Jakob Nielsen is one of the big online usability experts. So I was surprised to read in his blog about pay per click extracting all the profits from a website. First, it seemed out of place and second — I think he’s wrong.

If you don’t want to click through to his blog, here are the basics. Companies pay up to the maximum that their margins permit for pay per click advertising. So, for example, if the product costs $100, the margin is 50%, and 2% of the people who click the ad make the purchase, the vendor can pay $1 per click — 100 clicks = $100, and he will have sold to two people, 2*$50=100. So he makes nothing and loses nothing. (So far, so good, but let’s remember that that is the most the vendor should pay — not the most that he actually does.)

Nielsen then goes on to say that, while you may increase your conversion rate and be able to make some money on the sale, your competition will get better too, ultimately driving up your price per keyword to the exact amount of your margin. He exhorts the audience to rely on alternative methods of driving traffic (and feels that organic search is not to be relied upon, since the engines can and do change the algorithms at any time.)

I was really shocked to read this. The notion that the bidding strategy on a PPC engine is perfect, i.e. your competition will force you up to exactly the price that pulls all the profit out of the sale, is as outdated as the idea that a company sells for a perfect price (the net present value of its future earnings.) Markets aren’t perfect and neither is competition.

Furthermore, Nielsen ignores the lifetime value of a customer. (Catalog companies pay way more than the expected marginal profit because they earn it back later.) If you are sitting there saying, “What lifetime value?” (online buyers are notoriously fickle), you should read this post I wrote last week about the loyalty program Travelocity created to create lifetime value (as heard at shop.org).

Robbin
LunaMetrics

You know, they are having a Web Analytics Wednesday in Sydney, Australia. In Johannesburg. In Amsterdam. In Utah. (Well, in Utah they probably have Omniture Hour.) But here in Pittsburgh, no one seems to be interested. I realize that everyone is preoccupied with winning the Super Bowl, but life still goes on. I think.

Just send me email if you’re interested.

Robbin
LunaMetrics

Apologies to readers who are more interested in analytics than in conversion practices (or in this case, conversion and retention), but I really want to finish up on the Shop.org conference.

Jeff Glueck, Chief Marketing Officer of Travelocity, spoke at FirstLook/Shop.org about online customer loyalty. His first rhetorical question was a great one, what loyalty? Online customers are notoriously fickle, especially in the travel industry, where they shop around for the best deal. (After all, what difference does it make if you purchase your Delta ticket on Orbitz or Travelocity?) As important as loyalty may be to the average e-tailer, it’s way more important to Jeff. The online travel industry is already maturing, so increased sales come from a) stealing customers from your competition and b) keeping your competition from stealing from you, i.e. loyalty.

With the help of their analytics and an overpriced consulting firm, Travelocity learned that 15% of their customers drove 65% of their variable profit. But they were doing a lousy job of holding onto them.

First, the team learned that they had to earn customer loyalty. Visitors believed that travel sites are just software — they aren’t companies with real people who care about customers. So Travelocity created a little humanoid character as part of their branding. More importantly, they created a Travelocity Customer Bill of Rights, which guarantees you have a good experience with them, according to their site. Jeff gave a great example of how the Bill of Rights works: soon after it was launched, Travelocity accidentally sold $2M of tickets to Fiji at $7.50/ticket. It wasn’t clear whether the site or the airline was to blame for tickets which should have cost $2000.00 or so, but Travelocity quickly decided to honor them. The cost to a thin-margin travel site was overwhelming, but it was soon made up for by the amazing publicity Travelocity got on blogs and in the traditional press. Apparently, it was the internal publicity that really meant the most to them — employees wrote the CEO and said, “I’m really proud to work for this company now.”

Finally, he talked about retention via experience. After you book your trip, they come back to you with all sorts of ideas, some of them free. There’s the standard, “Do you need a hotel?” but they also have “Get your free Las Vegas survival kit” courtesy of Travelocity and ShowTickets.com. When you come home you can print a limited number of your digital or chemical pictures on Snapfish for free. And so on.

At lunch, I complained to Jeff about one of their services — they keep shopping for you. “Look,” I said to him, “I surfed your site for tickets to DC but bought them on Orbitz, the same day. For two weeks, you kept coming back to me with email about fares to DC.” He pointed out that, while I found it irritating to get email for a flight they couldn’t know I’d already booked, that kind of personal email converts better than almost anything else his team does. His comment is very much in line with what other retailers are finding about conversion and email: batch and blast converts really poorly compared to “You left this in your shopping cart” (which is what I had pretty much done.)

Travelocity came from behind (they were really hurting earlier in the decade) to post seven quarters of revenue growth in excess of 25% and an operating margin of 15%.

