Elsewhere: Google Analytics Visitor Counts by Hour of Day and Day of Week

I just posted Google Analytics Visitor Counts by Hour of Day and Day of Week on BoldInteractive.com.

Good news and bad news. I’m a pessimist realist, so we’ll start with the bad news. Bad news: Your web team says the site has to go down for maintenance. For several hours in the near future, your company’s presence on the web will be invisible to any potential customer. Good news: You get to […]

Read the whole article on BoldInteractive.com

How to Evaluate an Online Marketing Service

If you’ve got a budget for website promotion, I know about a hundred companies that want your business. Be it PPC, banner ads, link placements, paid blogging, search engine optimization, or any of a dozen other industry buzz words, they all have different strategies for driving traffic. With so many choices, it can be hard to know which ones are valuable. Answer the following questions, however, and you’ll have a good idea whether a service deserves your money.

How is the price of the service determined?
There are many different cost metrics thrown around in online marketing: CPC, CPA, CPM, flat rate, etc. Ultimately, the only important metric is how much the service costs versus how much value it delivers (i.e., ROI). However, different cost metrics elicit different quality concerns. With CPA, you have to be more cautious about the quality of conversions, with CPC, the conversion rate, and with CPM, the clickthrough rate. Always be on guard that the provider might be illegitimately inflating your costs.

How likely is the traffic to convert?
The question here is whether the visitors’ demographics and intent match your site’s conversion goal. What is the age range of the visitors? What are their interests? If you’re selling something, how much disposable income do they have and where are they at in the buying cycle? This will require testing to verify, but you can often get a good idea of traffic quality by asking where that traffic is coming from before it reaches your site.

How much volume can the service drive?
It’s possible for a channel to deliver a great ROI, but only at a low volume. If a channel doesn’t produce enough traffic and/or conversions, it may not be worth the trouble to manage in the first place.

How well does the service scale?
Business needs have an tendency to change. The best online marketing services can scale in cost and volume to meet those needs. Often, scalability is the key to retaining a service over the long term.

Does the service use affiliates?
Depending on the nature of your conversions and the cost metric involved, affiliates may be useful. For example, when e-commerce transactions are required, affiliates are generally safe. However, if your conversions are, say, form submissions, fraud becomes a chief concern. In situations like this, affiliate-based services are best used cautiously or avoided all together.

Is the traffic incentivized in any way?
As with affiliates, incentivized traffic may or may not be useful depending on the nature of your conversion and the cost metric involved. Generally speaking, though, you want visitors who are interested in your offer, not visitors who just clicked through to get a flat screen TV.

Does the service offer online utilities?
Speaking from experience, nothing is more frustrating than managing a service that doesn’t offer online reporting and management utilities. Services that lack online utilities are suspect, either because they aren’t willing to give you transparency and control, or because they lack the technical savvy to create them.

Does the service include a campaign manager?
Although management and reporting utilities are the ideal, when large-scale adjustments need to be made, a dedicated human being can help reduce your management overhead. Obviously, you should prefer campaign managers that are good at achieving your goals.

How easily can the service be terminated?
When it comes to online marketing services, a contract is almost always involved. Depending on your faith in the service, you’ll want to be sure that the contract can be terminated to minimize losses if the ROI turns sour.

A Value-Based Approach to Website Strategy

If you examine a website closely enough, you can almost always identify some form of commercial interest behind it. This is often overt, as in e-commerce websites or content sites with advertising. Sometimes, however, the commercial interest is more subtle. Ward on the Web, for instance, is a professional blog devoid of advertising. Because its purpose is to promote my professional standing, however, there is commercial interest in the usefulness and authority of my message. My strategy here is to show off what I know in the hopes that it will help me achieve greater career growth.

The first thing to realize is that a website is a tool for creating value. My very first post on Ward on the Web spoke about defining your site’s purpose, with the objective of building a website to achieve that purpose as well as possible. Does your website generate revenue? Does it attract attention? Does it build a reputation? Does it gather information?

Whatever your website does, it’s important to realize that the ultimate goal is always a resource that’s valuable to you. Money has obvious commercial value, but so does a large audience or a stellar reputation. It may be more difficult to tack on a dollar amount, but there’s no denying that even non-monetary resources have potential monetary value.

