Operating successful paid search campaigns is a complex, ever-evolving and often maddening endeavor. The only constant is change, whether it's the algorithms that power Google's Adwords and Bing's Adcenter, competitors coming in and out of the auction marketplace, shifting goals, etc.
Despite the complexities of paid search, there are 10 best practices we employ at Business.com. These have been stalwarts for years, even if their implementation requires periodic refinement. Here are the first 5:
1. Goal Setting
This seems relatively intuitive, but I'm continually amazed at how many paid search marketers are unclear on their primary (and perhaps secondary or tertiary) objectives. Before launching any SEM campaign, it's absolutely critical to know (1) the purpose of your campaign (e.g. traffic volume, pageviews, site engagement, registrations, conversions), (2) the metric(s) that will be used to gauge success (e.g. revenue, ROI, gross profit, CPA, CPM, cost per registration) and (3) the numerical goal associated with your performance metric(s).
Do not keep your objectives and goals to yourself. Nothing ruins a paid search campaign faster than a disconnect between the paid search team's understanding of their mission and management's. Before spending dollar one on paid search, it's imperative that the objective is clearly outlined, goals are established, and everyone is on board (Tweet This!).
2. Performance Measurement
While goal setting is vital, it's a meaningless exercise if you can't accurately measure your performance against defined targets. Think of a stock portfolio -- you can make adjustments based upon stock analysts' recommendations, market trends, your intuition, etc., but if you can't measure the results and how they compare against the broader market, you are investing blindly. Similarly, launching paid search campaigns without reporting capabilities is a fool's mission. Importantly, like a stock portfolio, which is comprised of individual securities, paid search campaigns are a collection of individual keywords. Just as you would manage your stock portfolio by tracking the performance of each security within it, paid search requires visibility at the individual keyword level. Without that granular level of reporting, you will never be able to fully optimize your campaigns.
Managing paid search is not a passive, set and forget, discipline. It require extremely close oversight, with quick, easy and highly reliable assessment capabilities. The platform used to manage your campaigns can be highly sophisticated (e.g. licensed software such as Marin, Kenshoo, Adobe), an in-house proprietary reporting system, or a series of manual data pulls that marry internal performance data with data exported from the search engines' reporting interfaces in an Excel spreadsheet. Bottom line, do not set goals that you cannot accurately and frequently report on.
3. A-B-T
Always-Be-Testing. The death knoll of any paid search campaign is dormancy. Simply put, if you aren't testing, you're losing (Tweet This!). The paid search landscape never ceases to evolve, and neither do your competitors' tactics. Fear of change when things are going well almost always spells long-term doom. So, what exactly should you be perpetually testing?
- Landing page design: All elements of a landing page should be included in your testing queue, from page look & feel and your call to action, to the articulation of your value proposition. The same holds true for subsequent pages for multi-page conversion flows. However, unless you are using a multivariate testing platform, it's critical that tests be launched sequentially, as opposed to simultaneously. Otherwise, it will be unclear which element drove the performance. It's also important to perform A/B tests against the current design, to normalize for seasonality and other external factors. Finally, resist the urge to be trigger happy. Patience is a virtue when it comes to any test, so make sure to allow your tests to reach statistical significance before declaring it a winner or loser. I cannot count the number of times a test looked like a failure out of the gate, only to ultimately prove successful once enough data was captured (or vice versa). Here is a calculator you can use to measure statistical significance: http://getdatadriven.com/ab-significance-test.
- Ad copy: An SEM best practice is to have at least 3 ads running within each ad group, testing subject lines, ad description copy and the display URL. For example, one very common test is the use of static subject lines vs. dynamic keyword insertion ("DKI"). Be sure to set your ads to rotate evenly (as opposed to allowing Google or Bing to select the best performing ads for you), because their determination of which ad is performing best is based purely on ad CTR, and they have been known to pick winners before tests have reached statistical significance
- Average Position: By adjusting bids, test and keep track of how your keywords perform at different positions. Depending upon your goals (revenue vs. gross profit vs. ROI), certain keywords will perform better at higher avg. positions, while others will perform worse.
- Google.com Search vs. Google's Search and Display Networks: Your ads are eligible to be featured within Google.com's paid search results, or through AdSense on non-Google properties via Google's Search and Display networks. Generally speaking, ads that appear within a search experience (on Google.com or throughout their Search Network) have higher CTRs and higher conversion rates, because they are user initiated instead of impulse based...but also carry higher average CPCs (ie better performance drives higher bids). Many advertisers elect to place their paid search ads exclusively on Google.com, but that leaves an enormous number of potential impressions and clicks on the sideline. Don't be afraid to test Google's network traffic, and by all means make sure you are managing the keywords within the Networks distinctly from traffic you generate from Google.com.
- Match type: While there are differing schools of thought on this, I highly recommend launching all 3 match types (broad, phrase and exact) for every keyword. Due to the matching algorithms, the three instances of a keyword are likely to perform very differently, and therefore require distinct bids. At Business.com, we employ this bid construct for new keywords: phrase match bids and broad match bids are set at 85% and 75% of exact match bids, respectively.
4. Don't Rely on Google Exclusively
It may surprise you how many companies use Adwords to drive traffic but ignore other high quality traffic sources such as Business.com, Bing/Yahoo. Sure, Google remains the 800lb gorilla, but vertical and general search engines still garners over 30% of the search market share (Tweet This!). That means if you are exclusively using Google, for every 1,000 clicks your site receives, you could be capturing an additional 500 clicks simply by mimicking your Adwords account across other search engines. Plus, CPCs tend to be lower than in Google, so many advertisers report a better ROI on Business.com than they do with Google.
5. Staffing
Whether you employ a third party SEM company to serve as a reporting and/or optimization platform, in-house management and oversight is absolutely mandatory. And this cannot fall on a marketing manager/director who has a cursory understanding of paid search or who doesn't have the bandwidth to pay it proper consideration. Paid search requires attention- constant attention.
So, what type of employee is best suited for this role? Having hired over 30 paid search professionals, I will only hire those with advanced qualitative and quantitative skills. No exceptions. Sure, SEM is part art, part science, but it's primarily the latter. I look for candidates with a mathematics, economics or engineering degree, and those who come from banking, insurance, consulting and finance industries. And, never will I consider someone who doesn't possess advanced excel skills. At Business.com, all SEM candidates are required to complete an Excel-based case study, and take a rigorous qualitative and quantitative aptitude test (we require a score of 90th percentile or above to proceed with the interview process). Intellectual horsepower cannot be overvalued when it comes to managing paid search.