Analytical - web analytics for blogmasters

Sitemeter Relaunch

Sitemeter, which offers a nice basic website statistics tracking service, relaunched this week with a new look and new features. It’s commonly used by bloggers to give visitors a look at basic visit information (though metrics can be password-protected).

The new version of the service supposedly gives demographic information, though I can’t see how short of surveys. It’s not hard to convert visitors’ IP addresses to a country, but certainly not into gender, age, education, and household income.

Google Analytics Relaunch

Google has systematically been converting the users of each of their purchased applications over to new versions, each requiring the use of GMail email addresses. The latest rollover is for Google Analytics, which also apparently received a nice new interface.

I’ve been using Google Analytics since pretty much the first month that they took over and started taking beta signups. For a long time, though, I stopped using when the old Performancing Metrics was available. Recently, I’ve started using Analytics again, though I haven’t seen the new features and look, despite using a GMail account.

PMetrics Up and Running

Performancing, under new management from SplashPress Media, has brought back a metrics package for bloggers. Dubbed pMetrics (affil link), it’s not the same one as previously, though the code for that may yet be released under an open source agreement.

The new PMetrics is actually a rebrand of GetClicky and, believe it or not, has a lot of great metrics useful to bloggers than Google Analytics does not. I find the per-visitor granularity very helpful, and a lot of the features that the old Performancing Metrics had is slowly being integrated in. What it doesn’t have is a graphical breakdown of a particular day’s visits, which is what I really like about Sitemeter. (Google has it, but it’s more complicated to access because of the giant menu.)

So, yeah, I use all three of these packages, depending on what metrics I want to see and how quickly.

Avinash Kaushik of Occam’s Razor says that path analysis of website visitors is not a good use of time. I know that in the past, I felt it was an extremely important indicator of a website visitor’s experience. That is, understanding which sequence of webpages each visitor visited.

Now, however, I feel that it’s only useful for particular types of websites: those with well-defined navigation paths on your site, and can “guide” visitors to where you want them to go.

For example, say you have several tutorials, each consisting of a sequence pages, and containing a bare minimum of navigation. Because of this, you are “channeling” your visitors to complete the tutorial. It’s worth it to know whether visitors actually are completing the tutorials. If they are not, there’s probably something wrong.

Another example is a checkout path, if you have an e-commerce website. Are visitors completing checkout? This is in fact the exception scenario for path analysis that Avinash discusses. He says that web pages with a structured experience are worth doing path analysis for. For anything else, there are far too many path page combinations - possibly thousands. And for some sites, it’s just not worth it. (Just my opinion, but I think that weblog style websites fall into this “not worth it” category.)

If you’ve been using Google Analytics to track and analyze your web metrics, you’ll probably already know that they’ve doubled the number of profiles you can track to 10. (If you didn’t know, you do know.) I’ve been waiting for this for a while. [via Google Analytics Blog]

Excellent advice from Avinash Kaushik: 7 best practices when measuring your conversion rate. [via Occam’s Razor]

Peter Da Vanzo says forget blackhat/ whitehat SEO: there are too many factors to say that such and such a step will help your hinder your rankings. He says the game is harder than it used to be. Nick Wilson of Performancing says SEO is not rocket science. By the way, Graywolf says he’s a grayhat SEO (link is to a pirates book review - so that might explain what he means).

Stuntdubl says he became a better SEO by learning to search better, and understanding how others search. Most people, while comfortable with search engines, generally do not use the more sophisticated commands. When I was a search engine webmaster in 1996, most people didn’t use double quotes around partial phrases, and I’m guessing they still don’t. Use that to your advantage.

While you’re at Stuntdubl’s site, check out this bit of wisdom: don’t market to everyone - pick your niche. And talk directly TO your readers, not AT them.

Blogger Skills has an overview of Google’s Matt Cutts and his SEO video podcasts. [via Blogger Skills] Blogger Skills are also looking for contributors. Check out their other great tips about blogging.

