In parts one and two of this series, I gave readers some background on upscale jewelry designer Carrie D Mader and the challenges being faced by both her and our team of volunteer marketing consultants. In part three of the series we started digging into some of the obstacles that small business owners face and how our content strategy was starting to come together.

Today, I'll be focusing in on the start of keyword research and on the changes that we needed to make to Carrie's Google AdWords account. I'll also touch a bit on the ongoing difficulties of getting good ROI tracking in place when you're working with an out-of-the-box shopping cart system like MIVA.

The Exhaustive Process of Keyword Research

While it was exciting to see that Carrie and Kari were starting to get a good idea of how they wanted t move forward with the style of the content on Carrie's site, we still needed to work on a good solid keyword list to use with that content.

We got started by logging in to Trellian's Keyword Discovery tool. (The generous folks at Trellian had offered up a free year's subscription for Carrie as part of the project, though the tool is still affordable for most small business owners. It runs about $40 a month.) The plan was to start pretty broad and to work our way down to more specific terms as we worked on content.

Thus, we started by searching for variations of Carrie's three main product lines: necklaces, earrings and bracelets. The Keyword Discovery tool does a great job of putting together an estimate of the number of daily searches for a phrase and also works well to suggest phrases based on your initial search term. For example, when I plugged in the very broad term "necklace," the tool spit back a giant list that included the following:

The idea was to create a large working list of keywords that Carrie could go through to start giving us some focus. To that end, I pulled down a ton of keyword data and dumped it into an Excel spreadsheet that I then shot over to Carrie. Her job was to go through the list pulling out the phrases that were relevant. The relevant phrases were then sub-divided into "relevant now" and "could be relevant." In other words, we would be building up a list of keywords that we could target based on what she already sells, but we'd also be building up a list of potential items that she could offer in the future based on what people were searching for now.

Once Carrie is able to get the new lists together, we'd be in an excellent position to sit down with Kari to go over integrating them into the content and to talk with Stoney about integrating them into the tags and the link structure.

Overhauling the Google AdWords Account

Our next step was to sit down and work our way through some major, yet relatively simple changes to her AdWords account. Carrie had setup the account back in the days where your bid price and your click thru rate were the only factors that played into your ranking and your cost per click. That meant that all of her keywords were gathered together in one advertising group and that she was running a single version of her ad copy and using a single landing page for all of those keywords.

Since Google now considers things like the relevancy of your ad text to your keyword and the relevancy of the page that visitors will land on, we needed to break up her one existing campaign into several smaller, more targeted campaigns. Thanks to Google Talk, we were able to work through this process with a little bit of real-time conversation to guide us.

The three areas of her campaign that we planned to tackle were:

1.) The ad groups: As I mentioned earlier, Carrie had all of her keywords placed into a single ad group. Although she was only running ads on keywords that were related to weddings, she was failing to take advantage of the subtle differences in her keyword groups. By segmenting out the her keywords a little further, we'd gain the ability to more closely target the ad text to the keywords.

2.) The ad text: Carrie was currently running a single version of her ad text on her campaign. This is a common mistake among small business owners, or advertisers that are new to the Google system and it's one that can easily be remedied. I explained to Carrie that we could setup multiple versions of the ad and that Google would randomly rotate their display. This would allow us to test different text to see which was the most effective.

3.) The landing pages: Carrie's ads were all pointing to the home page of her site, another fairly common mistake among new pay per click advertisers. While the ideal solution is to craft actual landing pages for your pay per click campaigns, the budget solution it so simply make sure that you're pointing your ads to the most relevant page that currently exists on your site.

Tackling the Ad Groups

Our first job was to go through the list of keywords that Carrie was currently advertising on and to figure out the best way to segment them out by topic. We debated the idea of breaking them down by topics like necklaces, bracelets or earrings, but the reality was that most of the phrases she was currently bidding on were more broad than that, most containing the word "jewelry" instead of the actual type of pieces. Since we weren't yet ready to expand the campaign to encompass even more keywords, we decided to divide the keywords into four categories: "bridal," "wedding," "bridesmaid" and "custom." This would allow us to focus on one specific type of product, while still letting Carrie learn to target different groups of keywords.

With this plan in place, Carrie went through her campaign and divvied up the phrases among these four categories. She then created new Ad Groups for each of the four categories and placed the appropriate keywords into each group. That left us ready for our next step...