Robbin
LunaMetrics

So here I am in Atlanta, finishing up at the First Look/Shop.org conference.

I was pretty disappointed that there were no sessions on web analytics, although the analytic software vendors were here in force. (You should have seen the WebTrends ice sculpture at the Atlanta Aquarium.)

The holiday recap, moderated by Chuck Davis of Shopzilla, was among the most interesting of all the sessions. Shop.org did a multi-phase survey of 110 vendors and 2000 consumers to measure success of the 2005 online holiday shopping season and to see how well their predictions, made a year ago, held up. Davis reviewed the survey, and the season, with three analysts (Zia Daniell Wigder from Jupiter, Carrie Johnson from Forrester and Gian Fulgoni from ComScore). Three retailers joined him too(Steve Goldsmith from Amazon, Andrea Gulli from J. Jill, and Robert Myers from QVC. So one was a pure play retailer, one was a mixed channel retailer, and one was a pseudo-cataloger/e-tailer mix.) Here’s what Shop.org found and what the panelists had to say:

• Women continue to shop more than men online - almost two thirds of online shoppers are female.
• 81% of consumers have broadband, and dial-up has dropped by 11 percentage points in one year. (No surprise there - I found that when looking at the analytics of my customers.)
• Here’s a surprise: Over 8 out of 10 consumers shopped online at home during the holidays. (This one was debated by the analysts - Nielsen’s research shows at home/at work shopping split pretty evenly during the holidays. Customers who shop at work are often reluctant to admit their use of company time or company computers, skewing the results.)
• Entertainment, electronics, apparel and toys were the categories that consumers indicated they had purchased more of compared to the last holiday season.
• Cyber Monday, Nov. 28, made up only 9% of total online holiday sales. The biggest day was actually Monday Dec. 12, which comprised 20% of all holiday season sales. The analyst from Comscore concurred with that result, and the Amazon panelist, who was incredibly unwilling to share much hard data, did admit that Monday the 12th was the biggest day at Amazon.

Moving on to predictions vs. actual:

Prediction #1, Growth. Last year, the analysts predicted growth as low as 15% (Forrester) and as high as 26% (ComScore.) Actual growth of holiday online sales was about 25%.

Predicton #2, Customer satisfaction. Analysts predicted satisfaction to be about the same or lower. Shop.com reported that satisfaction hit record levels, especially among the customers that they interviewed right after the holidays. They pointed out that record number of orders were placed during the last week before the holidays, because customers trusted that the gifts would arrive. However, some of the attendees around me doubted that satisfaction was up, or at least, up strongly.

Prediction #3, Free Shipping. Earlier predictions were all over the board: a year ago, Forrester felt free shipping would rear its ugly head but few would provide it, while Jupiter thought it would matter more than ever. In fact, free shipping without conditions continues to be the most popular promotional offer for influencing conversion, followed by free shipping with conditions (e.g. minimum order size, free shipping on certain items, etc.) 74% of vendors plan conditional free shipping in 2006 (that number probably includes vendors with unconditional free shipping too, but wasn’t broken out.)

Prediction #4, Margins will suffer this year. In fact, 40% of surveyed merchants said margins stayed the same and 47% said they increased. So, if you believe the numbers, 87% did not see margins get squeezed. On the other hand, they only surveyed about 110 vendors, so all of 14 were willing to write, “My margins decreased somewhat or significantly” — now how statistically significant is that?

Prediction #5, Retailers will discount early but customers will still wait until the end of the holidays to shop. This looks like the prediction that hit the mark - about 30% of consumers started shopping online earlier this year.

Prediction #6, Paid search will be too expensive by holiday ‘05 so the real winners will be the retailers who are optimized organically. In fact, the retailers on the panel did not feel that keywords were too expensive, and the survey showed that 86% of retailers found that paid search drove traffic effectively, while only 44% found that SEO was most successful driving visitors.

Prediction #7, Pure online brands will hold their own but multi-channel retailers will post the highest growth. In fact, online-only had the stronger revenue growth - 77% of online retailers reported growth of 30% or more, while 70% of multi-channel vendors reported the same. (Now let’s think about the sample size. If they used only about 110 retailers, of which half were pure play and half were multi-channel, that’s 55 retailers per group. 70% of the 55 multi-channel retailers makes 38 retailers. 77% would be 43 retailers, a difference of five companies. How significant is that?)

