Tested Advertising Strategies Respun For SEO

"I have seen one advertisment sell 19-1/2 times as much goods as another" - John Caples

I've been browsing through a pile of my old marketing books looking for tried-n-true techniques that could be applied to SEO in 2008. Here are some examples of ads that worked last century:

"They laughed when I sat down at the piano - but when I started to play!". And "When thin film covers teeth, smiles lose fascination" . A personal favorite "How a strange accident saved me from baldness".

Trouble is, in 2008, these ads come across as hokey.

However, much of the underlying psychology of these time-tested techniques is pure gold, and can be directly applied to search marketing and social media. Over the next few days, I'll look at strategies that can help you grab and hold visitor attention. Some of this will be old-hat to SEO pros, but hopefully it will help those new to the game :)

First up - the value of testing.

The Key To Success Lies In Perpetual Testing Of All Variables

John Caples, author of "Tested Advertising Strategies", outlined two classes of advertising:

1. The Testers
2. The Non-Testers

The idea is simple: decide on a desired action you wish the user to take i.e. making a purchase; then link this action back to the advertising spend/keyword term. You also test the wording of one page against another. Run with the winners, and cut the losers. Repeat.

This is known as split-run testing. This process was developed by direct marketers, and has a natural fit with search marketing. An ocean of material has been written about how to do this, so I won't reinvent the wheel by repeating it.

Here are three great resources regarding split-run, and more accurate, but complex, multivariate testing:

• Tested Advertising Methods (Prentice Hall Business Classics) , John Caples
• Split Run Testing
• MultivariateTesting

There are problems with this type of testing, however. You've got to watch out for small sample sizes and statistically small variations. Sometimes measuring the exact same page against itself produces different data!

The payoff in split/run testing is in the big swings in user action. If you're not seeing big improvements, then you've probably got your landing pages about right, and split run testing will offer incremental value, at best.

SEO: What Data Do I Test?

We're spolit for data.

The speed at which we can test and obtain data regarding visitor behaviour patterns is unprecidented. Before the internet, direct marketers used to run a series of trial campaigns in print. Can you imagine how difficult it was to measure response? And how much time it all took? These days, we've got detailed, automated analytics in the form of server stats, and we can build, test and analyse campaings, often within hours.

One trap those new to SEO often fall into is using the wrong data and metrics. One terrible, but often-cited metric, is ranking, as ranking doesn't tell you anything about utility. For example, how much traffic will the ranking generate? If the ranking does result in traffic, is it the type of traffic I need to fulfill my objectives?

We can find this information out by running a few simple tests.

How Can Testing Be Applied To SEO?

Keyword Testing With PPC

How do you know if the keyword you intend targeting is really worth targeting? A keyword term may have considerable volume and little competition, but if those searchers are intent on research, and you are trying to sell something, then your effort is wasted.

One way you can test user intent is with a short PPC campaign.

PPC offers you some valuable data points. Firstly, you can test actual search volume, as opposed to estimated volume, simply by running an ad. The Google data provides these numbers.

Secondly, you can measure the intent of the query by measuring click activity.

Determining searcher intent is important. If you aim is to sell via your site, then you want to target people who buy. Often, this information is contained within the keyword query itself. For example, the intent behind "buy x online" is clear. However, the intent behind "San Francisco Houses", less so. How do you measure intent if it is unclear from the search phrase?

You can do this by crafting different adwords ads - i.e. some commercial in nature, some informational - and testing them against one other. You can further test visitor intent by crafting landing pages that demand the visitor commits to an action that causes them some level of "pain". i.e. filling out a form. If they aren't prepared to engage after clicking a PPC ad, they're unlikely to do so just because they found your pages in the organic listings, either. If you find a PPC term with a high level of user engagement, chances are that keyword term is gold on the organic side, too, and therefore a great keyword to target.

After you've validated keyword phrases in this way, you can then set off on your SEO campaign, armed with the knowledge that your keyword terms should underpin your business objectives.

One flaw in this approach is that the searcher has clicked on a PPC ad vs an organic listing. This very action tends to indicate a commercial intent as people who do not have a commercial intent tend to ignore search advertising altogether. However, it will give you a ballpark idea and could save valuable time and effort, especially if you're targeting generic, non-specific keyword terms that don't clearly convey intent.

Keyword Spinning

I spotted this technique a while back on BlueHatSEO. It is a fantastic technique for testing and refining SEO on big sites.

It can be difficult to know what keyword variant attracts the most visitors. For example, does "Myspace Pimps" get all the traffic, or does the variant "Pimps On Myspace"? Keyword tools often aren't sensitive enough to reveal this data, and it can be time consuming to monitor, test, and change thousands of individual web pages.

Instead, try adding a counter to each page and decide on a delimiter. Say, 5 visits per page per month. If the page views are less than this figure, use an automated script to scramble the title tag and the headings to produce a different keyword order. Reset the delimiter, and see if the new keyword order receives more page views than the previous order. Your site will self-optimise, based on the results of each test.

This is an excellent application of an established advertising method - it's an automated split/run test - and applying it to SEO.

Useful Tools & Resources

• A detailed analytics program. Google Analytics is free.
• Split-run testing software. Google's Website Optimizer is also free - here is a great Google Website Optimizer introduction.
• Coding ability is useful, but not essential. If you're stuck, there are plenty of coders available for hire at Elance.