Pod1 Blog

Magento commerce has the ability built in to run test with Google Website Optimizer. But these are limited to the default style templates and limited to the category and product detail pages. This is less than ideal if you want to go onto the big guns and test your checkout for example.

At Pod1, the Magento Development team and the E-Commerce Strategy Team have come up with a simply way of testing all aspects of a page or section within magento.

We simply use the cascading element of Cascading Style Sheets, CSS, which allows us to overwrite styles. We leave the control variant untouched and simply use Google Website Optimizer scripts to load an additional stylesheet with corresponding styles in. These of course are altered in a way that allows us to change the layout of the page quick and efficiently. Either hiding elements or repositioning them or adding to them. If setup that way, you can do the same with JavaScript using the DOM model. Change the “Add to bag” button to a “Buy now” without the need for having graphical representations produced.

This way, we are able to reduce the cost of producing variants dramatically as all you need is a front-end developer and not the whole team. This makes for remarkably quick turnaround that yield results much faster than changing the whole site using all skill sets of your agency and implementing costly multivariate software that requires you to produce every page two or three times.

If you’re interested how to get the most out of your Magento Commerce Package, have a click around our blog.

IBM have bought Coremetrics, a major vendor of hosted analytics services. Since I’ve been implementing eCommerce sites, Coremetrics has been one of the few serious alternatives to the (free) Google Analytics service that most people use, and it’s now going to join the IBM world. The competition in the analytics space is now dominated by big Internet companies:

• Adobe own Omniture
• Yahoo own IndexTools
• Google owns Google Analytics (which used to be called Urchin)

And now IBM own Omniture. Webtrends remains independent, but it’s telling that these major companies have decided they need to own serious, serious analytics offerings. Data is key to CRM, to site personalisation, to content optimisation, to so many things, so the company that has the best data has a strong competitive advantage.

Incidentally, this is an interesting turn in the evolution of the IBM product set as well. They’ve been acquiring companies for years and adding their capabilities to Websphere, and this is another one of those stories. What that has led to (according to the experts I speak to) is a rather loosely integrated set of software tools, that don’t have the same origin architecturally, and aren’t therefore all that consistent in user interface or structure. They do get assimilated to an extent, but obviously one doesn’t spend hundreds of millions of dollars on a company to rewrite its code, so the whole collection is a bit of a mixed bag. Very functionally rich, but quite hard to work with.

It’s been a while since I last posted IE6 stats here. The good news is that Internet Explorer 6 is losing more and more ground. Not least, one would hope, because of Microsofts push to upgrade to IE8. The analogy of drinking 9 year old milk is pretty much what overcomes our Designers and Developers here when we talk about IE6 as well, so well done Microsoft Australia.

Another factor could be that Microsoft was forced to deliver the “Browser Choice” option to desktop machines where many users probably upgraded or switched. I don’t think that seeing Google ban Windows from their office machines will have an effect though. Interestingly IE6 is loosing it’s share in terms of revenue faster than it looses it’s share in traffic. We actually have the first clients pulling IE6 support from their retainers and invest the money saved into better marketing or a better strategy.

However, I don’t see IE6 vanishing as fast as we would like to but we’re on our way.

Below are the figures of our sample of about 3 Million p/m visits across the Pod1 Client Portfolio

Published on New Media Knowledge by David Hefendehl, March 2010.

Hands up who uses Google Analytics and gets most Organic- or PPC Revenue from branded terms (Brand name that is). All of you? Thought so.

Here we often have discussions around why you should do PPC & SEO on branded terms when you get most of your traffic and possibly revenue from organic branded terms anyway. I guess we need to take a closer look at what’s happening here.

There are two reasons why you get more traffic from branded terms when using the wonderfully free Google Analytics.

Convenience
Many, (many many) users have Google or any other search engine setup as their start page. Rather than typing your URL into the address bar they just type it into the nice and friendly, large text field in the middle of their screen. Ask a random sample of your users if they know what the address bar actually is. Ask your office mates. I swear most won’t know what you’re talking about.

