This is the final installment of our PPC copywriting series. Part one discussed the importance of unique selling propositions in ad copy to communicate why a customer should buy from you rather than competitors. We followed this up with strategies for continual ad copy split testing. Today’s post assumes you’ve found a clear winning ad by testing one variable at a time, and you now want to test it against radically different copy.

I learned of this tip through a fantastic free video from Stompernet explaining the AdWords Triangulation Method (38:50 minutes through the video).

Here’s the danger in testing a winning ad (as your control) against experimental copy - if your new creative is a dud, you could lose 25-50% of your click through rate potential if your ads are showing evenly, depending on how many ads you’re testing.

A smarter approach recommended by Stompernet is to create 4 identical copies of your control ad, and one test ad. This ensures that your “safe” ad will show 80% of the time, and your test ad only 20%. Plus, the control ad’s solid click through history will cause it to rank at higher positions than brand new ads, so creating new copies reduces this bias.

I think you could also pause the control ad, and create 4 new copies to eliminate the bias altogether. But it’s nice to have one copy that’s going to achieve better positions at lower click through rates, especially if you’re continually testing. If you can tolerate more risk, you could create just one copy of your control so you’re running 3 ads - 66% of the time you’re running your control copy, 33% of the time you’re getting higher positions and lower CPC from the click through history of the established ad, and 33% is your experimental ad. Then you can segment out the control ad with the history to compare performance of your control copy vs. new copy, without the ranking bias.

Yesterday we talked about persuasive copywriting for PPC ads, and as promised, today we’re going to talk about strategies for split testing ads.

I’ve heard most PPC experts recommend you test headlines first, because they are most visible and are believed to influence click through rate more than body copy (because people have a tendency not to read things if they don’t have to). But I’m going to go against the gurus and suggest how you could test offers and value propositions first, using a hypothetical campaign for “wireless headphones.”

The Strategy

This approach uses Dynamic Keyword Insertion in headlines for your first few rounds of testing. You have more room to market yourself in ad copy than in the headline. You can test offers, calls to action, value propositions — copy that answers the question “why should I buy from your site?”

Dynamic Keyword Insertion means Google will match your headline to the searcher’s query. If Jimmy searches for “wireless head phones” or “cordless headphones” the ad would adjust itself accordingly for maximum relevance so long as that keyword is in your Ad Group and fits in the headline. If Jimmy searches “I’m looking for killer wireless headphones” (a long tail term), as long as “wireless headphones” is broad or phrase matched, a default keyword would display as the headline. In this case, simply “wireless headphones.” You’ll see an example of how to use this feature in the first screenshot, or read more in Google’s DKI tutorial.

But we’re not going to use keyword insertion in the ad text - there’s no need to keyword stuff. We want as much space for testing compelling offers, not redundant keywords. It’s not SEO, it’s not the year 2001 and we speak English here. Here’s an example of totally useless keyword repetition (real ad):

Alright, let’s get started with a hypothetical example of 5 rounds of PPC ad testing:

Round 1

First you’re going to pick 2-4 versions of an ad to run evenly against each other.

Headline: Use DKI for your headline

Ad Copy Line 2: Make sure you have your best unique value proposition included in the ad. I recommend using it as your second line, and using the top line as your testing line. A unique value proposition is the most compelling strength or offer you as a retailer have that sets you apart. If you offer free shipping, free returns, flat-rate shipping, no hassle returns, money back guarantees, have won awards, or anything else that builds trust or adds value - use it!

Ad Copy Line 1: Here’s where you microtest your ad. Many tests have found including numbers - whether a price, percentage or otherwise makes your ad stand out and improves click through. If you offer low prices, it may make your ad more attractive than others. If you want to deter bargain shoppers, a higher price can save you money on clicks and improve your conversion rate. You could here test price vs. non-price, but for this example, we’re going to test variations of showing a price.

Display URL: Some advertisers use dynamic keyword insertion at the end of the display URL (so long as it fits) to boost the keyword relevance, and it may or may not improve click through. An argument could be made that shorter URLs have higher click through. Also, www and non-www URLs may be tested. You really need to test this yourself, but not in round one. If you have too many variables in your testing, you won’t know whether it’s the ad text or the display URL that’s performing better.