Viewed in this context, decisions about website strategy tend to become simpler. Say, for instance, that you’re writing a blog and toying with the idea of adding advertising. You’re afraid that this will turn off your visitors, reducing the size and responsiveness of your audience.

Remember, though, that your website is a tool for creating value. In this case, you’re examining the prospect of trading one type of value (audience size and engagement) for another (advertising revenue). Obviously, how much of one you’ll end up trading for the other is a complex question, but the basic proposition is simple. Do you stand to derive more value with advertising or without?

Even small website strategy decisions are best when advised by a value-based approach. For example, say you’re testing a landing page on a lead generation website. You want to know which of two calls to action are more effective. Generally, you’d do a simple split test (or multivariate analysis if you have other variables in mind). Let’s say your testing reveals that the first call to action converts 10% of visitors and the second converts 12%. Which do you use?

Of course you choose the second. But why? Because it makes your website a more effective lead generation tool. Because it adds more value. Thanks to analysis, this example is cut and dry; you simply take the greater of two values.

The important thing to realize, however, is that all website strategy decisions are value-based comparisons. Which provides more value, option A or option B? If you can answer this question with confidence, your choice should be obvious every time.

Social Media Experiment Update, Week Six

If you’ve been reading awhile, you may remember the social media experiment I started a few weeks ago. Well, after a month and a half, I figured it was about time to check on my Frankenstein monster and see how it was progressing.

Social network search performance over time

So far, most of the readings are negative. It’s not lurching around. It’s not terrorizing peasants. All it appears to be doing it twitching and grumbling on the table. Here are my findings to date.

  • Naymz was very effective. As you can see, my Naymz profile is the earliest and highest result of the lot. It even enjoyed a brief stint on page one during the second week. At the time of this writing, it still ranks at a respectable 17. This is especially remarkable when you consider how little effort it took to set up and maintain.
  • Facebook was also very effective. Although I enjoy a lot of competition in Facebook (419 results for a profile search with “Stephen Ward” in the name), it turned out to be one of the best performers. It never reached page one like Naymz, but it’s held its position on page two for several weeks now. At the time of this writing, it appears to be holding around #15. Granted, I had to put more work into Facebook than Naymz, and it doesn’t do much for links or branding, but it still proved to be fairly handy.
  • LinkedIn and MySpace could have worked. Much like Facebook, there’s generally a lot of competition for name rankings on these sites. More importantly, entries for both of LinkedIn and Facebook consistently ranked on page one or two. The only problem is that they were some other Stephen Ward’s profiles. If I could manage to make Google think that I’m the most important Stephen Ward on those sites, my profile would likely replace theirs.
  • Blogcatalog and Technorati were so-so. I mention them because they were the only other profiles to show up on the radar. Blogcatalog built up from page 15 to page four only to mysteriously fall off and never recover. Technorati languished between pages 13 and 18 before it also fell off. Neither performed phenomenally well, but I was surprised to see blog listing pages showing up at all.
  • Everything else failed. The down side is that most of my monster didn’t so much as twitch. I invested the most time in social news sites like Digg and Reddit and social bookmarking services like Delicious and StumbleUpon, but the profiles never showed up in the top 200. Even my underdog industry profiles, DZone and SEOmoz, didn’t make the cut.

I don’t think I’ll abandon the experiment just yet. As I already pointed out, LinkedIn and MySpace show promise. I’ll keep things going, redistribute my efforts, and maybe add a new part or two. Who knows? With a little patience, I may get my monster up and dancing yet.

7 Questions to Ask Your Marketing Analyst

“You don’t know what works until you test it, so test everything.”
Adam Schultz

Over the years, my marketing mentors have instilled in me the understanding that marketing is 10% creativity and 90% testing. Examined through this paradigm, marketing takes on an almost scientific nature, where every piece of data is scrutinized until the truth becomes undeniable from the weight of empirical evidence.

As you can probably guess, this involves more than a little knowledge of statistics, which may or may not be your specialty. Of course, analysis is pointless if you can’t understand it. Ask your marketing analyst these seven questions, though, and you’ll have a pretty good picture of your test results, regardless of their complexity.