Last click counts
This is the second reason why you get so many “head keywords”. Users come to your site in different stages of the purchase cycle. They most likely start of on a long tail keyword like: “Pink polka dot jumper”. They find a few pages, check them out and might even land on your page if you’ve done your Search Engine Optimisation right. Now they do something else, go back to work; take care of the kids; go out for a coffee and then remember that website where they saw that great Pink Polka Dot Jumper. Most likely they have no idea how they got there or what the product was actually called (I’ll do a post about slang later this month). What they do remember will be your brand name. So instead of going back to Google and using Pink Polka Dot Jumper then click through endless results, they type in your brand name and, shazam, you have a head term conversion.

The problem with Google Analytics is that it only counts the last click from which a user came to your site. So even if they have come from a natural long tail keyword they are likely to convert on an organic head term or even branded PPC term. Analytics will only count this last click, but not the first one or second one or third one. So your stats will be skewed and it might look like only brand terms convert. There is of course and additional problem that many users will use their lunch break to research but convert at home, i.e. switch computers. Forget about these users; it’s going to be a while until we get those tracked.

So should you do SEO and PPC on your own brand name or the other way round: “Should you even bother with this nasty long tail optimisation?” Yes is the short answer. You actually have to do it as otherwise you’ll simply die a virtual death. Always think of the purchase cycle your customers are in. Unless you’re a really really really big brand (Coca-Cola level I’d say) your customers will start of on a long tail keyword search. Most searches are 3 or more keywords long (see: http://image.exct.net/lib/fefc1774726706/d/1/SearchEngines_Jan09.pdf) so it is fair to assume that is the same for your customers. Brand names tend to be a bit shorter. If they don’t find you for Pink Polka Dot Jumper they won’t remember your brand name as they won’t get in contact with it. Not doing PPC or SEO for your own brand terms (PPC even for the sake of getting cheap clicks and preventing competitors to bid on it) will lead to them, if they do find you on the long tail search, not finding you on a branded search.

Now that it is clear that you must do SEO for both, long tail and head keywords what should you attribute the traffic to?

I guess the only way to find out is to launch a PPC campaign on your own brand name and see how far organic traffic drops. The difference should be your convenience clickers, the rest probably very engaged customers.

Discussion more than welcome!

P.S. Yes there are ways to hack 1^st touch into GA, check this out: http://www.seomoz.org/blog/how-to-get-past-last-touch-attribution-with-google-analytic

This article at Internet Retailing reports research that suggests millions of online purchases were abandoned last year, due to poor website performance.

It’s previously been reported that a 1 second dip in page load times can lead to a 7% dip in on site conversion rates, and it’s clear that a subjectively fast website makes more money than a subjectively slow one. The problem is that infrastructure can be expensive, especially if it needs to be maintained all year for the occasional seasonal / promotional spike.

There are solutions to this problem, though. We are working with content delivery networks to deliver the high volume, large files containing images, Javascript, style sheets and such like very cost-effectively, meaning that the load on the core web-server isn’t so heavy. Our smaller clients pool their hosting resources, so they all benefit from load-balancing, multiple servers, and the performance and fault tolerance that comes with that. This year more retailers will be able to benefit from a technology called ‘private clouds’ which will enable companies like Pod1 to offer scaleable hosting on demand, on their own farms of web-servers.

It’s important to be proactive about this, if you’re to avoid revenue slipping away that could be caught. We would suggest something like the following:

• Assess what your maximum load might be on your website (this is a simple spreadsheet task that we do frequently)
• Load test your current infrastructure, capturing performance data and lots of log file data during the test
• Fix the performance issues, upgrade the hardware where necessary, but often simple application changes can help, as can cheap upgrades like content delivery networks
• Re-run the load test

Load tests should be run periodically: upgrades can cause performance issues sometimes, and other factors (like the performance of the company running your domain name services) can change over time without being immediately recognised.

