Isn't January great for traffic?

I was sitting down doing my weekly reporting for the board (plus the several thousand other people who having discovered I do it decided that they wanted a piece of the action), when it occurred to me that January is great. Great for traffic to your site. Great for your web owners. Great for your web analytics managers who can take all the fame and fortune.

But why are they so great and where does all this extra traffic come from? These are the questions that you need to ask yourself and look into the data to find out why:

January is great because December is generally accepted (bar the e-tailers) to be rubbish. Everybody is out Christmas shopping and far too interested in their Christmas parties and eating turkeys and... Basically everything except going to your website to find the latest news, buy their car insurance, looking for a job, whatever-your-site-does-unless-it-sells-presents. You may be asking 'so what? How does that affect my traffic in January?'.

Here is a graph showing traffic to my imaginary website. It has a 10% increase in traffic each month and I am now getting 3.5 visits (page views, visitors, whatever) a month instead of the 1 I was getting last January.


That is all well and good, but we think that December had a 10% decrease, just because everyone is in party mood. What does that do to our graph? Well, all that happens is that December is 10% lower than November, but January (not affected by Christmas) should remain the same, right?



Now it looks like January has had a real big increase and all I've done is changed one month. January is up 30% on December and is breaking all records. Great - give Alec a raise.

30% increase again next month and suddenly we are away and this is the greatest website on earth. Except one thing. February is a short month. It only has 28 days (except for leap years, etc, and so on - trust 2008 to come back to haunt me). That is 10% fewer days than January. So if we continue our model of 10% increase each month, we'd probably expect February to equal about the same as January (given month ends).

Lets look at this chart again. No increase in Feb? "That raise wasn't deserved Alec. Give it back."

"Woah, hang on, look at March and the mega traffic again - give me another raise!"

Ok - this is starting to get silly now. Lets go back to what we really want to do and find out how to do it best.

Are monthly comparisons working? No.
Why not? Because not all months are equal in size or what is actually happening in the real world
How do we get around it? Giving Alec another hefty raise and getting him to come to the party with a slightly different chart (or do it yourself and get your own raise):

• Use your charts weekly. All weeks are 7 days long
• Some times of year are different. Don't look at week on week trends - look at an overall picture
• Go year on year. You can see if the overall trend is continuing.
• Work out why some weeks were higher than others and label them in your charts (and in your web analytics solution if you can)
• Remember bank holidays, public holidays, Olympics, World Cups, etc
• Remember your marketing, site downtime, etc

The first thing you should do is to get that chart out (week on week for the last 52 weeks, one for page views, one for visits and one for visitors) and get excel to draw a linear line through the points (it's easy - right click on the data in the graph and choose 'Linear'). Linear lines will show up any points that are above or below that look unusual and will show you a trend. Other types of trends just show up that you've got an A-Level in maths and can draw exponentials.


Now shorten it to the last 26 weeks. Does it still have the same upward trend? What about the last 13 weeks? 6 weeks? Now you can really tell if January is still going upwards or whether your 10% upward trend is dying off and you should do something critical about it.

Having said all that - sometimes January just is great. How much have you used the web this year? More than ever? So have all the people who have just been given new laptops for Christmas. So have all those people who are trying new websites as a new years resolution. So have all those people who have been given an end of year review telling them that this year they are going to be web first. Your job is to work out why they are using it more often and what they are doing on there (so that you can improve it).

Omniture's SiteCatalyst HBX

Well two months ago I wrote about how the end of HBX wasn't quite here yet. And it still isn't here, quite yet. However it may be a lot closer than I anticipated two months ago with the announcement on Thursday 17th that the merger had officially completed. I'll take you through the interesting points of the press release and the 'welcome' email that we got on the same day.

HBX Analytics will be rebranded as Omniture SiteCatalyst HBX, and will continue to be supported until the key features have been integrated into Omniture SiteCatalyst.

May be a bit too easy to read between the lines on that one. The email told us that they are "going to continue to support the HBX product line" which kind of follows from the excerpt from the press release above. However this doesn't appear to move away from the "migration program".

Hmm all interesting, but what does it mean? Well as I mentioned last time there are a couple of options going forward for the customers and it really depends on their user base (of the tool, not the site) and the purpose of the site.

Let me put this implicitly at the start - moving analytics tools is not simple. I've done enough implementations in my time to know that even if you can create a template to follow, you'll spend long periods afterwards tweaking it to make it right. So when we redesigned a website recently, we went through the whole rigmarole of working out what our users wanted and marrying them up with our business goals. In theory, choosing your web analytics tool should be no different. You need to know who is going to be using it and what they want to get out of it. Then you have to marry that up with the goals of the website (or company if there are many websites).