Prediction #8, Christmas falling on a Sunday would encourage last minute online shopping. This seemed to have a little ring of truth to it — in 2004, 7% of customers had not started their online shopping until the last week, and this past year, 10% did not start until the last week. (The male Shopzilla moderator mused that men are guilty of waiting until the last minute. Despite my many years as a gift e-tailer, I will not comment.) Some of the panelists had interesting observations here, pointing out how the NYC transit strike affected promotions, for example. Amazon was able to move their standard shipping to Dec. 20 (i.e. the last day you could order, use standard shipping and still expect your gift to arrive on time) and J. Jill moved theirs (at their own expense) to Dec. 21.

Whew! I’ll have to blog more tomorrow.

Robbin
LunaMetrics

One of my loyal readers wrote me earlier this week. She has a professional services firm, and was using her website to create sales. Trouble was, the sales were few and far between, despite decent traffic. The conversion rate was about .3%. She asked for an analysis of her site.

Before I looked at her analytics, I made two important observations. 1) Her home page didn’t seem to reflect what her company did and 2) This was a scenario (professional services) where the customer usually likes to say, “I’m interested, tell me more,” but she was asking them to say, “I’m interested, I’m ready to buy.”
Then I looked at her analytics. This is what I wrote to her:

I am not sure that I am right about your home page [needing to be revised]. You have a somewhat high bounce rate — 44% of the people who enter on that page turn around and leave before
looking at another page. But it is not awful compared to other pages that I work with… Furthermore, the average customer spends 1:26 minutes on Home, which means that some customers spend significantly longer (after all, you have to have some people spending a lot more time to balance out the ones who turn around and leave within 5 seconds.) Also, [your most important referrer] shows a screenshot of your home page - if you change your page, that will be disorienting to those visitors…

But I think I am right about your ordering page, although it is hard to know if the issue is the length of the form or the fact that they have to actual order. While 8% of your visitors get to that page, only a tiny percentage push the submit button. About a third of them leave at that point, and the other two thirds poke around your site before leaving…

I suggested that next to the “Order now!” form, she create a shorter, “I’m interested” form.

She thanked me very profusely, but I wasn’t sure that I would hear from her again. Then I got this email:

Robbin,

I just wanted to say THANK YOU! I implemented your advice about having a separate contact form two days ago.Since then we have had [a number of] companies inquire about our services and partnerships.

So once again, I learned that intuition and analytics have to work together. On the one hand, it’s not good enough to think we (you and I) know the answer, it needs to be backed up with the analytics. On this case, I called one of them correctly and one of them wrong (a good batting average for baseball but perhaps not wonderful in the world of web analysis). On the other hand, analytics create a lot of data, so without best practices and intuition, and of course, key performance indicators, increasing conversion can sometimes be like finding a needle in a haystack.

Robbin
LunaMetrics

Since your best customer is the one you already have - and the customer who takes the trouble to bookmark you is an excellent customer — make sure that your web designer creates a favicon for your website.

What’s a favicon? It’s a tiny little picture, usually 16×16 pixels — probably smaller than the nail on your little finger. If you put code like this on your webpage…

link rel=”icon” href=”http://www.lunametrics.com/img/logo-url.gif” type=”image/gif”/

… your website will know to do two things.

1) Your visitors who are using Mozilla will see your favicon in the address bar with every page of your site — a nice form of branding
2) More importantly, if your customers bookmark your website, they will see your favicon next to your company name in their bookmarks. This is supposed to be true in Internet Explorer (although it seems rather flaky) and is definitely true in Mozilla. So just play customer for a moment and imagine looking at a long list of shopping booksmarks. You are most likely to choose the one that suits your needs the best, but if two are equivalent, your eye will be drawn to the one with the little “billboard” — the favicon. You increase your conversion rate by drawing back your best customers, i.e. the ones who have bookmarked your site.

(We never claim to be a techie blog, but you should know that if you put your favicon in a different directory, you need another line of code to point your website to it.)

Now, the interesting question is, can you measure bookmarking by measuring requests for your favicons file? Nothing that I have read makes me feel very confident about this ability, because browsers like Mozilla request your favicon for every page. Even when I read, “It’s the direction of requests that counts,” I just think that is mostly a function of page views and visitors, not necessarily bookmarks.

Well, even if you can’t measure with favicons, get out there and create some so that I can understand all these bookmarks in my browser. Please.

Robbin
LunaMetrics

Marketing Sherpa just released their “What I learned in online marketing in 2005.” They ask viewers not to cut and paste but since I wrote one of the snippets, I figure that I have the right to copy that one here:

Web analytics rock! We are now seeing customers wake up the fact that measurable is better. We also learned [in 2005] that Alexa data can be useful. We had a potential new customer who had been running an e-commerce site, for a very large company, with no analytics. Despite their lack of numbers, we compared their Alexa data to Alexa numbers from a group of our other customers whose traffic we really do know, and triangulated to create a Unique Visitor count. Using the e-commerce company’s actual sales and average order size, we computed a conversion rate of only .3%. When their new webmaster installed an analytics package, the company found that our guesstimate was within 10% of actual. Not only did we close the sale, but the CEO became a data head. They are now making decisions based on questions like, “If someone touches the “About Us” page, how much more likely are they to convert?

