Used to be that you could optimize your site for search, buy some keywords and reap the benefits of plentiful and profitable search traffic. Turns out those days are long gone. A couple months ago I wrote about how incredibly complex and nuanced SEM and SEO tactics have become, and how perhaps some of our time was better spent working on social media instead. You’ll be relieved to find out that I have since regained my senses and have decided to take a deeper look at some of the more advanced optimization tactics in search marketing, and how they apply to large, complex programs, like some of the ones we run here at Yahoo!.
For the purposes of this column, I’m going to assume that you’ve already covered the basics. That is, you’ve put some solid effort into your search marketing programs, to the point that your campaign performance has reached a plateau, and now you’re looking for ways to make the incremental improvements required to boost your returns. What you now need is next-level optimization.
Keep Your Eyes On The Prize
To take on an advanced optimization project, the first question to ask yourself is: what business metric are you trying to optimize to? Is it CPA, revenue, ROI, or profit? You’ll need to have a clear vision of this before you get started. Hint: It’s going to be the metric that best supports your company’s goals. It’s also likely to change over time, but let’s start with what’s valuable today.
Stand In The Place Where You Are
Once you know what you’re going after, you’re going to need data, and lots of it. Make sure you have a good web analytics platform or other source of rich conversion and revenue data. Next, I believe you’ll want to start your optimization with your paid search program, because it offers both the most data and the most moving parts to be optimized. As well, I think that the same discipline and technique used in optimizing paid search campaigns will play well once we expand our scope to other types of media.
Hit Your Targets
One of the things that has drastically changed about paid search over time is the explosion of different targeting options available to advertisers. Targeting is a double-edged sword, however, because the more you use these options, the more data you suddenly have to analyze. Not a problem with small or medium-sized campaigns, but with large-scale programs this quickly creates data overload for the search marketer. Before we talk about that, let’s take a quick look at the major targeting options and some of their best practices in action:
Match Types: For high-volume keywords, add (and track) multiple match types. Don’t forget to use your keyword as an exact negative against your broad match version of the keyword to ensure proper traffic segmentation.
Geo-Targets: Again, for high-volume words, use geo-targets to segment your traffic by locale—you’ll likely have local competitors in some markets, so you’ll need to address these markets differently than your nationwide campaigns.
Network Targeting: Build out separate campaigns for search and content targeting. If you still have time, also segment syndicated and non-syndicated traffic. They perform differently and thus should be optimized separately.
There are additional targeting tactics such as day- or week-parting and single-keyword ad groups and campaigns that can also help drive incremental goodness, but I’ll save these for another time.
Automation, that’s what. As I mentioned above, what you’ve likely done now is created a giant headache for yourself, the cause of which is the mountain of data you’ve just manufactured. What we face—and what most large advertisers face—is the need to build automation around paid search management. There are simply too many bits of data for humans to look at them all. But wait a minute: how do you automate optimization?
Look At Everything
If you’re really going after an automated optimized campaign, you’re going to have to go beyond the usual pieces of data search marketers look at—impressions, clicks, conversions, etc. The reason is that search marketers actually think about much more data than this. Accordingly, you’ll want to incorporate additional data like bidding history, quality score and competitive landscape data. If you’re going to rely on machines to make decisions, you need to inform them with types of data that human search marketers take into account when managing campaigns: when did I last bid (and how much), how is my quality score today (as apposed to last week) and what are my competitors doing right now? This means that you’ll need to build (or buy) the infrastructure necessary to capture all this data and aggregate it so it can be consumed, analyzed and made actionable by bidding and/or management algorithms that are tuned to optimize to your target business metrics.
One Step At A Time
While there are some very good tools on the market for paid search management, most of them fall well short of end-to-end optimization. What this means is that you’re going to have to take a close look at the trade-offs in a build vs. buy world. If you choose to build (or perhaps even if you don’t), you’re going to need to prioritize your efforts very carefully. That’s OK, let the (pile of) data point you in the right direction! In other words, you’ll want to find automated solutions for either the most onerous or costly portions of your portfolio. You may want to develop solutions to automate control of your head keywords because this is the most cost-effective use of technology, for example. Conversely, you may want to start with the tail because you find it impossible to manage manually. Whatever your chosen course, take a measured approach so you can apply your learnings as you go. Believe me, I’ve managed many different kinds of paid search and no two programs are quite alike, so you’ll definitely get smarter the more work you do and the further you venture down the path.
Next time we’ll try to step out of our paid search silo and begin to look at how to think about optimizing search in the context of other media. Until then, happy searching!
Opinions expressed in the article are those of the guest author and not necessarily Search Engine Land.