Friday, May 28, 2010

Remember, Google makes its money serving search results

Garrett Rogers made an interesting point today about how Google was making it harder for site owners to collect precise site statistics. He cited two recent moves:

  • New browser plugins that allow users to opt out of having their data collected by Google Analytics.
  • A new secure search site that prevents people from sniffing your search queries over open WiFi networks.

The latter apparently causes search referrals to your site to be inaccurately tabulated by analytics.

My own perspective on this is that analytics is the feedback loop that lets you know how you're doing. It can play a critical role in making the web a better place by allowing authors to get more in touch with their audience.

Why would Google lessen the accuracy of analytics? There are a number of reasons, but the key thing to keep in mind is that Google makes its money serving search results, not off of its free Analytics package. Therefore, it's likely to be more willing to compromise on this kind of non-core activity.

Wednesday, May 19, 2010

If you're doing Google Analytics, you need to be aware of the developer docs

If you're following any of my other streams, you'll know I'm at Google I/O this week. It's a fire hose of learning. One key discovery was the Analytics talk that I managed to attend at yesterday's bootcamp. There's a new async bit of code for the analytics tracker that pretty much everyone should use. One of the key issues of using analytics with your web site is that fetching the javascript code that powers it can slow down page loads. The async code pretty much solves that issue on very recent browsers and keeps things ok on slightly older browsers.

One issue I'll bring up with analytics. Google needs to stop iterating the basic code snippet. I understand why they're doing it, but now that they have a mechanism that seems to work well and be fairly future proof, they should stop. Every time you have to go through and redo the basics, it worries casual users and provides just one more thing to do for advanced users.

Here's a link to the analytics api site. It looks really good.

Thursday, May 13, 2010

Language, Culture, and Modified Broad Match Keywords


Several years ago, I heard Colin Camerer give a talk in which he asserted that language and word choice were what defined a culture. Well, if they don't define a culture, they certainly define a search-based market segment.

Google is currently running a beta test for something called Modified Broad Match keywords, and some of the non-profits that work with our students are eligible. The idea behind Modified Broad Match keywords is to give you more control over how Google matches searchers' terms to the the keywords you bid on. As the diagram above shows, Google's default Broad Match will be quite liberal in matching your terms to synonyms, mispellings, plurals, and what not. The problem is that people who use words very different from the ones you've chosen may not at all be the people you're looking to attract. By that very different word choice, they've identified themselves as not part of your segment.

On the other end of the scale, Google has traditionally provided you the option to bid only on exact terms alone or appearing as a phrase in the search term. While this eliminates the possibility of matching people who use very different terms, it also requires you to specify exactly every term variation.

Modified Broad Match is a middle ground. It matches misspellings, stemmings (i.e., different suffixes and prefixes), and abbreviations but not synonyms.

For the careful search marketer, this seems like it could be the right targeting tool to go after those carefully defined segments. I suspect it will take some skill though, and I will be interested to see if my students find it useful.

Wednesday, May 12, 2010

Google's Conversion Optimizer

I discovered this video in response to a student question about how Google AdWords could guarantee a conversion ratio. The question came in response to a somewhat cryptic statement in the Google AdWords certification materials that the student, James Freed, references in his post. The short answer is that Google can't and doesn't guarantee a conversion ratio. Now, let me explain a little.

The real issue in search advertising is that you pay up front for clicks (that's why search advertising is called pay-per-click), but you only get paid when the visitor completes some action on your site. On average, only a very small percentage of visitors you pay for ever get to this point, typically on the order of 1–5%. Put differently, you're paying for a hundred visitors to click in order to achieve one to five sales on your site.

Three things should be obvious at this point:

  • If you're not making enough off of each sale to cover the number of clicks you have to pay for to make the sale, your return on advertising investment is negative, making the effort not worthwhile.
  • The resulting amount of advertising you can afford is going to be different for each and every product you sell. It will also differ by company.
  • You should therefore prefer to manage your advertising by cost per customer acquisition for each product versus cost per click for each visitor.

Google's answer to these observations is conversion optimizer. Essentially, conversion optimizer allows you to specify a target maximum cost per customer acquisition. Google then analyzes a myriad of factors in each of your visitors determining which ones matter for conversions on your site. It next alters your keyword bids (i.e., the maximum amount you are willing to pay for each visitor) to maximize the likelihood that you will get visitors more likely to convert into customers while staying within your budget. The frequent result is that advertisers using this tool both increase conversions and lower the amount they pay per conversion.

A few things to note:

  • You have to be using Google AdWords' conversion tracking tool and be getting at least 15 conversions per month to even use conversion optimizer.
  • You still pay per click. Google has not moved to a pay per acquisition model.
  • You still set your daily budget based on how much you're willing to spend on clicks.
  • Google does not guarantee it will hit below your maximum cost per acquisition since conversion depends on factors beyond Google's control. However, it appears that this tool is an honest effort to help advertisers manage to this metric.

On a final note, conversion optimizer focuses purely on the customer acquisition end of things. Far more significant gains can often be achieved by revisiting your website design and determining whether factors there are inhibiting your visitors from becoming customers.

Tuesday, May 11, 2010

What I do here

I'm Bud Gibson, a college professor who has worked with over 60 organizations in developing Eastern Michigan University's search marketing program. I also piloted the first run of Google's AdWords in the Curriculum initiative. I'm an expert at training absolute beginners to become effective search advertisers.

The range of my training programs spans from advertising strategies to landing page optimization. I can also connect you with capable hands-on help.

I will be using this space to talk about things I discover as I develop my college's search marketing program and work with clients.