Web metrics and analytics can be a fairly complex topic, but for the purposes of analyzing the performance of your weblog (or website), you only need to understand a few key metrics, how to analyze them, then how to improve your website. At least as a starting point.

But which metrics are important? Someone asked me to write an e-report early this year and list some key website/ weblog metrics. It was not an easy task. Who was I to decide which metrics were important and which were not?

Why so difficult? Because which metrics are important to you may not be important to me, and vice versa. Similarly, there is a difference in what is key to a webmaster and what is key to a blogmaster.

For example, I run on budget hosting. I have a certain allowable amount of bandwidth allowed on some of my sites. If I go over, it’s going to cost me. That means I probably should be keeping an eye on my monthly bandwidth metrics, including plotting a daily chart and watching the trends. (Unfortunately, budget hosts do not give you access to your web server logs, which I learned the hard way.) You, on the other hand, may be responsible for a corporate site and thus have a professional hosting plan. Bandwidth may not be an issue.

Another example: I as a blogger may be more concerned with the number of comments I’m getting, or the number of times my free e-book has been downloaded. You might be concerned with the number of repeat visitors you are getting, because your metrics data shows that that generates leads and, eventually, sales.

Still, there are some metrics which most webmasters and blogmasters would consider to be “key”. Within a certain set of values, at least some of these will be important to pretty much everyone. Assume that these sites are generating revenue, either through advertising, affiliate programs, direct online sales, or sales leads. Let’s have a look at what those might be, in alphabetical order. Keep in mind that these are general metrics. To make the discussion simpler, I have typically have not mentioned whethere we’re talking daily, weekly, monthly, seasonal, or yearly metrics. Usually we’re talking daily metrics.

Ad clicks

This is only valuable to those sites that run pay-per-click advertising. That is, if the advertiser pays you for each legitimate click on one of their ads running on your website.

Ad impressions

If you are, or plan to be, running CPM ads (paid prorated for every 1000 impressions), then this is a metric you want to collect. In fact, you may have to report it to an advertiser or ad network. This metric can be misleading. If I have two ads from the same ad network on every page of my site, that typically means that the number of ad impressions is double the number of pageviews. For pay-per-click advertising, this metric is not that important.

Bandwidth

If your host allows you X gigabytes of bandwidth each month for your website, and you have reason to believe you might exceed that limit, then you should track this metric, and calculate peak periods, not just daily averages. For example, the video-serving site youtube.com generates humungous amounts of monthly bandwidth but no income. Which makes people wonder how they’ll pay for it all.

Comments

This is a weblog-specific metric. Most “regular” websites do not have commenting capabilities. This metric varies in importance level from blogger to blogger, and should also include trackbacks. Trackbacks are comments that are auto-generated and are actually a snippet of text from a web page on another site that links to one of your pages. For example, if you post an article that links to this very page, a trackback comment will eventually show up below (provided it’s not spam). When you view that trackback, you’ll recognize that its text is an excerpt from your page

CPC

Cost per click. For an advertiser, this metric refers to how much they are spending, on average for every click of one of their ads, regardless of the website. For a webmaster or blogmaster, it means how much you are earning, on average, for every click of a PPC (pay-per-click) ad running on your website.

CPM

Cost per M, M=1000 ad impressions. Again, has slightly different meanings depending on whether you are an advertiser or site owner. For site owners, this metric is a prorated value indicating what you are earning, on average, for every thousand ad impressions or pageviews. So if you had 300 page views today and earned $6 in ad revenue (either PPC or CPM), then your CPM is ($6/300) x 1000 =~ $20.

Of course, CPM can be a highly misleading value. Just because you earned $6 for 300 page views (or ad impressions, or a combo) does not mean you’ll earn anything for the remaining 700 pageviews. So if you are tracking the CPM of one of your sites throughout the day, the value is probably dropping as the day progresses.

