By the e-book: $7
Buy Online: $19.95
Note: The analytics world has changed completely since I wrote this book. Have a look at the web analytics section of my blog for the latest. Also take a look at the tutorials section.
So, you’ve finished your site. It’s launched, it looks great, and you’re ready to start taking orders, or getting the word out, or doing whatever it is that your internet-communications campaign is supposed to do. All done, right? Nope. The conversation’s just getting started.
What makes a good conversation? If all participants have something to talk about, and they’re polite, intelligent communicators, what keeps things going? The give-and-take. If you are at a party, chatting with someone, you’re constantly observing his response to what you say and do, and adjusting the conversation accordingly. You do this pretty much unconsciously.
You can do almost the same thing on your web site. By tracking a few basics, you can gather information about your audience’s response to your campaign, and adjust accordingly. Trust me: Your audience will love you for it because you’ll provide a more relevant conversation that feels as though it’s directed right at them. Nothing keeps you in a conversation like knowing it’s just for you.
Let’s start with the simple stuff. You can always measure how many folks visit your site, which pages they like the most (and least), where they come from, and how often they return.
No matter how you set up your web site, it’ll be stored on a web server. That server answers requests from visitors’ web browsers and delivers pages to them. And every request is logged in a file.
If you opened that log file in a word processor, it’d be pretty hard to read. Impossible, actually, unless you have a lot of spare time on your hands. Fortunately, there’s a whole category of software out there called web traffic analyzers. Some are free, some are outrageously expensive, but all will take the babbling randomness of your traffic logs and turn it into a sensible, readable report. Chances are you already have a tool like this at your disposal. If not, keep reading — I have a few recommendations at the end of this section.
You can observe without intruding. Every example and principle outlined in this chapter is completely anonymous, and won’t violate the privacy of those visiting your site.
Once you have access to your web traffic report, there are five statistics to focus on.
Hits. A hit is one download of one file from your web site. So, if your home page consists of three images, a style sheet (style .css, for example), and the page itself (let’s say index.html), then every time a browser downloads that page, five hits are recorded. On some sites, a single downloaded page can account for dozens of hits. Hits are a good measure of overall server usage, but while it may sound impressive to say, “My site got 10,000 hits this month,” hits don’t tell you much about site traffic.
Sessions. A session is one browser visiting your site, one time. So, if I come to your site ten times in one day, that counts for ten sessions. This is a better measure of traffic than a hit, clearly, but there’s a problem that I can illustrate with this example: If everyone in your office has your site set as their home page, then every time they start their browser, they generate a new session on your server. Think about it. That means that five or six people can generate a lot of sessions, which doesn’t necessarily give you an accurate look at your site’s appeal. It’s great to have folks come back to your site, but you need to know how many individuals see it, too.
Visits. That’s where visits — or more specifically, unique visits its — come in. A unique visit is one visitor coming to your site any number of times in a given time period. So, if I come to your site 300 times in a month, I’ll still only count as a single unique visitor for that month. You need to know how many individuals arrive at your site, because each unique visit starts a conversation, and each conversation is an opportunity. (Note that some reporting tools call a unique visit a unique session — it’s confusing, I know, but if you see “unique,” chances are you’re on the right track.)
Page Views. A page view is one browser downloading one page, one time. Combined with visits, page views can tell you how successful your conversations are. If you have 100 unique visitors a day, and only 100 page views, that’s a bad sign — each visitor is only viewing a single page of your site. If, on the other hand, you have 100 unique visitors and 500 page views, that’s better — it means that visitors are spending a little time looking around your site. Generally, a high page-view-to-unique-visit ratio is a good thing. Don’t get carried away, though — if visitors are viewing a lot of pages per visit, they may be getting lost.
Referrers. The most underused statistic, and the most important one, a referrer records when a visitor clicks a link taking him from another web site to yours. If a potential customer types “custom road bicycle” into Google, and then clicks a link to MorgansBikes.com, Google will show up in the traffic report as a referrer. Even better, most traffic-analysis tools will also show that the phrase “custom road bicycle” generated traffic to the site. That’s valuable intelligence — it tells you what your audience was thinking about when they came looking for you.