  1. What do the results indicate?
    Here, you’re asking for the objective findings. The results might indicate, for example, that the first ad variation did significantly better than the second. The answer to this question should include cold, hard facts supported by statistical measurements.
  2. What conclusions would you draw?
    In contrast to the first question, here you are asking for subjective results. Over time, many marketing analysts develop a sense of what a data trend looks like. Thus, while there might be no empirically provable findings, your analyst will likely be able to offer his or her gut instinct as to what can be learned.
  3. Did the results meet with your expectations?
    Results that fall in line with expectations probably don’t require any additional scrutiny. After all, you just confirmed what you already suspected to be true. Results that fly in the face of expectations, however, often demand further questioning, especially if they go against the expectations of an experienced marketing analyst.
  4. How were these results produced?
    Assuming there is any real knowledge to glean from all the data, it’s worthwhile to question that data’s integrity. What sampling method was used? How large was the sample size? What sort of statistical comparison was performed? By asking your marketing analyst about the methods used to obtain the data, you can often get a sense of its reliability.
  5. How much do you trust the results?
    Much like the first two questions, it’s always worthwhile to switch back and forth from hard facts to gut instinct. After all, the research methods and results may sound reasonable even while your marketing analyst has shortcomings about them. Always be sure to ask his or her opinion on the data’s legitimacy.
  6. What next actions would you recommend?
    Based on the conclusions and how much trust you place on them, you probably have some ideas of how to move forward. Then again, your marketing analyst probably has some good ideas, too, especially if further testing is called for.
  7. What do you predict will happen if we take these actions?
    As the scientific method dictates, research is cyclical. Now that you’ve figured out what the data indicates, how much you trust the findings, and what you want to do with that knowledge, it’s worthwhile to speculate what those actions might produce. This feeds back into question three when you’re going over the next round of results.

IE to Lose Dominance Among Web-Savvy Users by Q3 2009

If I had written the script for Pinocchio, Jiminy Cricket would have sung, “Let your intellectual curiosity be your guide.” It’s part of my appeal as an analyst; I just can’t help but try to answer any question that crosses my path.

The most recent item to pique my curiosity was the browser statistics chart at W3Schools. This chart shows the share of users surfing W3Schools with a particular web browser month by month over the past six years. Obviously, the chart itself doesn’t do the data much justice. When you graph it out, however, it paints a very interesting picture.

W3Schools Browser Usage Over Time

There are a lot of interesting points in the chart above. You can clearly see how IE 6 has grown, plateaued, and declined as the dominant browser over the last few years, at once taking share from the IE 5 user base and then ceding it to the IE 7 user base as one would expect. You can see how Netscape gave way to its successor, the Mozilla Suite, in 2003, and how the Mozilla Suite in turn gave way to Firefox when it launched at the end of 2004. At the same time, you can see how other browsers like Opera and Safari have struggled to even show up on the radar.

Perhaps the most striking trend, of course, requires a simpler view.

W3Schools Grouped Browser Usage Over Time

In the chart above, I’ve combined all versions of Internet Explorer into one trend line and all versions of Netscape, Firefox, and the Mozilla Suite into another. As you can see, IE has declined over time as the Mozilla/Firefox family has risen in popularity. In fact, if you apply some linear regression, you see that the two lines cross in the not-too-distant future.

If these trends continue, Internet Explorer will continue to lose an average of 0.51% of the browser market per month. The vast majority of those users (all but 0.01%) will migrate to a browser in the Mozilla/Firefox/Netscape family. Then, some time in Q3 of 2009, Internet Explorer will be overtaken as the dominant browser.

Discussion

Naturally, this shouldn’t come as much of a surprise for anyone who’s kept an eye on browser usage over the past few years. Thanks to a passionate community of open source supporters, Firefox has grown into a phenomenon among the web-savvy. Just last month, it set out to get in the Guinness Book of World Records for the most downloads in a single day with the release of Firefox 3.

Of course, this data only speaks to the web-savvy audience that visits W3Schools. For the vast majority of web users, it’ll still be a long time before Firefox attains widespread usage. Much to the disdain of us Firefox fans, Internet Explorer is too common and familiar among average web users to oust from the number one spot so quickly.

Disclaimer

This was an off-the-cuff analysis of a single website’s publicly-reported browser usage statistics. Naturally, it shouldn’t be taken as authoritative evidence of anything. It is nothing more or less than one analyst’s predictions based on a limited dataset. It’s interesting, for sure, but try not to blow it out of proportion. 😉