CTR

Click-through rate. Measures the average number of ad clicks for every 100 pageviews or impressions - depending on whether you are running PPC or CPM ads. You should measure these separately, to track the effectiveness of different types of ads.

# Leads

This could be offline data, but usually it’ll come from a “fill out your details” type of web form. This could also be turned into a lead percentage, by dividing the number of leads generated by the number of pageviews or visitors. This value may or may not be key to determining ROI (Return On Investment), discussed below.

Pageviews

This is not the same as hits. Hits are an inflated value that includes all types of files, including images, that are served up when someone views your webpages. A pageview metric is more honest and typically smaller, unless you have no graphics whatsoever. For example, if you have a website with a single page and it displays four images, then each pageview generates 1+4 = 5 hits. Thus, hits are meaningless, except maybe to calculate bandwidth.

Referrals

Are other sites linking to you? Fine. Now is anyone visiting your website from one or more of those sites? How many people per site, and in total? This is the number of referrals.

Revenue

The revenue metric can be measured in numerous ways, depending on how your site is set up. Maybe you’re getting paid for CPC or CPM ads, earning affiliate revenue for CPA (Cost Per Action) activities (say leads, subscriptions, purchases). Or maye you are selling something online, or even offline. This is one of the few metrics that while it could be estimated daily, is probably calculated monthly, once all revenue information is in.

ROI

ROI means Return On Investment. It can be calculated in many ways. Let’s say that you are a webmaster responsible for a corporate site (it’s not as much of an issue for bloggers). You know that total monthly revenue for your employer, either directly due to ads, or due to sales from leads and/or online purchases totals $10,000/m. Say that your salary, the cost of equipment, rent and hosting totals about $7,500/m. The return on investment is (10000 - 7500)/10000 = .25 = 25% ROI.

Visitors

The number of visitors you have might not be as important as the number of repeat visitors. Does the revenue on your site depend on people coming back? Are they coming back? Is a graph of the number of returning visitors increasing, on average? Related to this is the number of times they come back during a certain time period, as well as how many pages they view in each visit (aka session).

Obviously, this is just a selection of all possible website metrics. It’s only my opinion, but I feel that the above are some of the key metrics that webmasters and blogmasters will/ should be concerned with. I’ll discuss these metrics individually in the future.

Blogging is about more than just writing and posting your articles to your weblog. If you have any desire to actually earn a living from your online publishing pursuits - be it through contextual advertising, CPM (Cost per M, M=1000) advertising, or affiliate programs, you have to measure and improve your weblog/ website’s performance.

Web metrics refers to the measure of various performance values, be it the number of visitors per day, pageviews per week, revenue per month, ad clicks, comments, bandwidth, CPM, CPC, CTR, etc. As you can see, when you combine measure with duration (day, week, month, season, year), there’s an alphabet’s soup of metrics related to websites. (I’ll discuss some key metrics in an upcoming post.)

Web analytics is the study of website metrics. It’s about watching the trends of metrics day by day or week by week, charting the daily changes, determining average performance (Moving Averages, Multiple Moving Averages), devising trendline analyses (using advanced statistical methods), trying to determine why the metrics are the way they are, and what you might do about it. This ties in directly with the next step in the process, optimization.

Web optimization is the act of devising and implementing a plan to improve your website’s metrics. You’ve measured and analyzed the metrics, and determined the possible (most likely) reason for poor metrics. Maybe you’re not posting enough. Your topic might not have enough interest, or there’s too much competition, or it’s seasonal (e.g., skiing).

Now you need to come up with a plan for improving those metrics, and going ahead and implementing the plan. Optimization is not as exact a science as web analytics, so its a topic of volatile discussion, especially SEO (Search Engine Optimization) and SEM (Search Engine Marketing) processes. Everyone has their own idea of why something happened and what to do about. It’s a matter of how they analyze and interpret the metrics collected.

Methods of optimization include SEO, SEM, visual redesign (page layout), information design, increased content and posting frequency, article marketing (publishing to specific article directory sites), contest promotions on the site, etc.

