3 ways to fight site abandonment

December 14, 2009
Source: Imedia Connection

Article Highlights:

  • Your content might be too rich, and it might take too long to download
  • Where an analytics tool will know whether a particular tag loaded, a "sniffer" will know many more details about the situation
  • Armed with load time data, your next task is to break the news to everyone on your web team

It's like the Swiffer commercial, where an old sponge-mop is outside a window, peeking in at a housewife who has moved on to better mops, while "Love Stinks" plays in the background. The old mop is trying to win her back, even going so far as to hire a Mariachi band to court its perfidious ex-partner.

In this case, you're the sponge-mop.

We're talking about site-abandonment, and especially early abandonment — the kind that leaves you feeling like a toddler dropped off at a nunnery, and you never know why, though you spend your life trying to find out.

Digital marketers, especially those who rely on campaigns to drive ecommerce, suffer greatly from abandonment rates, which can run as high as 80 percent for a typical landing page. All marketers are affected by it, even if their KPIs don't involve selling something online. In fact, any failure to complete a scenario — any loss of traffic anywhere in the funnel — is a symptom of abandonment: The target audience just didn't want to proceed. And isn't that the entire reason we perform web analytics at all? To find out why we can't get every single site visitor to do what we want them to do?

Perhaps a thousand types of measures have been devised about visitor behavior. The sheer number of report choices can nearly suffocate any search for the truth. There are so many interconnected measures that, at a certain point, it becomes like a house of mirrors — no one can tell what's the real number or which measure can be relied upon for some type of objectivity.

In your search for objectivity, have you ever wondered if people's behavior is affected by how long they have to wait for your page to load? Have you ever abandoned because of load time?

Here are three ways, based on understanding load time, to help put abandonment in perspective — and perhaps even decrease your abandonment rates:

Step 1: Admit you have a problem
The latest data shows that abandonment increases dramatically after a wait time of four seconds. In net time, four seconds is like four hours of driving: "Are we there yet?" becomes "Let's just pull over." And all of your marketing efforts suddenly become road-kill.

Slow load time can be caused by a number of problems, some of them obvious, some of them hidden, and all of them addressable.

The first thing you'll want to look at is content richness.


You may be thinking content richness is what everybody is looking for. How can you survive without Flash, without embedded video, without whirling, dancing eye-candy for a user base that seems to think a chance browser encounter should be as glittery as a stroll down the Las Vegas Strip?

Your content might be too rich, and yes, it might take too long to download. It might ask for too many plug-ins. It might be unsupported by your target user's browser.

Or, your page load problem may be caused by something that has nothing to do with over-rich content. It could turn out that your server is slower than a dinosaur in a dunce-cap; then again, you might have a peerage issue somewhere between you and Timbuktu. The hunt for KPI completion does not begin and end with earning kudos from your colleagues in a closed environment where everything runs at optimum speed and every link works just the way it ought to.

You need to get out of the conference room and into the real world. That's where the real story of page load begins and ends. And, as noted above, the trouble may not even be content richness. But how do you measure that?  Contact ICND to fix this problem now!

Step 2: Measure with the right yardstick
There are a number of tools available to measure load time: Gomez, Keynote, TeaLeaf, Atomic Labs — these are just a few of the companies with load time measurement offerings. There are also some clever ways to tag pages for load time information, but few of these options are bundled products as yet (though they may be available from your web analytics professional services vendor).

The products mentioned above tend to work in a somewhat different way than the typical analytics tools do. In other words, they don't concentrate on click-stream data and don't need tagging or log files in order to function. In fact, they are, generally speaking, separate architectures that look at all of your traffic, all of the time, and do a variety of things with that information. Sometimes they are called "sniffers" because they sit somewhere between your server and your user and pick up all the traffic (kind of the way a bloodhound sniffs an evidentiary trail).

So for instance, where an analytics tool will know whether a particular tag loaded, a "sniffer" will know much more about how long it took for the page to be requested, how long it took the server to get it out the door, and how long it took for the various packets sent by the server to show up in the remote user's browser.

I had indicated above that some of the tools have significant other uses. TeaLeaf, in particular, has a core function that allows the marketer to play back, literally, a browsing session. Some folks call it "Tivo for web analytics." It may be a bit much to deploy all of that solely for the purpose of locating a page-load problem, but if it is taking place, TeaLeaf will probably see it.

A tool like Gomez, on the other hand, sets up a network (around the world) of synthetic testing nodes, enabling them essentially to mimic requests from anywhere and measure your response time for anything and everything you want to measure. You can look at certain parts of your transactions and find out, for instance, how long that is taking in Albuquerque, Inchon, and Luxembourg all at once. The only drawback is that generally these tests are synthetic (in other words, not involving actual users).

A clever tagging solution will have its drawbacks too; for instance, it won't be able to know when the server request was made, or how long it took for the first tag to load. But then, the moon is not the sun, yet both have recognizable assets.

Linking load time directly to abandonment typically requires an additional integration step. For instance, if you are trying to find out how a sniffer-tracked load time problem is affecting your campaign ROI, you should be prepared to do some fancy dancing with your analytics tool — but with the right kind of integration, the dancing can lead to marriage, and you'll see stars.

The important takeaway is this: While I'm not going to claim it's easy, you can certainly find out how long your pages take to load, and you can, with the application of some elbow grease, understand whether that is affecting abandonment and conversion overall. In all likelihood, load time is affecting abandonment. Now you'll know how much, and perhaps even why. 

Step 3: Come back armed with the facts
This part is probably the least fun, but most rewarding once it's done and accepted. Armed with load time data, your task now becomes a matter of breaking the news.

Your creatives, your developers, and/or your IT folks may all dislike your news intensely, but you've got bigger worries than that if people aren't completing scenarios.

I know from experience that creatives have a soft, mushy spot in their hearts for visual extravaganza. For them, a day without Flash is like a week in the pokey, and they will start to complain that they can't deliver value without it — and some of that may even be correct.

I also know from experience that IT folks will defend their architecture to the bitter end — or at least until management coughs up enough bucks for some faster servers. But even the best server jockeys must sometimes face the fact their horse is not in the money.

Developers will say it can't be fixed. The work that's been put into your current site cannot be taken apart and reassembled in a more efficient way "just like that." And they are right (but they were, perhaps, wrong for building it "like that" in the first place.)

In any case, while your developers may be architects of the best parts of your site, they are also architects of the worst; and when people are bailing before your pages load, that's the worst. As a marketer, you may need to force-feed them some facts, and hold them accountable for making sure you don't have to tell management that the site's most important pages aren't even being seen by the audience.

You've probably long sensed there was some "other" problem out there that was keeping users from delivering for you. You could never be sure what it was. There were fingers pointing 'round the table during your last review, and nobody seemed to be at fault. Knowing about load time will help reduce the blank stares at meeting time, and set you up for better success at conversion.

Often, it's the unturned stone in the path that hides the all-important clue. My suggestion: You should at least turn over that stone.

Andrew Edwards is managing partner of Technology Leaders. 

Leave a Reply

Your email address will not be published. Required fields are marked *