrics follow whether a customer stays, searches, or actually makes a conversion:

• Home to purchase: The abandonment rate for visitors going through the sales path
• Search to purchase: The abandonment rate for visitors coming from a site search
• Special offer to purchase: The effect of various merchandising and pricing options
• Lead generation: The abandonment rate when personal data is requested

Establishing site objectives or goals and all the parts that make up these objectives (the who, how, where, what, and why) is essential when tracking the conversions on your site. One of these factors could contribute to the success of your campaign — or just as easily derail it.

Building conversion funnels

After your site objectives are established, you can measure your progress through the use of a conversion funnel. In Chapter 1 of this minibook, we define the four basic websites: e-commerce, content, lead generation, and self service. On an e-commerce site, a conversion is obviously a sale. For a content site, it might be the number of newsletter subscriptions. Lead-generation sites try to gather information for later contact. Self-service sites are targeted at solving a customer's problems, so the measure might be time spent on the site.

In the conversion funnel in Figure 2-1, each step in the sales process on the way to conversion is fraught with visitor drop-off. (Steps in the funnel differ based on the type of business and conversion that you're seeking.)

Each block on the conversion funnel becomes smaller as you go down the sales (or conversion) path. This represents the amount of users you lose along the way to a conversion, for whatever reason.

The point of using a conversion funnel is to figure out where you are getting the most drop-off. In a perfect world, there would be no conversion funnels because all visitors to your site would perform your desired action and you would have a conversion column. But because this isn’t a perfect world, your main goal is for the drop-off rate to be as low as possible.

Measuring your website's conversion rate is a challenge because a number of steps lead to that final action, and sometimes visitors are thwarted in their quest to complete an intended action. You can hope that you lost them just because their browser crashed, but sometimes they simply didn’t find what they were looking for, or the site was too confusing, or it took too long for them to get to their objective — and so they left. Additionally, many sites measure only their final conversion rate. This does not give webmasters the opportunity to improve their drop-off rates by analyzing the sales path and finding the bottlenecks in order to make the site improvements that result in higher conversions.

Don't measure your end-result conversion rate without tracking the path that your customers take to conversion.

Preventing conversion funnel drop-off

In a typical conversion funnel, visitors drop off along the way to the final step that completes the sale or achieves the desired action. The good news is that when your analytics program (such as Adobe Analytics or Google Analytics, which we discuss in Chapter 1 of this minibook) tracks the micro-steps required to reach the final conversion act, it reveals data that can be used to prevent drop-off. The analytics package you have does the work and the analysis so that you don’t have to. Just be sure to implement the changes it recommends.

One of the things you can do is to eliminate all the unnecessary steps to visitor conversion to reduce the conversion funnel drop-off. The fewer steps needed for a visitor to convert, the greater the likelihood of a conversion. You should create an effective call-to-action for every step in the sales path. Your conversion rate reflects your ability to persuade visitors to complete their intended actions.

Analyzing your conversion funnel

Your conversion funnel is the path a user follows on your site on the way to a purchase. It’s important to follow the conversion funnel closely and analyze where you're losing the most people by percentage. It's very unlikely that 100 percent of your visitors will continue on through every step, but you do want a high percentage of visitors to continue on your conversion path.

Say you have an e-commerce site that gets 2,000 visitors per month, your site has a three-step sales path, and your average sale is $11 per item. If half your site visitors enter the sales path, that means 1,000 prospects drop off at the first step. A 50 percent drop-off rate at the first step could be due to an impediment such as requiring site registration. If 40 percent of that total drops off at the second step, and 30 percent of that group completes the sale, you have $1,980 in sales at a 9 percent conversion rate because only 180 of the original 2,000 prospects made a purchase.

When people drop off, they have not found what they were looking for on your site. By identifying high abandonment pages, you can take a closer look to see what might be making visitors leave and test for ways that would make them want to stick around and continue on the conversion funnel. By properly analyzing this data, you can make sure you won't lose as many people along the conversion funnel. More people convert, which means more money for you.

Making site improvements

Using the math from the example in the preceding section, if you can improve the final step of the sales path by just 10 percent, it would bring you an additional $198 in sales, upping your conversion rate to 9.9 percent. However, if you can make improvements at the first step of the sales path, reducing your 50 percent drop-off rate to 25 percent, you can increase your sales by $5,940, resulting in a 36 percent conversion rate.

However, if you do not know what to measure and why, or you haven't a clue as to what indicators to evaluate in your analytics reports, you can't take the necessary actions to improve your site performance. So take the time to figure out the data to analyze based on your site objectives, and then follow up on the data revealed through the use of your analytics software.

Simply picking out indicators that look good at first glance — such as increasing numbers of referrals from Google and Yahoo or increasing the number of page views — might not help you improve site performance. It's not that these numbers are worthless; they just might not be the right metrics to improve your site. Knowing the basic analytic principles ensures that you know what metrics to check for when making your business decisions.

In the preceding sections in this chapter, we talk about overall site objectives, but you also need to consider objectives for the individual pages within your site, which we discuss in the following section.

Assigning web page objectives

Assign individual objectives to each page, especially the ones that require the user to perform an action. Every page should be designed to have a user perform an action, even if that action is something as simple as clicking over to the next page.

To effectively implement this approach, every page on your website that requires action should answer the following three questions:

• What action is required? Examples are clicking to the next page, playing a video, or reading the text on the website.
• Who must take that action?
• What information does your visitor need to take the required action?

By answering these questions, you can define your objectives and apply good analytics solutions to test and optimize your pages for improved results. The same principles you used for site optimization can also work for page optimization.