Many log analyzers provide other bells and whistles, too: Click paths, least popular pages, exit pages, and entrance pages may all be tallied and tabulated for your viewing pleasure. As you get a feel for web traffic reporting, you can put all of this to good use. But you can learn a lot with the five basics.
Morgan’s new site is live! With a clean design and easy access to lots of useful information, the site gets rave reviews. She’s receiving lots of e-mail inquiries from shop owners she knows, and from members of her local bicycle club. But she’s not getting a lot of love from the vast audience of cyclists who don’t already know her. She’s not sure why, so she takes a look at her traffic reports for the first month. Here’s a quick summary:
This sheds some light on the problem. The hit count looks great, but doesn’t mean very much. Visits and referrers highlight the issue. She’s only had 210 unique visitors in a single month. And 99 percent of her traffic had no referrer — that’s a fancy way of saying that 99 percent of her visitors came to her site by typing “morgansbikes.com” into their web browsers. Only two visitors found her through a link from another site. In internet marketing terms, she’s invisible.
Morgan digs a little deeper and looks at the most popular pages on her site: the top four pages are her home page, then three dealer-information pages. The issue becomes clearer now. The only folks who are finding her are dealers (and maybe a few consumers), because they receive her print catalog. No one’s finding her in search engines, or from other web links.
Morgan’s educated now, but not discouraged. She’s learned two valuable facts:
She knows why some folks are joining the conversation, and why others aren’t. She can use that information to refine her marketing campaign.
If you’re hosting your web site with a web-hosting company, or if you have an IT department at your disposal, you should already have access to a traffic-analysis tool. If you don’t, find a new hosting company (or a new IT person!). Analysis tools are easy to find and set up. A few good programs are:
Google Analytics: My personal favorite. The basic setup provides accurate measurement of sessions, page views, hits, and referring traffic. Google Analytics used to be Urchin. Google bought them and made the product free. Plus they added some whiz-bang new features, like automatic integration with Google Adwords. You can learn more about Google Analytics at www.google.com/analytics.
WebTrends: Now owned by NetIQ, WebTrends is one of the most popular web-log analysis packages. It provides the same toolset as Urchin — preference is based on personal taste. I recommend comparing them if you’re going to be buying a log analyzer.
Webalizer: This ubiquitous log-analysis tool provides the basics. It doesn’t do a good job of measuring unique visitors, and the referrer analysis leaves a lot to be desired, but many hosting providers include this package for free, and it doesn't require any changes to your site.
Or, you can hire a full-service internet marketing company and let them worry about the details.
The one thing you can’t do is ignore your web traffic statistics. Doing so is like talking solely to make noise, or ignoring the other people in the conversation. Conversation Marketing cannot work if you don’t observe at least basic traffic statistics.
Tracking how many folks come to your site, and where they come from, is great. It can provide you with a very basic measure of success. But the real measure of a successful marketing investment is the return — the ability of that campaign to deliver conversions. A conversion happens anytime someone answers your call to action.
Whether you’re selling, persuading, discussing, or entertaining, you have an end in mind — a call to action. Remember all of those calls to action in chapter 3? Each call to action is an opportunity to measure how well your internet marketing campaign is performing. Every time a prospective dealer looks for another dealer to contact, views search data, or, even better, contacts Morgan with a question or to request the newsletter, that indicates a level of interest beyond a simple glance through her site.
The next step, then, is measuring four different things:
That last one can be daunting, but bear with me, and you’ll see you can almost always figure it out.
The first step is measuring how often visitors respond to a specific call to action. Typically, a call to action means someone completes some kind of form, or downloads a file, or takes some other action and then lands on a page on your web site that they wouldn’t otherwise see. Here’s an example:
If a dealer contacts Morgan using her web site, that dealer fills out a form and clicks Send. They then land on a thank-you page.
That thank-you page is a unique web page that isn’t delivered to a web browser unless a dealer has completed the Dealership Opportunities form. So, if her web traffic report shows a page view for that page, Morgan knows that someone has completed this form. Voilà! By using the tools already at her disposal, she knows when someone responds to that specific call to action.
Wait a minute, you say. That’s an easy one. What about when someone checks dealer search data? They’re not filling out a form, then, are they?