So lets start from the beginning. I suspect that most sites will fall into a couple of groups (but there may be other ones that I am missing out on).

eCommerce, many analysts, few users - this is probably a common situation. When I say many analysts, I'm probably still thinking of between 2 and 10. What happens in this case is that the analysts use HBX to produce reports, analysis and recommendations to the Business. The Business don't use HBX, because the analysts understand it better and can provide better recommendations or they tried to once in the past, got the wrong end of the stick and caused a storm in a teacup when they blamed a whole department for not working properly. In this case it is entirely up to the analysts what they want to use. It's personal preference.

Site Catalyst works quite well and a migration would be smoothed by the fact they don't have to explain to anyone who to use it. Also because they are now the same company as HBX they won't have to think about paying for two tools at the same time whilst they migrate. Equally they could think about taking up core metrics on their offer of migrating for free. Both these tools give a few more reports that HBX doesn't, although they may cost more. The other advantage this lot have is that whilst the analysts work their socks off to ensure that it is right, the users of the reports (the Business) may never know what has happened.

eCommerce, few analysts, many users - given a lack analysts skills and the ease of use of HBX as an interface this is also common. Here you have not only the issue of migrating to the new tags (not easy without analysts), but you have to train more users on how to use the system. This means you need to give this more thought - either the users think using it is Business critical or they use it for ease. Before you know it, they've got fed up and are going back to the data adwords or affiliates give them and don't worry about conversions.

Critical in this situation is firstly overlap of tools so that you can migrate the users in your own time. Secondly is that you have someone who understands the Business enough that they can explain to users (who aren't going to be as web savvy) where to find the things they used to look at and why they should look at new metrics. Here the advantage that Core Metric has is the added consultancy, but the disadvantage is the less ease of use. The people may choose to move to a tool like Web Trends which is fairly intuitive as to where things are - but this will require consultancy to set up properly in the first place. The SiteCatalyst is probably a good option too - because this will reduce the costs of the overlap period.

eCommerce, few analysts, few users - it wouldn't surprise me if this was fairly common as well. There are a couple of simple options here (apart from giving up and going home). You can either hang around on HBX for as long as possible and hope someone develops an interest so that you can move into one of the two above (that'll give you a good excuse to leave HBX behind). You can question why you are paying any money at all and move onto Google Analytics or the soon to be released Microsoft Gatineau. You won't be able to do as much in depth analysis (probably) but then you weren't doing much anyway.

Media, few analysts, many users - ok, this is a web first world when it comes to media these days. If you don't fall into this category I'd be amazed. Unless of course you happen to be in the fortuitous position of having many analysts. Here it is a bit complicated. If you're anything like my company, we have 50 odd websites, with 3 different sections on most of them, all owned by three different people. In the end you have 500+ users when you add in dev, creative, user research, sales, Marketing, etc. Also because you are a media site you probably want to keep costs down to a minumum.

You have two options: Run into SiteCatalyst as fast as you can and hope that you can ride the wave of help from the Omniture 'migration program'. This will hopefully allow you to give your users the ability to do simple things very quickly (good). Meanwhile going hell for leather with the developers to retag your site in as similar way to how you set it up in HBX. Things to think about here are your API linking data into the front end may have to become manual for a little while, as will your report builder reports.

Or, alternatively, you can hold out on HBX as long as you can until you are the only one left that they are supporting and they tell you that you can't stay or they find a way of transferring your data without you changing your tags. This way at least you can spend a long quality time looking at SiteCatalyst and working out how to set up your tags in the optimum way for all of your users. Plus it gives you licence to work out how to set up the API from Omniture and set up any excel reports.

Easy decision? Don't bet on it.

Hitwise Part Deux - for the Sales team

We'll continue in the theme of Hitwise after last week's 6 tips. This week we are going to go into some of the other features of Hitwise that allow you to do some demographic and lifestyle based queries on yours and other sites. I'm also hopefully going to go into a bit more detail on it and hopefully it will show you some of the uses that you can get out of it after the wow factor.

Lets revisit what we said last week:

Hitwise has developed proprietary software that Internet Service Providers (ISPs) use to analyze website usage logs created on their network. The anonymous data sent to Hitwise from the ISPs include a range of industry standard metrics relating to the viewing of websites including page requests, visits and average visit length.

Ok - what that means is that as well as having huge volumes of data about which sites each of the users have, they can also start doing some investigation into the location of the users. And if you know the location of a user then you can start running all sorts of socio-demographic reports. Also because of the way Hitwise queries the ISPs logs, they can also extract information based on the bill payer for that particular usage giving it huge volumes of demographic data.

