Recently, someone wrote me and explained that he was launching a new site, and all the advertisers that he was interested in using (except Google) had pay per impression plans. Old fashion advertising, you might call it. That’s the way magazines charge for their ads, and they audit their circulation to prove that they are faithful to their rates.

He did a comparison of pay per impression to pay per click, and then wrote me back:

Let’s say I pay $25 per 1,000 impressions. If spend $500K, I should get 20 million impressions. If I get 4% click through, that is 800,000 users on my site that I may or may not convert to purchasers. If I pay $1.5 per click, I get 333,333 users that may or may not convert. Now, I believe all number to be within the industry standards but as you can see, pay per click looks terrible unattractive. What am I missing?

Since he kept the 4% click through the same for both scenarios, it’s not worth quibbling over what the actual CTR will be (and I mislaid my crystal ball, anyway.) The real issue, I wrote in my reply, was that he was paying too much for the click. At $.625 for a click, the two models are financial identical. (So, I suggested a few ways to decrease his cost per click.)

This leads right into the question, how much do you pay for a click? It’s an easy calculation, but to do it well, you need to understand your profitability, your average order size, and your conversion rate.

Let’s say your widget sells for $100 and has a gross margin of 45%. (I use the word “widget” loosely - you can use this for a service too. Or, scale it up by adding zeros and use the same equations to value PPC bids for your million dollar enterprise software license.) 1.5% of all visits to the site turn into customers, and they purchase, on average, 1.2 widgets when they convert (so the average order size is 1.2*$100 = $120)

To figure your maximum price per click:

Average Order Size x Gross Margin x Conversion Rate = Maximum Price/Click
$120 x 45% x 1.5% = $.81

Explanation: Average order size times gross margin percentage gives you the average gross margin per customer. When you multiply that by the conversion rate, you now have average gross margin per visitor (instead of, per customer). When your average gross margin per visitor is the same as the amount you are paying for each click, you break even. But just.

Robbin Steif
LunaMetrics

At Pittsburgh’s Web Analytics Wednesday, I told the story of this post whereby I mistook someone’s email address for an alternative website address. Dan Halpern of Duquesne University said, “Why do people circumvent the bots with ‘myname at gmail.com’ text, when they can just encrypt their email address?” He then sent me a link to this email encoder site. (It looks like there are a lot of encoders out there, just type in spambot encoder.)

I tried it and it works (but I don’t know if bots caught onto this one long ago.) Here’s what this particular encoder says about themselves:

This form will allow you to encode your e-mail address through the use of Character Entities, transforming your ascii email address into its equivalent decimal entity. Simply enter your regular e-mail address in the first text box, click the encode button, and then highlight and copy the resulting code produced in the second text box. This encoded e-mail address can be read and translated back into its original ascii text by almost any web browser without any further action on your part. Just replace all instances of your e-mail address on your pages with the code, and you won’t have to worry about spam lists.

Robbin Steif
LunaMetrics

Since Google makes news faster than other Internet companies, almost everyone already knows about Google Trends, announced two days ago. In case you’ve been busy doing real work over the past two days: it enables you to type in up to five keyword phrases and shows you the relative number of searches done for those phrases on Google (but there are no numbers, so it is a graph without coordinates.) They also show major news events for those terms and do bar graphs for various countries.

Here was my first graph:

I was very surprised to see these results, because on Trellian’s Keyword Discovery, which captures a full year’s worth of data but only samples about 1/11th of the Internet, I got these results, showing “embedded linux” to be the larger term:

It’s possible, I thought, that Google is picking up all searches that included the term (as if they were doing what Google AdWords calls broad match), because KeyWordDiscovery shows many more derivative terms containing cygwin. However, that guess was wrong wrt Google Trends: if you want a broad match, you have to separate your terms with a pipe like this: (embedded linux)|cygwin.

The real issue seems to be about pitfalls that a service like KeywordDiscovery faces when they sample, and the skew of various kinds of words. KD’s FAQs say that they do use Google as part of their sample, but they also use many other engines. This probably works well for them for most situations except for the one I happend to try. After all, if your company is involved with embedded linux, or you want to understand the market or the competition, you may easily be a manager who could use Google or Yahoo or MSN, typing in embedded linux. On the other hand, my limited experience with real geeks (the kinds who would type in cygwin) is that they use Google almost exclusively.

Robbin Steif
LunaMetrics

When I was at Jared Spool’s UIE Roadshow, Christine Perfetti talked about sites that load quickly vs. those that load slowly. They found that the user perceives a site as loading quickly if she finds what she needs.

Having said that, if your site loads too slowly, the visitor will never have a chance to find out. I know you are reading this and saying, sure sure, I know this already — but do you really?

Today is Thursday and Mother’s Day is Sunday. So my sister called me from Florida and we looked together (in a figurative sort of way) at every jewelry site in the country where our mother lives (we figured, if we buy it there, she might actually get it in time…) Since we were doing this by cell phone and during the work day, we just wanted to find a great pair of earrings for less than $300 in under an hour.