In round one, we’re going to use www URLs for every ad variation, with capitalization, as there has been many studies that do suggest this gets higher click through. Since I want to play it safe here, we’re going to always use capitalization in every round.

Okay, here are the ads we’re starting off with - but there’s an error, can you spot it?

Answer: We’re testing pricing offers, so we want all else to be identical. We can test “name brands” in a later round.

Don’t forget to edit your campaign settings and select “show ads more evenly” so you can properly split-test.

Now we get to make up some results. Let’s say that “$29.99 and Up” had low click through rate of 0.41%, but the other two versions had 3.0% and 3.72%. And let’s assume a quick check with SplitTester.com’s confidence tool declared “From $29.99″ the statistical winner. Before you decide it’s a winner, make sure that it also has a good conversion rate - that these clicks end up buying. You don’t want to pick an ad only costs you money and doesn’t make you any.

If you have an ad that rocks click through rate AND conversion rate, you can move on to round 2.

Round 2

Pause the ads that didn’t “win,” and let’s ad new ads to test. If you delete your ads, you’ll never be able to see their history again.

Now I want to know the impact of URL versions as follows:

I’m going to pretend that the non-www, non-trailing keyword version performed best. On to round 3…

Round 3

I’m trying out more creative copy in the first line. I want to know what seems to be most compelling about wireless headphones. Is it range and clarity of sound? Is it better bass? Is it cool styles? Are name brands important?

OK, again, totally arbitrary - let’s assume the test resulted in “Stylish, Wicked Bass” as the winning creative. Can this be further tweaked?

Round 4

At this point (or in any round), there might not be a clear winner. You might want to keep testing different micro-variations until you get a definite star ad before you move onto the next round - testing headlines. Or, you might decide that they’re all performing decent and you can choose one you like best and move on to the next stage.

Round 5

Now I’m ready to test headlines. There are common modifiers PPC advertisers will use in headlines - the ones I’ve seen most are:

• unique

• discount
• cool
• cheap
• find
• buy
• compare
• on sale
• sale
• save on
• for less
• get
• need [keyword]?
• reviews
• [brand] [keyword]
• [keyword] [country]

You might want to try these, but remember that once you add the bargain hunter modifiers, you could attract clicks you don’t want (that’s why including price, even if higher than competitors helps). But definitely avoid testing different brand names. Unless they are doing specific branded search, you don’t want to use brand names in your headline. It makes your selection seem too restricted like you only offer one brand (less selection) or you have made a decision for them. Unless the searcher is familiar with the brand, it’s not effective. Your branded products should have their own Ad Groups, anyway.

If you’ve had a clear winning ad copy in previous rounds, you might want to inject these words into your headline, or try your unique value proposition in the headline, like “$0 Shipping” to stand out. Just make sure you’re not wasting ad copy by repeating your offer again in the lines of text. And use your dynamic keyword insertion headline as the “control” version - you want to see if headline variations outperform it.

Here’s an example test, can you spot the error?

If you noticed that one of the ads does not have “wireless headphones” in the ad text or headline, you’re correct. You don’t want to sacrifice keyword relevance for catchy headlines. Double check your ad text, it’s easy to forget the details.

Conclusion

Again, there are many ways you can approach your testing, this is just an example of how you could go about it. This strategy involves testing micro-changes. You can also test really different copy against each other. Tomorrow we’re going to show you how to reduce your risk when doing such tests.

Until tomorrow, here are the key points:

1. Keep testing. You can always “beat the control” - so long as you keep trying.
2. Don’t test more than one variable at a time.
3. Use your unique value propositions whenever you can.
4. Use prices when it makes sense, or other numbers.
5. Use SplitTester.com to check if your test has produced a winner or not.
6. Don’t make your decision based on click through rate alone. Make sure the ad converts.
7. Capitalize The First Letters of Your Words And URLs, It Is A Proven Strategy.
8. Avoid using brand names in headlines unless your ad group and landing page are targeted to only that brand.
9. Triple check your work for spelling and other errors. Make sure your landing pages are tied up correctly.
10. Keep reading Get Elastic for more tips for online marketing. Subscribe if you haven’t already!