Regardless of your method, web analytics and optimization is an ongoing process. Each time you to optimize, you have to go back and re-measure and re-analyze your metrics. It’s a cyclic process.

I’ll be talking about web metrics and analytics on this site. Someone has asked me to write about optimization elsewhere. Although I have a lot to learn and that project has not been unveiled yet. I’ll announce when it is revealed.

The free MMA spreadsheet (MMA = Multiple Moving Averages) from my last post can be used in a number of ways. It’s currently set up with MA (Moving Average) windows of 7d, 14, 28d and multiples of 28 days. When analyzing the stock market, there’s a tendency to use windows of 100 d and 200 d when forecasting a stock market index or the price of a stock. Why 100 or 200 days? Well keep reading.

The way the spreadsheet is set up right now, the graph gives you too much information that’s hard to make heads or tails of. There are too many Moving Averages displayed, and the resulting graph is a muddle. Reducing the number of windows to 3 windows clarifies things for medium- and long-term patterns.

In truth, when you’re looking at trendlines, you should focus on just a few. What’s important to you? For me, short-term trends are a curiousity. I plot MAs of 7d, 14d, and 28d just to look at them. But it’s the long-term trends that are much more important. For example, if after 1 year, a website/ weblog is showing steadily rising 100d and 200d MA curves, then I’m very happy. And at the same time, I can see how each day’s traffic or revenue is doing against 100d and 200d curves.

I’ve tweaked my own copy of the spreadsheet to plot daily pageviews against 100 d and 200 d MA windows. It’s not that hard for you to do, if you know how to use Excel or OpenOffice, etc. But if you want me to post a new version, just ask.

Pick your own windows, depending on the durations that are important to.

Several months ago, I made a promise to put together a spreadsheet that bloggers could use to calculate MMAs (Multiple Moving Averages). It’s been a while but I finally have a free one for anyone interested.

I use my MMA charts to get an idea of what’s been happening, as well as what type of pageviews or ad revenue that I MIGHT experience in the near future. In other words, MMAs are trendlines, but they’ve been “averaged” out to smooth out spikes.

MMAs give you insight into patterns that your website/ weblog metrics (be it pageviews, ad clicks, revenue, CTR), etc., are experiencing, and what might be coming up. It’s hard to pick these patterns off by looking at a list of numbers. You need a graph for a better view. And because there are several “rolling averages” being calculated, this isn’t something you want to calculate by hand. Which is why I use a spreadsheet. (I could write a program, but a spreadsheet is simple and pretty much any blogger can use it.)

I use MMAs for a lot of uses. As I’ve mentioned in my (incomplete) series of articles on Moving Averages and Multiple Moving Averages, I’ve used MMAs to correctly predict stock market turns (along with other data).

At the very least, they give you a “reverse crystal ball” view of what’s been happening with your weblogs and websites. They also show you that you shouldn’t worry too much about what happens on any given day, because there are always “transients” in the metrics. Worrying about day-to-day changes might give you ulcers. It’s the medium- and long-term patterns that are more important. (Not to say they won’t make you ill too.)

So here is my MMA Multiple Moving Average spreadsheet, using the publicly-shared, web-based ZohoSheet spreadsheet format. You should NOT need an account at zohosheet.com, but you may want one. (They also have the web-based zohowriter, which I’ve been using for two days. It’s compatible with MS-Word.)

The ZohoSheet is compatible with both MS-Excel and OpenOffice, and you can export to HTML or PDF. Go to the link above and view the spreadsheet, then find the export link in the displayed page and choose your spreadsheet flavour.

You’ll see that I’ve set up the sheet from Jan 01, 06 to Dec 31, 06. (The date format is a quirk of ZohoSheet. Feel free to change, as well as the duration. Replace the data in the column that says Pgvws with your own values. They can be pageviews, ad clicks, ad revenue or hey, even stock share prices. Whatever you like.