True enough. But again, the only time someone will view the Dealer Searches in My Area page should be when he or she is figuring out whether there’s enough demand in the area to warrant selling the Road Special. Sure, folks might randomly happen by that page, but if Morgan’s site is well designed, consumers should stick to their side of the fence, and the flow of traffic to the Dealer Searches in My Area page should be fairly pure.
OK, fine, you say. But what about dealers who just pick up the phone after seeing the site? That’s a valid issue — you never record 100 percent of your conversions. But I always prefer to err toward the conservative, and so treating web conversions as your sole measure of success will play it safe. In a face-to-face conversation, some folks don’t let on that they’re happy/sad/mad about what you’re saying. Some do, though, and it’s prudent to watch for those responses. Same thing goes in Conversation Marketing.
Each response to a call to action has a cost for you. Whether you’re using a banner ad, or making pay-per-click bids on Google or Yahoo Search Marketing, or hiring an agency to build your web site, you pay money for that potential customer/voter/ member/devotee.
How do you quantify the cost of conversion? Sometimes it’s easy. Sometimes it’s not.
Easy. You’re selling a product. Each product sells for $40, retail. It costs you $10 to make, stock, and ship each one. Your web site cost $10,000, and your advertising is another $1,000 per month. Your site sells 80 items in one month. Loosely figured, this comes out to:
$833.33/month your site ($10,000 divided by 12 months) $1,000.00/month advertising
$1,833.33/month total for internet marketing
$1,833.33/80 items = $22.92 cost to you per sale
Since you earn $30 per product sold, you’re getting by.
Less easy. You’re collecting leads — either for callbacks to sell a product or to register folks to vote, or for something else. You can’t necessarily attach each conversion to a specific profit or value — if you’re trying to acquire members or persuade someone, there’s not necessarily any intrinsic dollar value. But you can compare the cost of conversion on the web to your cost of conversion off the web. This doesn’t have to be exact, but here’s the gist:
Let's say that Morgan wants to compare her rate of new dealer signups from offline and online marketing. Here's how she'd go about it:
Offline Conversions: Road Special Dealers
$6,000/year: Magazine ad
$6,000/year: Interbike visit, hospitality suite, hotel
$1,000/year: Phone calls to dealers
$13,000/year total
100 dealer inquiries in 2004
$13,000/100 leads = $130/lead
10 new dealers acquired
$13,000/10 dealers = $1,300/dealer offline in 2004
Online Conversions: Road Special Dealers
$833.33/month: Web Site (see above — I’m being lazy)
$100/month: Pay-per-click ads
$933.33/month: Total for dealer marketing
10 dealer inquiries
$933.33/10 inquiries = $93.33/lead in the first month
2 dealers acquired online
$933.33/2 dealers = $466.67/dealer
The web’s not converting a lot yet, but the cost is lower, so it balances out. If you were measuring conversions and return on investment (ROI) for something that has no final value, you could stop here. But Morgan can take things a step further. She can keep track of where each dealer first contacted her (on- or offline) and see whether dealers contacting her through one medium or the other do a better job of selling her product. Then she can truly measure the return on her internet marketing investment, and adjust her campaign accordingly.
Finally, you want to know which specific online marketing efforts are generating responses. Are your e-mails garnering more conversions? Or those banner ads you bought on Yahoo?
This is one of the toughest areas of internet marketing. If you’re a web geek, you can build a measurement solution of your own that’ll track conversions back to individual sources. But there are a lot of pre-built tools out there these days that can do the job for you.
Yahoo Search Marketing has a built-in conversion tracking tool. With a little cutting-and-pasting you can measure whether your pay-per-click campaigns are generating conversions, and the cost.
Google AdWords has their own solution, of course, that lets you measure conversions and ROI for AdWords campaigns. they also have Google Analytics, which is free, but invitation-only at this writing.
Google Analytics lets you track conversions and cost per conversion, right down to the search engine and keyword. It also lets you track conversions from banners, links, and other assets.
And, of course, you can always hire an internet marketing agency to not only track conversions but analyze and provide advice based on that data.