Just a heads up for Google Analytics users, your data may not be accurate for April 30 - May 5, 2008:

“System Message: Analytics Processing Delay from April 30th to May 5th

Google Analytics experienced a data processing error from April 30th to May 5th. Almost all of the data has been recovered and is currently being reprocessed. The recovered data will be reflected in your reports within a few days. Please note that a small percentage of data, particularly in the area of e-commerce reporting, was not recoverable from those dates.

We sincerely apologize for this processing issue and are taking every precaution to prevent such disruptions from occurring again in the future. For more information, please read through our common questions.

The Google Analytics Team”

With PPC, unless you’ve tapped into niche long-tail keywords your search ads compete with 20 other links per search result page (organic and paid ads) - or even more if you count Google Maps or Google Shopping results.

And often the products and offers are so similar between retailers, what makes a shopper ultimately buy is which retailer he or she prefers.

That’s why you should include some statement that explains why someone should buy from you in your ad copy. You’re not just selling the product, you are selling yourself as a retailer. Most ads don’t do this at all, they just focus on the product.

What if Ads Were Ranked By Persuasion?

Let’s have a look at search ads for “coffee mugs.” If you search for this and refresh the page a few times, you’ll notice results bounce around quite a bit (we’ll explain why this is later). The left screenshot is how Google really ranked the ads when I performed the search, and the right side how I would rank the ads by effectiveness.

Now let’s deconstruct the ads:

Continue Reading:
PPC Advertising: Are You Selling Yourself In Your Ads? »

What?!! It’s true. Many of the biggest and most popular online retailers with fat SEO budgets are not showing up for their own names or valuable keywords in Google search results. Most don’t even know it. How can this be?

All the major search engines offer a .com search engine and a number of country-specific engines, like Google.ca, .co.uk, .com.au, .fr, .de and so on. These are local search engines, and often use geo-IP targeting to show the local search engine as the default when a searcher lives outside the United States.

If you’ve never seen a localized search engine, this is what it looks like:

As you can see, a searcher has the option to restrict search results to only pages from his or her country. This is particularly helpful for searchers who are performing transactional searches - they’re looking for products to buy. Using the general “search the web” will often deliver US sites which requires the shopper to dig through the sites looking for shipping information and costs. Searching only pages from their native land, searchers can save time and discover online stores they purchase from over and over again.

Continue Reading:
Top Online Retailers Not Showing Up in Google! »

While in London this week, Jason Billingsley spotted something different in Google search results.

If you look closely, you’ll notice the display URL in Adwords ads are above the ad copy, not below. Ad copy also appears on the same line as the display URL in some cases.

I did a quick search in Google Blogs (search within only blogs) to see if anyone has blogged about this yet or may know what’s going on. Over here in Vancouver, I still see the traditional ad display so all I can do is speculate that Google is shakin’ things up and testing the impact of different ad structure.

Jason also observed that the text appears larger than usual which indicates Google may be testing out new formatting of not only paid search but also organic.

Of course it could be handywork of an internal prankster at Google, or the equivalent to Jason’s computer trying to drive on the opposite side of the road while across the pond. (PS, if you’re wondering what the stars are, that’s a feature of StumbleUpon - you can see which pages have been reviewed and rated by Stumblers, and if one of your friends gave it a “thumb,” that will show up too.)

Anyone have more information on this?

How should you approach split-testing your checkout process?

This question was asked of Bryan Eisenberg in yesterday’s Google Website Optimizer webinar. Bryan recommended split-path testing, reducing the number of steps in your process and using your analytics data to determine what part of your checkout path needs attention.

What is Split-Path Testing?

The definition of a split-path test, according to GrokDotCom:

Split-Path Test — This test will split your traffic among different linear paths containing multiple pages for each path. This is different in that you’re testing the performance of grouped pages against other grouped pages. For example, you could test a checkout process by splitting it into two variations; one with four steps (or pages), and another with only three steps. Each variation of grouped pages will have the same Goal Page (e.g., order confirmation page). Once the data is collected, the winning checkout process will be the one that converted a higher percentage of visitors.