I’ve used MMA “windows” of 7 days, 14d, 28d, and then multiples of 28 days. I’ve only shown the MMA graph up to 56 days (approx. 2 months). [The data in the sample is some of mine from last year, transposed to this year. You can see the severe short-fluctuations that get “normalized” or evened out.]

If you want to change the sliding “window” durations, feel free. But make sure you look at the cell formulas first to understand what I’m doing before you try to extrapolate. Same goes for shortening or lengthening the data collection duration.

Good luck. Feel free to ask questions here, but being currently busy, it may take me a while to respond (at least during weekdays).

Don’t let the title of this post scare you. The general math formulas that you need are relatively simple. It’s understanding the analysis that takes some effort. This first in a series of posts is directed at those of you bloggers and website publishers that are obsessively looking at your site traffic and related stats and lamenting the poor results. Let’s start with a little story that may help you feel a bit better about your fluctuating traffic and revenue, as well as steer you back to focusing on the important aspects of online publishing.

Leonardo Pisano, a member of the Bonacci family in Pisa, now part of Italy, was the son of a diplomat and received an education in mathematics and other topics in North Africa. He was later referred to as Leonardo Fibonacci, the mathematician, and was credited for a very simple sequence of integers known as the Fibonacci Sequence - which you may have learned in high school. Truth is, written evidence shows that this sequence pre-dated Fibonacci. It’s a sequence that is lurking in the proportions of architecture, nature, even the stock market.

If dozens of websites are to be believed, numerous stock traders have used Fibonacci Trading to determine stock prices and have made a fortune buying and selling short, day trading, etc. Mathematically, the idea is sound. The question is, does the Fibonacci sequence truly manifest itself in pretty much everything in nature. Experience suggests that it does, but it’s a matter finding the precise sequence and determining how to apply it. That’s the part that takes effort.

We’re not going to get too deep into Fibonacci sequences yet, and this is a long-term series. However, a refresher of the best-known Fibonacci sequence is in order. The idea is simple. Start with the values f0 = 0 and f1 = 1 for the first two whole numbers in the sequence. Then f2 = f0 + f1 = 0 + 1 = 1. Also, f3 = f1 + f2 = 1 + 1 = 2.

The general formula is fn = F(n) = F(n-1) + F(n-2), if n > 1. If n=0, F(0) = 0. If n=1, F(1) = 1. So then f4 = F(4) = F(4-1) + F(4-2) = F(3) + F(2) = f3 + f2 = 2 + 1 = 3. The sequence so far is 0, 1, 1, 2, 3. What is the next number in the sequence? It’s the sum of the last two numbers, 2 and 3, or 5. Proof: F(5) = F(5-1) + F(5-2) = F4 + F3 = f4 + f3 = 3 + 2 = 5. The next, then, will be 3 + 5 = 8.

In ancient architecture and in examples of beautiful proportions, there is something called the Golden Ratio, which is related to the primary Fibonacci sequence (i.e., the one above). Approximations to the Golden Ratio are calculated by taking the larger ratio between any two consecutive numbers in the Fibonacci Sequence. For example, 3/2 = 1.5; 5/3 =~ 1.66666; 8/5 = 1.6; 13/8 =~ 1.625; etc. The actual Golden Ratio, phi, is approximately 1.618.

On the flip side, the smaller ratio between two consecutive numbers are also valuable. They are used in stock marketing and other analyses. For example, 1/2 = .5; 2/3 =~ 0.666666; 3/5 = 0.6, 5/8 = .625, 8/13 =~ 0.615384; etc. Now here’s where things get interesting. It is not these ratios that are used in analysis but something a little more complex: 1 - (1/phi) = 1 - (1/1.618) =~ 1 - 0.618 = 0.382.

This latter value is one of a set of magic ratios that I’ll talk about in later posts. I’m boldly hypothesizing that you can use 0.382 to calculate future highs and lows for your website traffic, and possibly even your daily revenues, provided you have enough data. What is the value in such an exercise? Simply that you can gain a true understanding that website traffic and revenue WILL fluctuate daily in a natural state, and when you might want to take actions (i.e., write and post articles) to positively influence this in the most efficient manner.