Social Grade

Firstly lets look at how Hitwise looks at social grading. I think there is a fairly accurate description of it on the wikipedia page, but I am going to give you a shortened version. People in social grade A are posh and people in social grade E are vagabonds and B, C1, C2 and D are everything in between. That's a bit of a simplification, but essentially your social groups of ABC1 (not the American television channel) spend lots of money and those in C2DE don't.

Your website will have a combination of ABC1C2DE users. You can have a look in Hitwise (it groups A/B and D/E) - below is FWi



Wow - look at all those D/Es. We've got loads of them. The first question someone would say is "why do we have so many of them?" and "Can we get more of those lovely A/Bs please?". That may be true, but you have to look at this one in context. Firstly these figures on their own are meaningless. You need to firstly compare to the online audience - do we actually get more than the online audience does (use the 'swing comparison'). Well in this case the answer is still yes, but it gives us a much better picture. Whilst our site is more D/E than the online population, it's actually a bit more A/B than the whole online audience as well.

Remember that much of this information is based on post code and we are aimed at farmers. "But so are our competitors!" I hear you cry. Well exactly, lets try comparing it to our 'Agricultural' group rather than the whole online population and suddenly we've got far fewer D/E's than the total population and the C2 category is much higher. That's far more likely - C2 included agricultural workers and that is where we are aimed at. In fact, we have more of these C2s than our nearest competitor as well when we went to a more advanced swing comparison.

So what can you do with this I hear? Well firstly you can use it to help sell your on-site advertising. If I'm going for a general advertiser that isn't going for my market, telling them that according to Hitwise we have more A/Bs than the whole population. This should make my site far more attractive. If we're doing more bespoke selling then we can point out to, say, John Deere that we've actually got a bit grouping for the agricultural workers. They are the ones that they should be focusing on and we have them. Spend money with us.

Mosaic Lifestyles groups

Ok - now we are getting somewhere but this isn't anywhere near enough info for advertisers. Can we go deeper? Well yes - this is where the mosaic lifestyles come in. Mosaic is a split of all users based on where they live into 11 (sorry US guys, I know you have more over there) groups of people who range from A - K. Have a look at this pdf, it briefly tells you about it. Here is FWi's:



Do you see that nice big spike at the end? That is mosaic group K. Mean anything? No - lets add some context. Mosaic group K is the 'rural isolation' group. It tells you about them in that pdf back up there. The summary of these people is that they who live in the country side, that has predominantly been farming in the past, but is starting to get picked up by commuters. Our users probably aren't the commuters, but it's interesting to see what they are interested in (look at the photo in the pdf). Again it is useful here to compare to the online audience because it turns out we've actually got a bit more C (Suburban Comfort) than the total online audience.

Nice info for you to put in a sales pack isn't it? And actually, just a bit more than that, because it should be giving you ideas for companies that you want to get advertising from - it will give the advertisers higher click throughs and should therefore give you the scope to charge more money.

Mosaic lifestyle types


More detail Alec? Surely we don't need that much. Probably not, but in this case it gives a nice example. Our Ks in this example are sub split down into more detailed K57 - K61. They relate to:

K57 Summer Playgrounds
K58 Greenbelt Guardians
K59 Parochial Villagers
K60 Pastoral Symphony
K61 Upland Hill Farmers

This nice little subset shows us that really we're not aiming at K57 - these are urbanites who just happen to have a second home in the country. They aren't interested in Farming and the demographics bear this out - this is less than one percent of our total population. The K58s, K59s and K60s however fit our sites target ideally. They are wealthy, they are farmers and they probably own their own farm. These are great people to sell to if you are in the farming industry and they visit our site in abundance. Far more than the online audience when we do our swing comparison and far more than the agricultural industry as well.

If you were trying to sell something in the farming industry and did a search on Hitwise to find the most farm-wise community - FWi would be one of the first places I'd go. However the farming industry people don't have access to Hitwise, so we have to tell them and show them the data.

This tells a nice story about FWi - but all other sites will be the same. Proving to your advertisers that they are better off advertising with you than someone else is difficult. However if you have these tools then it can really help your sales pitch.

6 tips for using Hitwise

Because I can, I am going to continue the theme of writing blog posts about things I've presented on recently. It kind of makes sense that I've gone to all the effort of putting together a power point presentation, so I might as well write a blog post about it as well - rather than wasting the effort. Anyway, this week we are going to touch on the subject of Hitwise (something I haven't really covered before). So I suppose that I should really start by giving a brief synopses of what Hitwise is before I give you the big six tips on how to use it.

Often quoted on the web as a reliable source of data (partly because of their very good Analyst's blogs), rarely do we see posts talking about the best way to use Hitwise. This is partly because they have an intrepid support team who go around the world telling you how to use it. The only downside being that they usually don't work in a business that needs to use it. So here is the spiel:

Hitwise has developed proprietary software that Internet Service Providers (ISPs) use to analyze website usage logs created on their network. The anonymous data sent to Hitwise from the ISPs include a range of industry standard metrics relating to the viewing of websites including page requests, visits and average visit length.