Reducing Checkout Steps

Different ecommerce stores have different checkout paths, ranging from one-page AJAX checkouts to 6 steps or more. Bryan believes less is more - in fact, he recommends going under 4 steps. But you can find out for yourself if this is so for your website by doing your own testing.

I’ve gathered some examples of checkout steps (many are generally the same aside from labeling) that can give you some ideas for how to simplify your process. For example, you may want to test a new path with a combined billing and shipping page vs. your existing separate steps. Or you may want to ditch a step that may be clogging your funnel, such as “Rewards Program.”

Continue Reading:
Checkout Process Split-Testing Tip from Bryan Eisenberg »

In case you missed it on TechCrunch the other day, Google is now showing search boxes within SERPs (Search Engine Results Pages) for some of the larger online retailers like Amazon, Zappos and Office Max.

The boxes only appear for certain keywords, for example “amazon” and “shop amazon” but not “amazon books.” Zappos shows up for “zappos shoes” but not “zappos shopping.” For other sites, adding “shop” or “shopping” to the site name won’t trigger a search box at all.

OfficeMax should be pleased that this works for them but not for Staples and Office Depot, at least it makes them seem a bit more important? I noticed that Target and Walmart get a search box, but not Sears. NewEgg, Radio Shack and BestBuy get one, but not Circuit City. Ebay and Overstock also are left out, which is a bit of a head-scratcher.

Continue Reading:
Search Within A Search - Good Idea? »

Okay, maybe this is more of an opportunity than a hack, but links from Page Rank 8 pages are hard to come by, and from Google itself nonetheless. But for Google Checkout merchants, you can get a link from the Google Checkout Merchants page, and no - Google doesn’t use nofollow or robots.txt to blog Page Rank from passing.

Now a general rule of thumb amongst SEOs when scoping out link opportunities is to find pages with less than 100 links. Obviously this page far more links than that. But don’t worry, here’s an explanation from Google’s Matt Cutts himself:

Matt’s exact words - The “keep the number of links to under 100″ is in the technical guideline section, not the quality guidelines section. That means we’re not going to remove a page if you have 101 or 102 links on the page. Think of this more as a rule of thumb. Originally, Google only indexed the first 100 kilobytes or so of web documents, so keeping the number of links under 100 was a good way to ensure that all those links would be seen by Google. These days I believe we index deeper within documents, so that’s less of an issue. But it is true that if users see 250 or 300 links on a page, that page is probably not as useful for them, so it’s a good idea to break a large list of links down (e.g. by category, topic, alphabetically, or chronologically) into multiple pages so that your links don’t overwhelm regular users.

Source: SEOmoz

I would imagine Google is going to crawl its own pages as far as they go. Yes, if you understand the Page Rank concept that “link juice” is divided among all the links on the page - the more links the less nectar you get. But Page Rank is exponential. 100 links on a Page Rank 8 page could pass more value than 50 links on a PR4. If I’m wrong here, please drop a comment.

Continue Reading:
Hack Week Part 3 - Scoring a PR8 Link From Google »

Google Analytics recently introduced an internal site search feature to its already kick-ass free stats program — aptly named “Site Search.”

This tool works with your existing site search and is invaluable to ecommerce marketers as it gives you so much insight into customer intent and your website’s success at delivering results. For example, you can use search log data to discover:

• What keywords people search for - what’s hot and what do they want that you don’t carry
• What search refinements are made, indicating possible “Results Not Found” messages or unsatisfactory results
• What pages the searches were made from, and where users clicked to

The next 30 days is when this information will be crucial. Customers can’t buy what they can’t find. Maybe you only use the term “notebook computer case” and your customers search for “laptop bags.” You can tweak your product pages and search engine for the various ways customers describe your product until the right pages show up when you test your site.

Continue Reading:
Why You Should Turn On Google Analytics Site Search Today »