Technorati Tags: mathblog, math blog, fibonacci sequences, website forecasting, website analysis, golden ratio

If you are running advertising on your website/ weblog, or plan to, one of the more interesting web metrics is CTR, or Click-Through Rate. It’s important from a financial point of view and refers to the percentage of clicks on an ad (or any link in fact) compared to the number of impressions said ad (or link) received. [NOTE: I should have posted this before my How Long Before I Make X Dollars Per Day? article. I’m posting this for posterity only.]

For example, if your home page receives 1000 pageviews for a given day, and the same 5 ads appear on that page all day, all the time, then each ad receives 1000 impressions for that day. To calculate CTR for each ad, you divide the number of clicks for that ad by the number of impressions. So if ad 1 received 20 clicks, then the CTR is (20/1000)x100% = 2.00%.

If you are calculating CTR for an entire channel, there a number of ways that this is calculated, but not all of them are accurate. A channel, by the way, can refer to a number of things, at least for Google AdSense. The simplest definition is that a channel refers to a domain, subdomain, or URL. For Google AdSense, you can also sub-divide pages by ad type or colour, etc., to define more granular custom channels.

Let’s talk about a domain channel. Track your channels throughout the day, maybe every hour or two hours. Just for one day. When you get the first click (or more) on a channel, record the indicated CTR, then watch the channel closely. As the day progresses, the CTR for that channel will change. Whether it goes up or down depends, of course, on further traffic and clicks. In my experience, CTR for a channel drops by the end of the day.

Overall, the day’s CTR for a given channel isn’t that important per se. What’s more important is a graph of the CTR for each of your channels over a range of time, say at least 6 months. You can also calculate the overall CTR for all your channels, and then graph it over time as well. The latter CTR, on a daily basis, is equal to the total ad clicks for all channels, divided by the total pageviews for all channels, times 100%.

If you are very serious about tracking your CTR, you will want to create an MA (Moving Average) or MMA (Multiple Moving Averages) graph for daily CTR values over a long period of time.

What I’ve noticed for my blogs is that my overall network CTR has been dropping gradually since its peak in Nov 2005. This is probably due to having too many blogs with just a few posts, which are not generating many clicks. It’s been rumoured that this lack of performance can result in lower AdSense earnings per click for your other weblogs. Consider looking at the CTR or CTR MMA graphs for each of your weblogs individually. For instance, I know that some of my blogs are actually doing better over time. I’ve considered removing advertising from those that are not performing.

Technorati Tags: mathblog, math blog, CTR, ad revenue, google adsense, MA, MMA, multiple moving average, adsense performance

Numerous bloggers have reported that, on occasion, they receive a spike in traffic from one of a number of sources: a popular blogger or a social bookmarking site. Either way, the result is sometimes a massive influx of visitors, typically all at once, or over several hours for a few days. The problem is that most of us bloggers/ online writers have purchased low-cost website hosting plans that allow only so much web traffic per month. Exceeding that costs more money. Exceeding the limit in just a few days can get your website shut down temporarily by your host. So what to do?

The answer is to use a bit of forecasting, as discussed in the last several posts, and to approximate the amount of bandwidth currently being used daily or monthly. Then you can approximate when you might exceed your current hosting plan under normal traffic growth. (Please first read my previous “forecasting” posts, if you have not already done so.)

For example, let’s say that you have 6 months of traffic data for your weblog. (If you have multiple sites, analyze one site at a time.) Your data shows that average daily pageviews for the past month is 250. Assume that each post contains 10Kb (Kilobytes) of text and a 40Kb image that links to a larger version at 110Kb. (I’m making up numbers.)