We must have looked at every problem a website could possibly have (including a site that had the most gorgeous product but no contact information or order information at all.) But the worst problems were the sites that loaded so slowly it was painful (and I have a fiber optic connection, so the wait is never at my end. I write about my fios connection so much, Verizon should pay me for the commercials. But I digress.) The worst were the sites that loaded slowly and then still didn’t have the information easily accessable (or in plain view. Or a search box.) So I guess I need to point out that Christine was right about this one - waiting and then not finding what you want is painful.

I used to pretend to be sympathetic when customers showed me their flash intros, as I gently tried to persuade them that the intro is so 1999. But after today, I have no sympathy. It has to load fast and show the visitor what they want. Really.

Robbin Steif
LunaMetrics

As everyone knows by now, teenagers spend their time on MySpace, which is a cross between Blogger and LinkedIn. And lots of parents know that their teens use stealth techniques to hide their MySpace space.

While researching an article on measuring consumer generated media, I found that IceRocket has a special myspace finder, that you can use to find anyone’s MySpace. Ostensibly. Like most CGM tools, it isn’t great. I did find my daughter’s account but didn’t succeed in finding the MySpace of two of her friends, even after she came over to my computer and typed their names in to be sure they weren’t misspelled.

Can’t wait till these CGM measurement tools actually work more often.

Robbin Steif
LunaMetrics

There’s a new keyword based RSS search engine, researchbuzz, and I wanted to submit my blog. So I went to their URL, and found the “Submit your site” link prominently placed in the left nav bar. I read all the instructions and then thought, “Now where do I submit?”

So I went back and read all the instructions again, and realized that there was an alternative site one had to use, and they hadn’t even hyperlinked to it:

Feel free to send information about your site to sitesubmit-at-researchbuzz.com

OK, I thought, clicks are precious and they don’t get it, but let’s get listed and get moving. So I cut and pasted the URL into my Firefox browser (which insisted on adding the www) and got page not found. I tried taking out the w’s, but nothing. Maybe it doesn’t work on Firefox, I thought, so I pulled out IE, tried it again, and still, page not found.

So it’s obvious how to make sure no one clicks on your links: Don’t hyperlink, and be sure that when you include them in text format, they are wrong.

Of course, it is possible that they really don’t want to get submissions, they just want the verisimilitude of customer service. But I doubt it.

Robbin Steif
LunaMetrics

“Bounces” , also known as one-time page loads, one time access rates, or single page entries, measure visitors who enter your website and leave without looking at a second page. Not all bounce is bad (maybe they did need to view your home page just to find your phone number.) However, when your bounce rate (single page visits divided by entries on that page) gets very high, you’ve got a strong signal that the page is not working for you — it’s time for user testing or to consider some various alternatives and do split path testing.

How high is too high? Conversion Chronicles editor (CEO?) Steve Jackson wrote a great answer on the Web Analytic Forum. He cautioned that the rates he cites are only based on his own experience.

Similarly, I looked at a site a couple months ago and evaluated their top four entry pages. The home page had a bounce rate in the thirties, and I was pretty comfortable with that. (In addition to all those customers who are just looking for your phone number, plenty of visitors stop by your site by accident.) But the top two category pages had bounce rates over 75%. It was clear that the pages were “broken” — at least in the eyes of most users.

You may find that your web analytics don’t calculate your bounce rate. I created this workaround last fall: total exits from that page divided by total entries on that page. It’s very rough — after all, people have to leave someplace, but it helps you when you are comparing similar types of pages, such as all category pages. (Note: I eventually found that this was the inverse of Hack #58 in Web Site Measurement Hacks.)

Robbin Steif
LunaMetrics

Read Part II, which leads you to a great Google Analytics “cookie creation” tool.

I really hate when analytics only allow me to remove my computer from the data by IP address. I have a dynamic address, as will more and more people in N. America soon (as we gradually catch up to Europe and Asia in the Internet Connectivity Competition.) So I can remove a range, or a piece of my address (”remove anything that includes the pattern…”) but those options will remove other users, too. There are software tools to block your own computer, but that doesn’t help my customers. It is particularly upsetting when little free packages can do what four thousand dollar packages cannot. But there is a reason why StatCounter and Blogbeat can easily exclude you from their analytics while you have to work at it with NetTracker’s software version:

Client-side (page tagging) analytics should always have the ability to work this magic by setting a cookie on the visitor’s (and your) computer. It then uses those cookies when you click the button that says, “Remove this computer from the analytics.” This is why, when you erase your cookies, you put yourself back into your analytics and have to repeat the process of removal.

Server-side (log file) analytics can never work this magic unless you pro-actively install your own cookies (and maybe a webserver plugin from the software company). They may the ability to do the plugin thing, but often just tell you to use an IP address when screening yourself out, so you have to dig to find out what their capabilities are.

