Understanding Trends in Google Analytics

I was talking in understanding segmentation post, we can use that segmentation functionality to evaluate performance. Say for example which traffic sources performing better, this is very important to know often all.

If you have one source converting at 94% and one at just 28% obviously one is more valuable to you than another or is it. Just like averages and aggregates lie we need to make sure we don’t consider metrics and isolation. We consider full context of the number we’re seeing to truly understand what’s going on. Now clearly in this case if I’ve have to choose one or the other I will take 28% any day over 94%. 28% from 80,000 is a lot and 94% of 17 is very little.

But context is about much more than just Web metrics, we tend to get so focused on that that we forget other aspects of our business that is interrelated. For example we have a lot of travel related clients some are in Hawaii and I can tell you that their website will convert a lot differently on a day that is sunny than on the day when it’s raining. So we wouldn’t want to include the changes we made to the site that they were a disaster and we should revert them right away without realizing those numbers into context of this external weather event.

And more than merely understanding why our performances that way, if we savvy about our analytics we can actually use that to our advantage. For example one hurricane Nina was blowing over one of our clients resorts with style huge spike in traffic, now digging into our analytics we found all due to hurricane Nina initially there was some concern that this was effectively bad press that would do damage to the brand. But some creativity allowed us to react and take advantage. For example in Hawaii resorts its common for hurricane guarantee. In this case the hurricane offers the chance to come back during better weather with all kinds of upgrades and freebies.

So by changing the homepage from the picture that is sunny paradise we should put a huge flash page all about hurricane Nina then they would be able to salvage that traffic, generate new stories in the press and get many folks associate in their mind the guarantee that their brand not just the scenes of horrific hurricanes all because they knew how to use analytics to their advantage.

Taken metrics into context is important in lots of ways. The CEO of the automobile company is ecstatic showing up this chart going down and showing how much money they were losing everyday. So what was he happy? Because he wasn’t losing as much as they expected to lose and not as much as their competitors were losing.

Keeping the date in relative context is always important. And you can imagine how marketing manager for this big hotel next to Ifle tower, bidding on keywords and ranking in the search engines just got whole lot tougher for a Hilton in Paris a few years back when searches on Paris Hilton suddenly got way more popular. And the point here is it has nothing to do with the website or the campaigns and analytics or even their business. In fact on the online keywords searches and we hear this all the time oh they are not our competitors, they just have the same name or we just share the same keywords umm well that you are the competition online. We called this accidental completion and it’s important to realize your competitors off-line have little to do with your competitors online.

In this graph above is two aspects of analysis that are critical to making correct decisions about our site. You can guess what industry this is well its actually travel as well not in my house booking travel is usually involves my wife and i doing some research if we get a chance but we actually not just booking until we both get a minute to sit down confirm and look at the calendars etc.

And this always seems to work out on Sunday night when there is no big plans and we actually have a few minutes. But just because it’s how it works in my house I can’t make the mistake think in my personal bias is the same as all my clients.

In fact it looks like based on this conversion graph but that’s not the case at all. Above graph tells me most people are role in to work on Monday morning and say I can’t handle this I need a vacation so they jump online.

So the key here is that the size of our personal bias cloud the data. We also need recognize the prevailing trend in our industry such as days of the week. We certainly don’t want to compare how are compaign did that run on the Sunday and the one that ran on the Monday. Because unless it was the massively different, monday will win every time and we will conclude the wrong thing.

Those trends are more than just the days of the week but seasonal as well. For example one of your keywords where you saw a huge spike around the beginning of July view assume that your new Adwords compaign did that? Ofcourse not, the searches go up for every one around that time because it’s fourth of July and that’s reasonably obvious though.

But what if we sow even a bigger spike earlier like in November, could’ve been from Halloween. Well we know the first step in the analysis is segmentation as i was talking in a previous post and we immediately see that all the traffic is coming from India and centered around happy diwali  keywords which like our Fourth of July causes the spike in fireworks searches.

Now this trend wasn’t initially obvious to us but by using the tools available we can understand whether we can claim success due to our marketing compaing or due to simply arising tied that floats all boats and have essentially nothing to do with our actual site or marketing. Now the key here is understanding these trends will allow us to compare apples to apples like at external factors and keeping our data in context.

Anyway now it’s time to go stretch your legs is been some time since you started to read this post so take a break and check out the following post.

~ Veto

Comments are closed.

Scroll to Top