Robbin
LunaMetrics

I had a very cool experience integrating web analytics, SEO and pay per click this week. It was pretty textbook – except that we don’t often get to do things “by the book.”

One of my customers runs a continuing education course in the area of spray technology. The range of industries that sends students is truly amazing: from companies who spray chocolate on the inside of ice cream cone drumsticks to automakers whose fuel injected engines spray out the fuel.

This week, I sat down with four months of analytics. They aren’t fancy/expensive analytics, but for a ten-page spray technology website, they do the job. I learned that, due to the vast range of industries sending students, keyword searches are rarely duplicated. One person will type in “Electrospray painting” and the next visitor types in “Aerosol medication.” (Bet you never thought that the way inhalers deposit asthma medication was related to the way that automakers deposit painting on a car. But I digress.) I did find one or two words that were often included within keyword phrases and which will make excellent candidates for a broad search pay per click campaign.

I took every single keyword phrase and ran it through Web Position Gold, which shows how the site ranks for a particular keyword phrase. I only chose to see rankings on Google, MSN and Yahoo. I learned that I had a couple of Page One rankings, such as “spray atomizer design” but that most of my search terms were on Page Two, Page Three, or later. You might reasonably ask, if a term is buried that deeply, how did someone get to the site using it? There are various answers, but the best one is, they didn’t necessarily use one of the engines I measured, and not all search engines rank the same way.

Then, I took the site and ran it through a keyword density analyzer to see how the on-page and off-page content match (or don’t match) preferred keywords. While the one-word occurrences weren’t that interesting (who cares that I used the phrase “need” twice?), the two- and three-word occurrences were very helpful. For example, I found that I had succeeded in using the phrase “Spray technology” five times on the home page, counting the page title and description, but only twice on the “What you will learn page.”

Next, I opened up my Macromedia Contribute (and if you don’t have it and wish you could make changes to a site that someone else designed for you, I strongly recommend that you fork over the $149, or at least download their 30 day free trial.) I went through each page and everywhere that I could reasonably add the one or two words that were embedded in many of the keyword searches, I did. I was only able to make a handful of additions, but they may move us out of Position 19 into Position 9, since there aren’t that many people fighting for phrases like “rheology” or “non-Newtonian liquids.”

Finally, having

• used my web analytics to learn what phrases customers care about
• used Web Position Gold to learn where the site ranks in the organic search for the phrases
• used a keyword density analyzer to learn what words potential customers are currently typing
• used Macromedia Contribute to better optimize my site

…I sat down to use Keyword Trellian (or, if you like, the free Overture tool) to decide which keywords were worth paying for.

Neat, huh?

Robbin
LunaMetrics

I can see from my web analytics that I often get Pittsburgh visitors (because we are blogrolled at pittsblog.) I’d like to host a Web Analytic Wednesday here in the ‘Burgh, is there any interest? Send me email or reply to this post.

Robbin
LunaMetrics

I do lots of key performance indicator and dashboard work. Much of it is longitudinal (this is stat geek speak for “over time”), and customers say, “I see the little dots, what is the trend?” If you have a big gorilla package like Sitecatalyst, you can ask to see the graphs smoothed. But we are not all so lucky, so smart or so rich as to have Omniture on our
side - so I finally figured out how to do it in Excel.

Although Microsoft keeps the ability to do linear regression, even with basic Office XP, pretty well-hidden, it is trivial once you figure it out.

First, create the data, or use mine:

57
68
48
52
78
79
58
80
83
98
67
64
93

Then, highlight the data in Excel and choose the graph wizard with the little chart wizard icon that is on the top of your screen or just choose Insert>Chart. The graph you want is Scatter, like the picture on the left, and you want the subtype that has the little dots only (it will probably already be highlighted, as it is in this screenshot.)

Click Next, Next, Finish (although you can add any options you like along the way, such as a title, legend, etc.)

Now comes the moment of truth. While your graph is highlighted, look at your Excel toolbar, and you should see a new menu, Graph, probably to the right of tools. Choose Chart>Add Trendline>Linear>OK.

Excel will will do the linear regression and add a line showing the trend of your datapoints. It should look something like this the screen shot on the left.

This is somewhat of a statistics hack. After all, we don’t know how well the data fits the line, which a package like MiniTab would tell us. But it’s free if you have Excel. You can do it tonight on your computer.

Robbin
LunaMetrics