Successful Web Analytics Approaches by Avinash Kaushik. It’s just over 1 hour long and it is worth your time. Avinash is a Google rep. He talks about how to report on data and what’s important.

Static vs Noise

If your own mother was in a crowd of 1000 people, you wouldn’t be able to pick her out because of all the other people around her. She’d be hard to see, especially with the occassional swim suit model, or clown walking about. With so many interesting looking people around, they can distract you from her. It would take time and patience to get to your mom. Eventually you’d get to her, but not before some time of hunting and searching.

That’s what most analytics programs do for you. They dump huge amounts of data on you, and you’re required to sift through them to find out what is meaningful.

But what’s meaningful? Most managers want to see a lot of data. The marketing department wants to see how many people viewed their press release, or how many clicks they got from the pay per click program. The sales department, if they’re not asking the same kinds of marketing questions, should be asking for how many leads or sales they received.

“How many hits did I get…” is one of the most popular yet time wasting questions ever invented in web 1.0 that we’re still suffering with. If the answer is 10 or 10,000, what can you do with the information? Today you got 100, yesterday you got 10. Who gives out bonuses for hits or visitors to a page? That’s where I want to work for sure.

The Internet is about making money for most companies. It’s how little companies get customers that aren’t local to them, not to mention that it’s a pretty level playing field. So if the bottom line is about making money, why do we get swarmed with questions that lead to no where? There’s change in the air, and that change involves educating corporate America to see through the forest to actionable data.

Actionable data sounds like this: “How many new leads did we get?”, “How many registrations did we get?” or “How much did we sell yesterday?” This actionable data type allows managers to make monthly decisions. With the answers to these questions, managers can plan the next month, the next week or the next year. If today you received 30 leads and yesterday we received 12, you’re increasing sales or sales potential in this case. Your actions here may be to do nothing for now or it may be that this progress rate is unacceptable. You have the power to change it either way. Knowing that you received 100 ‘hits’ onto your fancy new landing page is worthless information, even if yesterday the hit count was 50. You simply do not know how many of those hits resulted in a registration or sale.

That’s the key to using web analytics; knowing which data points directly benefit the bottom line. That’s how you see through the crowd to find your mother in the above example, you simply focus on what is important, in favor of what is not.

I submit to you that there are only a few important metrics to focus on from a managerial standpoint:

1. Are we getting more or fewer unique visitors than we have in the past? This shows growth. This is a good progress indicator.
2. Are we getting more or less orders from the internet than we have in the past? This is a big business indicator of success online.
3. Of those that place orders with us, where did they come from? This tells you where to spend your money and time in the future.

Web Metrics are not an exact science. In the big sea of data, you’ll find tons of search engine spiders coming from search engines you never heard of. Then there are hackers making requests to your server at random, viruses and computer trojan horse programs on auto pilot looking for a place to park. Plus you’ve got your competition visiting and they’ll never turn into customers either. So you need someone to manage them who understands who’s visiting your site daily. They have to filter the data looking constantly for data types to remove (new search spiders for one, or nigerian IPs. Most US companies don’t ship to Africa anyway and Nigerians are usually only there to cause trouble and they just end up skewing the data if you leave them in.) (Read my other article about Nigerian IPs. I’ve collected the IP addresses for the whole continent of Africa. you can take those IPs and have them banned from accessing the site. If you run an ecommerce site, this will help prevent you from having to ship products to Africa for free and will also keep them from leaving messages for free products or sending other spam taking up your valuable day.)

Let me define 3 web analytics definitions so you can get a better idea of how ify they can be:

1. “Hits” is an indicator of how many requests are made to the server overall. When one web page can contribute 20-40 ‘hits” per request, this makes Hits a worthless metric. People have been using the word ‘hits’ when they mean visitors, so you need to clarify with them when you’re discussing it.
2. “Visitors” is a  little tricky. This is an indicator of the number of visitors. However, this metric includes some garbage with it, because what it really measures is computers who are accessing your data. This includes search engine spiders, computer virus or trojan horses polling the internet looking for a place to inhabit, real people and there are probably more. Another flaw in this number is that if 1 real human accesses your site 8 times a day every day, it counts them all. Therefore increasing the number of “visitors” has to be taken with a grain of salt.
3. “Unique Visitors” is much like visitors but it removes most of the repetitive actions of each human visitor. It takes humans to bring credit cards to make purchases. This is a more accurate count of real paying human beings on your site, but it ain’t perfect either.

I hope you’ve learned something from this. If I can be of some help, please let me know via comments or email.