The total data bandwidth your site uses for the most recent month, then, is about 30 d/m x 250 pv/d x (10 Kb + 40 Kb + 110 Kb)/pv = 7500 pv/m x 160 Kb/pv = 1,200,000 Kb/m = 1200 Mb/m = 1.2 Gb/m (Gigabytes/month). Now that is pretty puny bandwidth, considering that most “economy” hosting plans allow 50 Gb/m or more. I currently use two different domain hosts. One of them offers 250 Gb/m for just US$3.95/m (less if I pay 12 months in advance). If I need 500 Gb/m, I pay only US$5.95/m. Or for US$13.95/m I can get 1 Tb/m (1 Terabyte = 1,000 Gb).

So under this sort of scenario, you probably don’t have anything to worry about. Notice that I am not factoring the amount of bandwidth you use for transferring files from your computer to your web server. That cost is paid to your ISP (Internet Service Provider). So how many pages would you need daily, on average, to exceed your monthly limit? Well, given the example data above, we would need to exceed 250 Gb/m. Let’s break this down.

Given data:
Monthly bandwidth limit = MBL = 250 Gb/m
Average pageview bandwith = APB = 160 Kb/pv = 0.160 Mb/pv
Days per month: 30 d/m

Calculate:
Daily bandwidth limit = DBL = MBL/days = (250 Gb/m) / (30 d/m) =~ 8.333 Gb/d = 8,333 Mb/d
Daily pageview limit = DPL = DBL/APB = 8,333 Mb/d / 0.160 Mb/pv =~ 52,083 pageviews/ day

On the other hand, if you started running podcasts with audio or video files on a regular basis, your APB (avg pageview bandwidth) will increase dramatically, depending on the file formats that you are using. For example, .WAV files in CD quality stereo sound take up about 10 Mb per minute of audio recording. MP3 files, on the other hand, are a typically a lower quality and thus take up less space - sometimes 1/3 of the WAV equivalent. For video, .WMV (used by Microsoft’s Windows Media Player) takes up as little as 1/10 of the space of the high quality .AVI and .MOV formats.

So if you want/ need to publish high-quality multimedia files on your website, and you expect your site to eventually gain popularity because of it, you’ll want to regularly check your bandwidth usage. (Or you could opt for a free or commercial storage solution.)

For example, say that you expect that 1 of every 100 pageviews will be to one of your video posts. A short, high-quality video file might be 30 Mb. If you’re getting 250 pv/d, then you’ll get 2-3 requests per day for your video file. That’s between 60-90 Mb/d, which still doesn’t exceed the DBL of 8,333 Mb/d that we calculated earlier.

But if because of your high-quality video, you get linked from some popular site. The link gives you an extra 1000 visitors in a single day, all of whom view the video. That’s at least 30,000 Mb bandwidth usage for that day, over and above your typical average. You’ve exceeded your average daily limit, but are still under your monthly limit. Unless this goes on for several days, which sometimes happens. At this rate, you can only handle about 8-9 days of this bandwidth before you use up your month’s limit. And you’ve still got the rest of month to cover.

What if you had posted several videos, and the incoming visitors know this. As a result, each visitor might view more than one video on spike day. You now might use up your month’s limit in just a day or two. What’s more, because of the sudden influx of traffic, your server will slow down and some visitors will watch parts of a video, then return at a different time, hoping to watch the whole thing.

You can see how your monthly bandwidth could be used up very quickly with just one incoming link from a popular site, especially if you are a featured link. Most domain hosts will shut you down until you pay for an upgrade plan. Some might force this on you even if you do not expect to get the same kind of traffic again for a long time.

Given the low hosting plan rates available these days, it’s not a lot of money to bump up your hosting plan. A few years ago, the same plans cost $60/m and up. Today, a very busy website might have hosting plan that costs US$139/m.

If you suspect the possibility that your website might get a huge influx of traffic someday, there are ways to protect against that. Well, if you are running a weblog using the WordPress blogging platform, you can use Elliot Back’s Digg Defender plugin (found via Stupid WordPress Tricks).

Technorati Tags: mathblog, math blog, website bandwidth