Just because you seem to track first time vs. returning visitors doesn’t mean you are setting a cookie. Your analytics may merely do matching — when the IP user string from one session matches an IP user string from another session, it marks that user as returning. If the individual had chosen, say, to use the same computer but a different browser, the analytics would not see them as the same person.

Robbin Steif
LunaMetrics

A while back, I did a post on “How to make sure no one replies to your email.” That addressed the issue of personal email that you open up and it looks like a form letter. Today, I’m talking about the opposite problem: someone who writes a personal email (or what looks like a personal email) and then doesn’t take the time to write a subject line that reflects the mail’s one-to-one nature. A lot has been written about testing of subject lines, but that’s generally for mass emailing, where you can split the audience and see what pulls the best. I don’t see that much on how to write personal email, but personal email is often where the sales call begins, especially for high-ticket items and/or in business-to-business companies.

Bad example #1: Last fall, I was hoping to work with the marketing committee of the Web Analytics Association (I still am.) Another web analyst arranged for the co-chair of the committee (at that time) to get in touch with me, and I was probably told the co-chair’s first name, although I promptly forgot it. When the co-chair’s email came through, it looked like this:

From: Firstname Lastname (which I didn’t remember/recognize)
Subject: Time for a chat?

This was a very one-to-one email, but the subject line looked so pink (like spam) that I promptly deleted it. I am posting this example becaue the email came from a high-level executive at a large web analytics company, and I just didn’t understand how someone “in the business” wouldn’t know better.

Bad example #2: This morning, I got email that looked like this:

From: John0700@yahoo.com
Subject: Recommendation from John0700@yahoo.com

This also looked very pink, but my Thunderbird email client does an incredible job of deleting junk mail, so I took a chance and opened it. It might have been personal and it might have been mass generated, but here was the opening:

Hi Robbin Steif,
I was visiting your profile today on blogger.com and thought you could use this for your blog, “Increasing your Website’s Conversion Rate”. Its a good resource for sticking files up on your blog and it doesnt cost anything.

He then went on to give me a link to his site. {I made up this email address, although it is very similar to the one I got.)

Now, maybe this last message was mass generated. If so (and if, as it seems, he was visiting blogs and sending all the blog owners the same message), why wasn’t the subject line, “Your blog,” a subject that would mean something to me.

If I go to a networking event, say, and John Smith tells me about an associate whom I should write, I never send email with “Recommendations from Robbin” or “Let’s talk” in the subject line. The recipient will trash it, 1-2-3, because he doesn’t know me and my subject line is meaningless to him. I work hard to craft a subject line that means something to him, usually, “John Smith gave me your name.” After all, conversion starts with a click.

Robbin Steif
LunaMetrics

After I posted about SalesGenius, I received a long email response from Lisa Dilg at PeskettPR, who handles (I believe) the PR for the product. She wanted to make some corrections, so here it is — I have not edited it, I’ve only done her the favor of keeping her cell phone and email address off of the web:

Robbin,

Thanks for the write up about SalesGenius. Great review of the product – and with screen shots too.

We really hope though that you will allow us to correct a few misconceptions that you have about the product however. I would really appreciate your time in reading some information from us on these issues.

1) SalesGenius is not a phishing product, and CANNOT be used to phish. Users CANNOT track visitors outside of their own domain – as you saw, you can only track prospects on their own site. (The definition of phishing is the fraudulent presentation of a web site – this is not) It is entirely legitimate for business users to track visitors on their own web site.

2) Yes, SalesGenius does proxy the LINKs and text of our user’s corporate web site. This method makes it VERY EASY for a Sales rep to interact with and understand his customers, and his own web site. Unlike any other Web analytics tool, SalesGenius requires NO TECHNICAL knowledge, a key design point when serving a Sales audience.

3) We have over 120 customers using our product, and we have not received a single complaint from any visitor. Tracking visitors in this way is entirely commonplace and expected on both web sites, and physical storefronts like Department stores, etc. The goal is to provide better service. Visitors are opted-in, and can easily opt-out. It’s all standard practice.

4) SalesGenius helps customers too: by helping Sales Reps focus just on their most interested prospects, SalesGenius fosters positive interaction between Sales Reps and prospects who actually need help with a solution. Just Friday we heard about a $6K deal from one of our users who said “The customer and I never would have found each other without SalesGenius.”

SalesGenius delivers the power of web analytics to a wide, non-technical audience. Our goal is to deliver personal web analytics data directly to the people who need/want to help customers most: the Sales Rep. Web Analytics should not be bottled up in an ivory tower in Marketing, with reports delivered once a month, and ignored. It should be personal.

Thanks for taking the time to learn and write about our product, and we hope your audience will appreciate the spirit within which we are trying to democratize the web analytics marketplace.

–Lisa

Lisa Dilg
PerkettPR for Genius Inc.

She makes a great point about analytics being bottled up with marketing. But that Point #1 bothers me. If the sales rep added at the bottom of the letter, “By the way, you will really be going through someone’s proxy server so that I can see your every move,” would the individual click?

Robbin Steif
LunaMetrics