Basically they take a bunch of data from ISPs and transform it in the way that a Web Analytics tool would. How does this differ from other competitor intelligence? Well Alexa, Nielsen, etc rely on you to download a toolbar to sit on your computer and then extrapolate the data - so you are only taking a small sample of people who know that have actively downloaded. Hitwise takes a large sample (up to a third of the UK population) so doesn't need to do any extrapolation.
Needless to say - Avinash has written about this before (a long time ago). What's more, I'm not even going to delve into the extras of the demographic or lifestyle information (which tends to be a nice to have, rather than a method for creating business decisions). What I am going to do is start with where everyone who uses Hitwise starts:

1. Use Hitwise to compare your traffic volume to your peers/competitors

It's all well and good going to your boss/CEO/whoever and telling them that you have an extra 25% visits/sales/sign ups this month. However if you go to your boss/CEO/whoever and tell them that you have an extra 25% traffic this month and your competitors have had a smaller increase, then you are putting some proper comparison in place. Hitwise allows you to use their traditional industry sectors or to create your own list of competitors/peers that you want to be compared against. It is important not to do this once and forget about it. You need to trend this over time to show how your traffic increases compare to your online industries (remember to look at market share and not rankings - rankings can be misleading if one site is way in front of another).

2. Use the Clickstream report to find out where your traffic comes from/goes to

Web Analytics tools will tell you when someone clicks on a link to your site. It's called the referrer. Hitwise will tell you the site that people were on immediately before coming to yours, whether they clicked on a link or typed your name into the address bar. This is important because it will tell you what your competitors are. The sites that you think are competitors rarely are - look at what your users think. They use your competitors. No really, they do. They also use your affiliates, your sites that may include brand marketing of yours or all sorts of things. This report will tell you the sort of mind frame a user was in when they came to your site. Track it over time - if you are trying to get the same traffic as a competitor through advertising (on or offline) look to see how the traffic coming from them changes over time. Now look at your downstream traffic. No web analytics tool can tell you where your users go afterwards.

3. Use the Clickstream report to find out where your competitors/peers traffic comes from/goes to

Wow - this is the first bit of real competitor intelligence that we've done. You can find out where the traffic to your competitor's website comes from. Look at it. Do you want that traffic? Does the source of traffic you thought was giving that website a real advantage only provide minimal amounts? Does an affiliate of theirs appear high? What websites can you steal from them? Are you a major downstream of their site? If not - who is and can you get their traffic as well? So many questions on this one I don't even know where to start. Actually I'd probably start approaching their highest ranking upstream site and see if you can get a sponsorship deal with them.

4. Find competitors/peers search terms

Do you know your own search terms? Of course you do, you've spent ages analysing that long tail over and over again to see what niche terms you can add to the list. But you're still worried that competitor.com appears higher than you in the rankings for the search term 'crazy blue widgets' and that is all you sell. Well why not look at the search terms that drive traffic to them and you might discover that, hey, you are below them, but they don't get much traffic from that term. They get it all from 'foolish blue widgets'. You don't even have that term on your site. Lets put it on the site and get in on some foolish blue widget action. In fact, why not go the whole hog and bring up a list of terms (using Hitwise answers) that they get traffic for and you don't. You could use them all (or all that are relevant).

5. Find search phrases around search terms

This is important. You've optimised your site for your key search terms. They are driving in traffic. But your competitor websites are still doing better than you. It's probably because they've farmed the long tail. Have a look at the keywords section where you can type in a term and it will come up with all the phrases that include that term and get volumes of traffic. Look at how much traffic the term gets compared to the tail. Get some of those search phrases in your website. Make them become relevant. Tailor your site, to what users are searching for. This isn't spamming - this is making your site more relevant to more users. So you've find the search term - check to see how much of the traffic you get from it. If it is a large proportion, you may want to focus on other search terms. Which sites are doing well for these terms? Why are they doing well? Find them and see what they've done.

6. Find what your competitors/peers are bidding on

Use the search facility in Hitwise to show you what search terms your competitors are bidding on. This is important if you are about to start bidding on terms yourself, you need a list of what all the other sites are bidding on so that you don't just start a bidding war. Choose your terms wisely - those nobody is bidding on will be cheap. Those that your competitors are bidding on may convert well. They may even be terms that you have the organic listings sown up in, but that you want to ensure you have the whole keyword space. Most importantly though, don't get bogged down by this - they are probably bidding on them, because they haven't worked out how to optimise their site for those terms.