The DMA (Direct Marketing Association) has released an insightful report on email marketing for 2009.

You can download the report here:

DMA National Client Email Marketing Report 2009

It features intersting and useful stats which give a broad understanding of the email marketing landscape in the UK.

Front cover

Excerpt from report

About the report:

“This is the third year that the Email Marketing Council has run this client survey providing
the supply side drivers and a valuable companion piece of research to the National Email
Benchmarking Report.

Whereas the National Email Benchmarking Report normally surveys the ESP’s and
therefore the demand drivers, this Client Email Marketing survey asked the client
marketers to provide their answers to a mixture of response rate data as well as
attitudinal questions thus giving the DMA membership a perfect companion piece to the
top line industry figures available in the quarterly National Email Benchmarking Reports.”

Google has just added a whole load of new functionality to its Analytics Tool. Watch the videos below for a demo of the new features and let us know what you think.

Many online retailers will see the busiest trading period of the year between now and Christmas. In lots of ways, they’ve been preparing for this all year: planning their product range, ordering extra stock, hiring temporary staff to cope with the extra demand, and keeping non-trading activities to an absolute minimum. Advertising has been designed and produced, and seasonal page designs for their websites are all being coded and tested.

Despite the economic downturn, Christmas will very likely be a busy time, but how many retailers have thought about whether their websites will be able to deal with all that extra traffic? Will they slow down? Will databases and web-servers be able to cope with the extra traffic? The extra shop staff have been hired so that sales doesn’t literally walk out of the door, but how ready are their online equivalents?

Clearly if your website is down because of extra traffic, it’s costing you money. But research shows a slow website can lead to a lot of lost sales too. Retailers should commission load tests that simulate large volumes of traffic, and these will highlight the issues before they start to cost real money. Extra server capacity can be commissioned short-term, and again this can be turned around fast, so it’s not too late to do something.

Having enough stock for peak trading is also a challenge. Too much stock means capital tied up for months and massive discounts in sales in the new year. Too little stock means lost sales and lost profit. Managing stock optimally across all channels and all stock holding locations makes a big difference here, and there are companies who we partner with who have excellent solutions in this area, allowing retailers to view and re-allocate stock against sales very quickly and efficiently. True multi-channel inventory and order management can save a retailer enormous amounts and generate substantially increased sales, and it’s within reach for a lot more companies than it used to be.

Planning to make the most of the seasonal uplift is an essential task. Particularly in these challenging times, retailers need to make the most of what’s coming up, whether it’s on the website or in their physical stores.

Omniture emailed me last night to tell me that they’re being acquired by Adobe, a very interesting acquisition reported by Bloomberg here.

At first sight, the development looks odd. Adobe, home for Photoshop, Illustrator, Flash and a whole host of web and print design tools, might not seem like the most obvious home for Omniture’s analytics, personalisation and merchandising tools for websites. Adobe’s core business is moving, though: most designers have traditionally worked with Apple computers, and Apple continues to make inroads into that business with its own products. Microsoft is even developing an offering of its own which isn’t that bad. Faced with that challenge, Adobe have been growing their web development business steadily, by acquiring Macromedia (the developers of Flash and the Coldfusion development suite) in 2005, and Scene 7 (the multimedia manipulation and presentation tool popular in eCommerce). The Adobe web offering is getting more and more extensive, so combining it with tools that measure, personalise and merchandise the web seems a logical next step. And the synergies between the tools look very interesting: consider a Flash website dynamically merchandised by Omniture Merchandising, for example.

It’s also worth remarking how well Adobe have done in combining past acquisitions with their own business: Macromedia and Scene7 seem to have been thoroughly assimilated. The management team must be confident they can repeat that success with Omniture.

It looks like this transaction could lead to lots of interesting new thinking and products, so we’ll be watching and waiting.