Tackling the Ad Text

A lot of Google advertisers have no idea that they can run multiple ads for each keyword group and Carrie was no exception. The basic idea of running multiple ads is to test different messages against each other. This works because Google will randomly show each ad an equal number of times and will spit back data on the click-thru rates. Add in a tracking code and you can even use your web analytics program (we're using ClickTracks) to find out which ad results in the most sales.

The benefit of this is that you can test minor variations and then adjust your ads and even your page copy based on those variations. For example, in Carrie's "bridal" ad group, we setup three different sets of ads. We selected three ad titles and then three different messages that would run underneath them. Her ads look like this:

Basically, we're testing both ad title and ad descriptions at the same time. You'll note that we've used the same three descriptions with each of the ad titles. That will let us test the impact of both the title and the description without getting too deep into a million variations. In fact, the ONLY difference in the ad descriptions right now is the second adjective in the first line. Thus, we're testing the words "gorgeous," "artisan" and "designer" in the title and the phrase "high quality," "hand crafted" and "romantic" in the ad description.

We'll run the ads for about two weeks, then will look at the click through rates and the conversion rates for each of them. Based on what we see, we'll likely keep the highest performing set of ads, drop the other two sets and add in another round to start testing. Over time, we'll want to make sure that our most successful ads are running the most frequently, but that we continue to test new variations just to see what happens. (Google will automatically increase the run time of the most clicked ad, but you can "force" them to run your highest ROI ad by entering a duplicate ad into the system more than once.)

Now, one problem that we ran into on setting up these unique ads was our ability to track the conversion rates of each of them. As I was explaining the concept of testing the different ads to Carrie, I made mention of the fact that Google would show us which ad had the highest click-thru rates. Before I could move on, she quickly jumped in to ask if we could use ClickTracks to find out which ads had the highest conversion rates, since she knew that customers were going to be more valuable to us than clicks. It was one of those moments that consultants love...the moment where you know for sure that the client "gets it."

I explained to Carrie that we could, in fact, track that data but that we'd need to add tracking parameters to the end of her landing page URLs. This is easy enough to do, you pretty much make up a code for each ad and append it to the end of the URL that you enter in the "destination" box. The problem was that whenever we would add a parameter to the end of the URL for Carrie's domain, the page would break. (Note: we could add a parameter to her home page, but not to any of her store pages) Apparently, MIVA runs into some type of formatting issue and tries to read the ad parameter as a store parameter. That meant that we'd need to hold off on adding the parameters until we spoke with her developer about correcting the problem.

A setback yes, but these are the potential pitfalls of an out of the box shopping cart. We decided to leave the ads, remove the parameters and to plan on adding them back in as soon as we fixed the URL issue.

Setting Up the Landing Pages

Our final task was to make sure that each ad was sending the shopper to the best landing page. For now, we directed the wedding, bridal and bridesmaid related phrases to the bridal jewelry section of her web site. We directed the "custom" related phrases to the page that talks about her custom jewelry design services.

Ideally, Carrie will eventually setup landing pages for each campaign that match the tone and content of the keywords and the ads themselves. For example, she might create two versions of her bridal jewelry page. One version might use the word "wedding" and the other version might use the word "bridal." Each version would serve as the landing page for shoppers that entered the corresponding search term. This should not only help lower her per click costs on Google (due to the quality score evaluation of landing pages) but should also help to increase her conversion rates by matching the words that were used in the original search.

Wrapping Things Up

After several hours of working in Google's AdWords interface we'd managed to completely overhaul her existing campaigns, to set her up so that should experience lower costs per click and higher conversions and to start to put some ROI tracking methods in place. We still had more work that needed to be done in AdWords and we'd yet to tackle Carrie's Yahoo Search Marketing ads, but Carrie had learned quite a bit about how she could make her pay per click campaigns perform better.

She'd also shown that she was really learning how this search marketing game works, which meant that she'd be able to carry on when the project had ended as opposed to being forced into hiring someone to continue to manage it for her. Since that's what this series is all about, we wrapped up the day feeling pretty confident in the progress that we'd made.

Discuss this article in the Small Business Ideas forum.

August 31, 2006

Jennifer Laycock is the Editor of Search Engine Guide, the Social Media Faculty Chair for MarketMotive and offers small business social media strategy & consulting. Jennifer enjoys the challenge of finding unique and creative ways to connect with consumers without spending a fortune in marketing dollars. Though she now prefers to work with small businesses, Jennifer’s clients have included companies like Verizon, American Greetings and Highlights for Children.

Search Engine Guide > Jennifer Laycock > Real Small Business, Real Big Ideas - The Carrie D. Mader Project Part 4