Search Engine Marketing Results Tracking

So far, we covered setting up the keywords and establishing metrics. Now, setting up your tracking to optimize SEM…

With a half way decent IT group it's not difficult to append each click or destination URL with keyword parameters that can be picked up and tracked in your system.  Google and Yahoo (even MSN) have auto tagging options that you can set up to append your destination URL with keywords; while you can set up your own parameters to track campaign and ad group information. Since the engines use different parameter names in the auto tracking, this can add a level of complexity that IT groups would rather not see. So, you can opt to use your own parameter names and set them equal to the token or key that pulls in the Keyword.

For Yahoo! the token is {ovkey} and for Google it is {keyword}. So your parameter would look like this:

page.html?keyword={keyword}&campaign=[camp name]&adgroup=[adgroup name]&engine=[search engine name]

This will populate your URL with some of the more basic tracking information. How you structure your tracking information should be based on how you set up your campaign, adgroups and creative. By mirroring the logic behind your set up, you will be able to more easily make sense of your results, especially if there are any odd things (and there will be).

Depending on your IT set up, this may be handled by any number of methods with JavaScript, php, asp etc. and dropped into your database. Once this information is in your system, you'll want to have it appended to the target event, preferably the sale.  IT is your best friend here. Learn what they like, candy, cookies, their beverage of choice… you will want to thank them from the start because it will be an on going effort.

Okay, so you don't have an IT group, or they don't have the resources to code tracking and put up data tables and pull reporting (we’ll get to this). If you can get blocks of code set (pasted) onto your pages, Google has an attractive option. Their analytics tool can track program advertising on any site, search or otherwise.  
By properly appending your ad urls with their code, not only can you see the conversions, you get a suite of site analytics related to the advertising campaign.

Straight from the Google Analytics answer page:

Campaign Source (utm_source) Example: utm_source=google	Required. Use utm_source to identify a search engine, newsletter name, or other source.
Campaign Medium (utm_medium) Example: utm_medium=cpc	Required. Use utm_medium to identify a medium such as email or cost-per- click.
Campaign Term (utm_term) Example: utm_term=running+shoes	Used for paid search. Use utm_term to note the keywords for this ad.
Campaign Content (utm_content) Examples: utm_content=logolink or utm_content=textlink	Used for A/B testing and content-targeted ads. Use utm_content to differentiate ads or links that point to the same URL.
Campaign Name (utm_campaign) Example: utm_campaign=spring_sale	Used for keyword analysis. Use utm_campaign to identify a specific product promotion or strategic campaign.

To auto generate your own URLs, go here. However, you have to have a Google Adwords account to generate your page level codes. Though you do not need to place actual ads to use the analytics, you do have to sign up for the account.

Okay, so why not just use this in the first place? There are multiple reasons, some less valid than others depending your perspective. For those who are just too gun shy about letting Google know what is going on inside their site, this tool can not be useful. There are those who believe that Google will use the information in ways not necessarily in the advertiser’s best interest. While I do not buy into this, given how much Google wants to ultimately collect (information that is), I am hard pressed to be derisive of those who do.

My reason for not getting too deeply depended on Google is that it does have data access limitations. Doing a dump of all tracking for deeper analysis is not easily possible. Crossing ad spend with ad results is difficult, and for those slightly less proficient with spreadsheets and databases, it might be impossible. While I like Google’s site side analytics and standard report views, it is not really a good SEM optimization tool. I found creating our own is better.

Ultimately, once you have collected target actions associated with engine, campaign, keywords, you also have to pull in ad spend by keyword. All the metrics ideas discussed previously, should be measure in aggregate and in minutia. Detailed engine reporting is the only way to do this. It may be necessary to be so mundane as to just manually enter the information each day into spreadsheets. If you’re fortunate to have a decent SEM management solution like Omniture’s, you can have all the reporting set up automatically. From engine level margin performance down to individual keyword performance, you can be kept informed easily.