Which tool’s right for you? If you’re on a tight budget and you’re carrying out a pay-per-click campaign on Google or Yahoo Search Marketing, stick with their tools. But Google Analytics, and tools like it, will let you keep a more complete picture of what’s driving conversions to your site, because these more complete tools track natural search, pay-per-click, banners, and unexpected lead sources, like that link John Doe just added to his fan site. And you can likely find a web hosting company that already has Urchin plus the campaign module, and save yourself some money.
Morgan knows her dealer-marketing efforts are doing pretty well. But her consumer marketing has a flat tire (sorry, couldn’t resist). So she runs a few low-cost pay-per-click ads on Google, sets up Google Analytics, and starts tracking performance.
She tracks how many consumers complete the Contact Me about the Road Special form on her site, and how many complete the Find a Dealer Near Me form, too. Every time a consumer completes one of those forms, it’ll count as a conversion. After one month, her stats reports show her the following for AdWords:
Road Special ads received 1000 clicks at $.10/click, costing her $100. 5 of those clicks became leads, for a cost of $20/lead.
Custom Road ads received 1000 clicks at $.50/click, costing her $500. Only 5 of those clicks became leads, for a cost of $100/lead.
Whoa. Those “custom road bikes” leads are costing her a fortune. She only sold three Road Specials last month. But she still doesn’t know whether those customers came from either of these PPC ads. And while $20 per lead is pretty steep, she makes about $2,000 per bike. It’s probably worth continuing that campaign.
Morgan decides to keep the campaign going, but reduces her AdWords bid for “custom road bikes” to $0.20. That reduces the clicks by a third, but cuts her cost by over 50 percent. That’s good.
By keeping track of cost per conversion, Morgan can control her marketing costs, and funnel resources (money) to what appears to be working. She can make sure that these conversations are worth having.
The final step in observing is to gather business intelligence you can act on. That’s a fancy way of saying, “Learn things you didn’t know before.”
Here’s an example: With a little extra programming, Morgan’s web guru can add a log of user information — zip code, state, and city — each time a user searches for a dealer. Then she can use that information to guide her marketing in specific parts of the country. Say she discovers that Princeton, N.J., is a hotbed of Road Special interest. She can advertise in the local paper, or even fly there for a special presentation to bike clubs. There are many opportunities to capture this kind of information on your site:
Be creative. Web sites, and internet marketing campaigns in general, are an invaluable source of intelligence that most folks ignore.
In any conversation, the more you can observe about your audience’s responses, the better you can refine your message (and know whether the conversation’s even worth having). In Conversation Marketing, watching those who come to your site, what they look at, and whether they take a desired action lets you do the same. As I said before, the internet offers a unique opportunity to make this kind of observation. Make the most of it.
So far, I’ve talked a lot about observing, and not very much about adjusting. Once you start collecting statistics, you have to do a little analysis. Some adjustments are obvious:
Some are subtler:
The possibilities are endless. There are three basic rules, though, that you can apply no matter how simple or sophisticated your web analytics efforts may be:
Over time, you’ll learn the patterns and flow of traffic on your site and be able to recognize when something has a good or bad effect. Then you can adjust accordingly, and keep the conversation going.
A lot of people want competitive intelligence; they want to see how their site’s doing compared to the sites of their competitors.
It is possible to get this kind of information, but you have to take it all with a grain of salt, and sometimes it can cost a pretty penny. I promise not to use any more clichés for at least two paragraphs.
For free, you can use Alexa.com. At the time of this writing, Alexa distributes a tool bar that lets folks do more convenient searching from their desktops. In exchange, Alexa collects data as to which sites those users visit. Alexa then provides that data on its search engine. If you go to Alexa.com and search for a web address, the search results will include a See traffic Details option. Then you can see the relative traffic rank for that site. By searching on your site and then on your competitors’, you can get a very general picture of how you’re doing. I say “very general” because Alexa only collects data from their toolbar users and users of their search engines. So this isn’t a global picture. But it’s a good start.
If you have some spare change, consider Hitwise.com and comScore.com. Both use broader measurement techniques to get you a more comprehensive measurement. Each company has something to recommend them, and I’ll leave it to their salespeople to convince you. The minimum you’ll pay, though, is around $750 for a three-month report.
