A Quick Guide to Reading Webstats

On Monday and Tuesday of this week, I was in Penticton visiting a client, Theytus Books. Theytus is Canada’s largest indigineous publisher and I’m part of a team doing work for them to analyze some of their business practices and make technology and web recommendations. Visiting Theytus’ office and business setup was a great experience and helped me contextualize the work I’m doing for them. The visit improved my understanding of Theytus’ technology and web practices, and it allowed me to get a better handle on how our recommendations will end up living in their world.

On the second day of our visit, I sat with Anita Large, Theytus’ Managing Publisher, and asked her to show me Theytus’ hosting account. We went through some of the details and I made notes. Then we opened up Theytus’ webstats. Anita knew about her webstats and knew they meant something about the performance and health of her website, but she didn’t have a great understanding of what each metric meant and so she had stopped looking at them.

I spent some time going through the Theytus webstats with Anita, and it was suggested to me by a colleague, Kiley, who also listened in on my webstats chat, that Anita was not alone in wanting a quick guide to webstats. I walked Anita through the stats in about half an hour and she loved being able to understand the words and figures. So, in the spirit of sharing, and because I think Kiley was right, here’s a quick guide to reading webstats.

Webstats Software

First, a quick note on the webstats packages you’re likely looking at if you use a basic web hosting package, such as the Mini-Me package I use for Work Industries from Nexcess.net, a service I recommend.

The webstats software you’re using is free and offered en masse to everyone. Common webstats packages used include AWStats and Webalizer, both of which are not great but will do the job for most people starting out or looking for an overview of their website’s performance.

Think of it this way: the webstats packages that come with basic hosting are like the standard gauges on a car - speedometer, oil temperature, battery amps, fuel volume. They tell you some basics and you’re able to start off and get around with them. A more advanced car comes with some more advanced gauges - tachometer, turbo-boost, oil pressure, fuel consumption, tire pressure. An expensive car built for a specific purpose - an F1 race car, a rally car, an endurance car - requires more specialized tools that provide greater volume and more granular detail of information. The more advanced tools also require skills and experience to interpret the information, relate it to the performance of the car, and turn it into a meaningful action plan to change the performance. But I’m getting ahead of myself. Here, let’s stick with the basics.

Just as with a car, some of the measures in your webstats are useful for managing and enhancing your website. Some of the measures are just added-on information to sell the car. They’ve been added because they make sense to the folks who built the webstats packages or simply because they could be added.

Knowing how to interpret the measures - to transform them into useful information - is far more important than the volume or detail of the measures. The more you invest in your car or website and the more you demand of its performance, the more you need to know, and the more you need to create a strong feedback loop that monitors performance. The reporting should be commensurate with both the rewards of analysis and the analyst’s level of expertise. A daily commute doesn’t require an F1 race car.

So which measures matter and which measures are just added-on information?

Finding the Meaning in the Glut

For most websites, here are the numbers that matter from the common webstats packages:

Meaningful Webstats

1. Unique Visitors: This is the single best measure of how many people are coming to your website. Under detailed scrutiny it’s flawed, but as a basic measure, it works very well. Keep your eye on it. Your Unique Visitors number should be growing month after month and year over year. See if you can find monthly patterns in this number, or seasonal patterns that correlate to something in your business. (This stat can also be called Uniques.)
2. Visits: Visits displays the total number of times people or spiders have come to your site. People are humans visiting with a web browser, while spiders (or bots) are computer programs used by websites like search engines to crawl your site and retrieve its content so it can be added to the search engine index. (Here is a good list of the search engine bots’ names.) The Visits stat is the second-most important thing to watch. (Visits can also be called Number of Visitors or Visitors.)
3. Page Views: The Page Views number shows you the total number of pages requested and served by your web server. If a user comes to your homepage (1), then clicks on a link to your About page (2), then clicks on Contact Us (3), then closes their browser window, they will show up in your stats as 3 page views, 1 unique visitor, and 1 visitor. (Page Views can also be called Pages.)
4. Referrers: These are the other websites and sources of links (called direct referrals, meaning bookmarks, email links, direct URLs typed into the browser) that are directly sending you traffic. Keep an eye on this stat because it will tell you where your users are coming from and who is linking to you.
5. Search Keywords: These are the searches people are using to find your website. When they click through to your website, the search engine tells your web server what search they used, and your webstats software reports on this.
6. HTTP Status Codes: These will tell you what errors your web server has logged. The most common and most important to stay current with are 404 errors - “file not found.“ These occur when someone tries a URL and doesn’t get anything back from the server. They often indicate a misspelled link or misreferenced file (internal) or a mistake from an external site pointing at your site. Either way, you want to track down each of these errors and see what’s causing it. Each one is a potential connection with a prospective client.

Kinda Meaningful Webstats

1. Days of Month: It’s interesting to watch which days attract more traffic, and to watch which new content or events on the website can drive user behaviours, but it’s very hard to correlate these stats back to events.
2. Days of Week: See above. But, the real value in looking at Days of Week is seeing which days people visit your site. Generally, the middle of the week will be busiest, with Tuesday, Wednesday, and Thursday probably your most popular days.
3. Hours: If you blindfolded yourself and had to guess when people were visiting your site, you could probably draw a graph pretty close to the way this stat appears. A business-related website will draw people during working hours and a website that caters to personal interests will draw people in the evening.