The only problem with the daily tracking and reporting processes is that they can lull you into a false sense of success. Keep all your metrics in line and your good, right? Wrong. Establishing and maintaining current metrics assumes you are already fully optimized. No one is. The target is always moving. There are always better ways to structure your keywords, deliver ad copy and marry expectations to experience. Nothing is 100 percent optimized. And even if you are the Einstein of SEM, and were aimed directly at the center of the target… it moved before you pulled the trigger. If the best you could do was a 22% margin when you put the plan together, something changed. There is an opportunity to make it 21.5% now. I know this is vague, and cliché and almost trite, but I have seen improvement come from too many angles to believe that we can not make it better. Set your metric, shoot for it, hit it, then make it harder and start all over.

A Final note on tracking: Set up your URLs NOW. Do not wait until your tracking is in place. At the very least, your information is being recorded on your server logs for later analysis. The longer you wait to append your URLs, the less information you will be able to retrieve.

Setting Search Engine Marketing Metrics - ROI

Metrics seem to be an obvious component of any SEM program. Unfortunately, it is too easy to set up quick metrics and call it a day. Some of the more common ones are:
-Click through rate (CTR)
-Cost per click (CPC)
-Impressions
-Page views
-Time spent

Some often talked about but seldom used, I mean truly used are:
-Conversions
-Cost per Sale
-Margin
-Units per Sale
-Revenue

In short, metrics tied to real ROI.

It's funny, but I have been in conversations with heads of marketing and advertising for some large companies.  They pushed us as an agency to prove the ROI of our program, which we were perfectly willing to do (this was off-line). We could set up unique tracking right into the store and online. They would simply have to provide sales related to the tracking. This is when they balked. Despite all the talk of ROI, most companies are not set up to truly track sales-to-efforts.

With search, we should be able to close the sales-to-effort gap. The preferred Methods are the direct methods. These include attaching source information to each visitor and tracking where they go and what they buy. 

While every company is about the sale, not all online activity is.  So the question to address is, "why are you doing search advertising?" Somewhere your activity online is expected create or assist in a sale. Are email newsletter subscribers your target? How about inquiries or phone calls (click to call on site, or track-able phone number)?  If you can only get as far us measuring visits, then nail down specific pages, or page views or time spent which tie as closely as possible to the sale.

Then know the percent who buy and the average amount of purchases or life time value.

You must also figure SEM's role in generating the sale.  Ask, "if we weren't doing SEM how many of these sale would have come any way?" And.. . "If we didn't have radio, direct mail, display or other Brand and promotional marketing, how many sales would SEM miss as a result?" Both are extremely difficult to answer.  But both should evoke a line of thinking that helps define the metrics that are as close as possible to real ROI. 

The ultimate goal here is to define your activities in terms of revenue generated or margin contributed rather than on some cost basis. Below are some scenarios based on direct sales and email newsletter subscriptions. These may reflect closely what your program is, or they may not. The point here is to let these help you figure a way to set metrics that truly show the value of your efforts.
 

The closer you can come to isolating your efforts and focus on your margin the better. But, we all know there are other factors, so you will avoid issues if your metrics account for these up front. If you have a good analytics group, they can help you out here. If not, build a consensus among other stakeholders in marketing, advertising and sales. Some methods are Unique sale - answering the question about how many sales does SEM generate that would not happen without it; Assigned Value which provides the amount, or value other media have on each sale, and the indirect where the action is not a sale, but will lead to sales.




  
Most likely, these will be estimates. But, if you continually monitor these assumptions, you can keep these numbers close to reality and show real value.

Beyond the ability to show the value of what you do, developing a margin based approach can open a different risk paradigm that allows you to find more opportunities. A while back I went off on the whole risk thing.

After you figure the metrics, the next step is setting up your tracking… more to come.

Google's Unversal Search

When Google announced the Universal Search this week, I was more than just a bit skeptical. From my perspective, they already made it easy for me to dive into images, or video, or news, etc. By placing all these items on the page which is currently web results, they have to guess (although with very sophisticated engineering) what I want and prioritize it accordingly. As a user, I am okay with clicking a category link for a deeper dive.

Then, I waffled. I saw some results of the universal search and it was not overwhelming. They have done a good job of keeping the results clean in this initial foray. Images are there on top, but not too many to be a nuisance. News appears at the bottom, with a few article links available. Not yet truly integrated by relevance. So, all in all, the results are not bad.

Then, I waffled some more. If I was truly interested in news, I don’t want one or two articles at the bottom of the results. And a couple of images, if that is what I wanted, placed at the top would not suffice. I realized that my initial reaction, that it was a good clean result, was based on the fact that I usually want the web results, and these added categories were not interfering in any major way.