Mostly Meaningless Webstats

1. Hits: Hits represent the most overused and abused term in website performance. Hits mean nothing. I’ve heard some people say that HITS are How Idiots Track Success. Simply, a hit is one file requested and served from your server. That file could be any type. A homepage consisting of 37 files (not unusual) will register 37 hits every time a person or bot requests the homepage. Hits mean next to nothing.
2. Bandwidth: This is the amount of data transferred from your web server to requestors. Unless you’re metered by bandwidth, it’s not too important. You’ll likely find that your homepage or most popular landing page is the top bandwidth user, unless you have high-bandwidth files on your server, such as audio, high-resolution images, or video files. Don’t pay much attention to bandwidth unless you’re close to your hosting limit.
3. Hosts: These are the ISPs used to connect to your site. Unless you can differentiate ISPs from URLs, skip spending any time on this stat.
4. Visitor Domains: Depending on your webstats, these have varying degrees of accuracy. Basically, every request to your web server comes from a requestor using an IP address. Visitor Domains groups these requestors by IP address, which are assigned in blocks to various organizations and countries. But the mapping of IP addresses to countries is very unreliable, so Visitor Domains ends up being unreliable.
5. Operating Systems: Microsoft Windows is the most popular operating system in the world. Do you need your stats package to tell you that? What are you going to do with that information? Nothing, unless you’re offering software for download.
6. Browsers: Microsoft’s Internet Explorer is the most popular web browser in the world, but its market share is eroding. Mozilla Firefox is second and growing. Apple’s Safari reach is small. So are Opera’s and a few others. The only important thing I have ever learned by looking at Browsers is that a mobile browser was accessing one of my websites. That .2% of my traffic was interesting to me. But I wasn’t going to do anything about it. Build your site according to the W3C web standards, employ a good web designer to ensure your site works with the quirks of the major browsers, and every few months glance at the overall market share of browsers.
7. Any other stats overlooked here.

Setting Up and Securing Your Webstats

In your hosting account, you likely have access to your stats page. It’ll be called webstats or the name of one of the software packages listed above. A few questions to ask are:

• Is your stats page secure or can anyone with the URL for it see it? This is important. Copy the URL of your stats page, log out, then paste the URL into the address bar of your web browser. Can you see the stats page? If so, it’s not secure and anyone can see it. You want it to be secure. Get in touch with your hosting company and tell them you want it secured. If they balk at this or try to charge you money for it, change hosting companies.
• Is your own activity tracked in your stats package? You probably visit your own website. Is your webstats package adding you or your company’s activity to its numbers? Ask your hosting provider. Or, look for a page that only gets accessed when you’re logged in. If one of these pages shows up in your stats as getting activity, then you are skewing your own stats. This could mean very little, if you have strong traffic volume, or it could mean a lot. Contact your hosting provider and ask them how to remove your own activity from your stats. There are ways to do this, such as not including the traffic of a specific IP address, so be persistent and get them working for you.
• Do you get your webstats on a regular basis? Are they run automatically? You should be able to answer “yes” to both of these questions. You also want to get your webstats on as regular a basis as possible. Weekly is good, daily is better.
• Never let your ISP or host throw away your webstats. Historical data are important and data storage is cheap. If they insist on removing your historical webstats, tell them you still want the data and ask them to export them to a file for you to keep.

Trends to Watch: How Your Webstats Relate to Your Efforts

Here is a first-draft list of things to watch for in your stats:

• Month-over-month trends for meaningful metrics (or week-over-week if you get weekly stats).
• Spike activity in unique visitors, visits, and bandwidth. The good scenario is when this correlates with a link to your site from an external site with lots of traffic. You’ve been found, deemed worthy of attention, and linked to. Watch your referrers to see where the traffic is coming from. The less-good scenario is when a spike correlates to poached content (other sites using your images to serve on their pages) or some kind of scraping of content (other sites literally copying and pasting your content onto their own site).
• Search keyword phrases: These will tell you what terms people are using to find your site in search engines. Some search keyword phrases will come as no surprise if you’ve built your site correctly, but others will be completely unexpected. This is both the pleasure and frustration of working on the web - getting found for what you offer and getting found completely randomly for something you never considered being found for. Search keyword phrases are leading indicators of areas of opportunity for your site.
• Referrers: Since referrers are the websites that are sending you traffic, of course you’re going to want to pay attention to them. Who are they? What made them link to you? Can you find how they linked to you? Go to their website and get in touch with them. The web is made of people. Web people love to get a “thank you for linking to me” email.
• Most popular entry pages: These are the pages people have bookmarked or linked to. Why? Are you an authority or resource for something you never intended to be? Sleuth around a little and find out why people are coming to that page of your site.
• Overall number of visitors: Unless you have large enough numbers - like in the hundreds or thousands - for the Meaningful Webstats listed above, don’t sweat changes in your overall numbers too much. Watch them, but don’t obsess. Small numbers are often statistically unreliable. Ignore the percentages of change until you have large enough numbers for them to mean something.

So that’s it. What do you think? Many of you out there have great knowledge of webstats packages, far more than I have. Feel free to take this quick guide and use it to educate others - just give me some credit for starting the snowball rolling!

If you have anything to add or ask, please do so. Together we’re much stronger than any of us alone or on islands. The web is about links and connecting information. Add a comment, send me an email or post on your own site and link to this post to continue the conversation.