However, if I want news or images, or videos or books, then this provided too little of it. I will click on the category link, making the results of the Universal search somewhat moot.  I understand that Google engineers are superb. However, when I type in Saturn, sometimes I am looking at the make of car, other times I am looking for pictures of the planet. Until I am actually searching, I don’t even know what I want.

Another chance to waffle… if Google can really evolve this tool (and it is just in its infancy), to provide results more heavily populated toward my intent, this will be great. But, until then, it is really cool engineering but not much more for the users.

The how behind the what in SEM - Optimizing paid search

Optimizing SEM for Google and Yahoo!, I mean really optimizing it, requires a wide range of tools. As I thought about some at my posts I realized that some of the tactics which seem basic may require core skill sets which are not so basic, or tools that are not used as commonly as I assumed.

So, over the next week or so, I will cover some of the tools we use, or have seen other use, and what they can do for your SEM. This is definitely not all- inclusive so any ideas you have please feel free to post. Also, one of the challenges I have is determining if these fall into the arena of basic or advanced SEM. Since both will be covered, I will try to keep each section to a pattern of starting basic and elevating to advanced. This will be interspersed with commentary about the value of escalating SEM sophistication.

We'll start with outlining and defining Categories of SEM strategies and tactics for which we employ tools. Then in following posts, will delve into the resources, tools and techniques to tackle them. Broadly, some of the areas I’ll hit on are:
   
-Keyword Selection
  Creation and evolution of keywords, Match types and their association to specific ads   and landing pages.
-Bid management
  The process of adjusting bids to optimize cost relative to a target event.
-Conversion -Tracking - Reporting 
  Setting up conversion metrics that are based on real margin / profits rather than click CPC and CTR.
-Copy Optimization
  Creating copy which improves conversion and front end CTR. 

Okay, there is one post that I did a ways back that I think will be good for starting the Keyword selection. If you’re not familiar with developing keyword lists, or just want a perspective on it, check it out. Then, also take a look at Google’s estimator tool. Based on your site, it can help identify keywords and provide estimates (without the proper match type , negative, etc, it can be far off on the estimates).

More to come...

added 5/23:Setting the metrics

Added 5/26: Tracking

Negative Keyword Tool...Google’s Expanded Broad Match Antidote?

If you have not seen it, keep an eye out for the Negative Keyword tool in your keyword tab on Google.  This handy addition will let you eliminate unwanted impressions and clicks that have crept into our programs with the expanded broad match.

While many of us can identify some of the big offenders to use for negative keywords, estimating the impact on impressions is kind of a pain. With the new tool, you can select specific keywords in your ad group, or the whole ad group. It will present you with a list of suggested negatives, along with an estimated reduction in impressions. You can then click to add them. This is great for streamlining your campaigns and tightening quality scores. However, before you jump on this, do some homework.

Take a look at your conversion data. At the very least, set a baseline for sales by keyword  "pre negative keyword tool implementation". Then, after the period of time you need to get a sufficient number of clicks / conversions, run the analysis again and compare it to the first. How is your conversion volume and ROI? Ideally, what you should have cut out are clicks that did not lead to sales and impressions that did not drive clicks. If, however, you find that your CTR, or post click conversion rate did not improve, or your order volume dropped, then the negative keyword tool may have filtered out good searches on a broad match of your keyword.

Ideally, your system captures the exact search phrase that triggered your ad. If you do this, then you can track it to the sale, and compare that information with the suggestions from the negative keyword tool to minimize unintended cuts is sales; if the suggested cut actually produced sales, obviously, don't add the negative (see below).  Yahoo! and MSN have this as part of their URL tagging also. If you do not have log file analytic tools that can help you pull the search query, you may be able to apply the Yahoo! and MSN information (to capture the exact search term) to the Google tool.

I mentioned above that if you can capture the literal search string, you should (obviously) not add the negative to your campaign. However, an experienced SEM should be calling this out as an opportunity to optimize the campaign. Take these search phrases and turn them into exact match keywords in their own adgroups. This is a great way to reduce costs and improve relevancy. The trouble is that most companies aren't set up to leverage this (it does require time to start and maintain). If you're one of them, then at just take the time to identify them and not to add as a negative.