Can Quality Score Be Gamed?

Like most complex games with many intelligent participants (such as financial markets, chess, and high-stakes poker), your first hunches about how to “get ahead of the pack” in the Google AdWords auction are likely to be trivial, clichéd, or just plain wrong.

This year, as ever, you’ll read plenty of articles focusing specifically on Quality Score and “what to do about it.” Conference sessions will teach you Quality Score “tips and tricks.”

Since this is about rank and CPCs, admittedly we’ll always be driven to crack the code in some way.

This dates all the way back to when Overture ran a pure PPC auction. Under those circumstances, would it make sense to write extremely restrictive, “filtering”-style ads to maximize the value of a click to your business, while garnering a lot of free impressions? Of course it would. High CTRs, in that instance, wouldn’t be desirable. Overture tried to address that problem with a laborious, cumbersome human editorial process. (Arggh.)

When AdWords finally introduced CTR into the ranking formula, it led to a great leap forward in relevance, and fewer opportunities to game the system. That being said, many of us enjoyed early-era tricks. Fun in a time machine set to 2002: come in guns blazing so you enjoy very high CTRs associated with high ad positions. Then, “lock in” that CTR history by doing this at a reasonable volume, then gradually walk your bids down, holding your position. That worked pretty well then. The system is much more sophisticated today.

People are still routinely coming up with “Why don’t I move my queen way over there now and put my opponent in check?” moves for AdWords. Those moves are usually neutralized by a more sophisticated algorithm. Google’s spokespeople – not always wanting to say much about the formula beyond the published overviews – have often felt compelled to dispel certain AdWords myths. The “don’t get your hopes up” points made by AdWords product developers in recent years have included:

• AdWords normalizes for match type. You’re not going to be penalized for using broad match or rewarded for using exact match.
• AdWords normalizes for ad position. Aim for the positions that make sense for your business. Lowering ad positions that by definition enjoy a worse CTR does not harm your keyword Quality Score or any account-wide factors.
• You can’t improve Quality Score with negative keywords. (Really?) Well, I’ve heard that said by a high-ranking Googler, and no, I don’t really believe it.
• While there is an account-wide Quality Score factor (enough bad history across the board can affect your whole account), one of Google’s top AdWords architects has denied that there is a specific factor at the ad group level. One badly chosen keyword in an ad group won’t “contaminate” others in an ad group.

Current industry consensus is that 50 percent to 75 percent of AdWords keyword Quality Score comes down to CTR, with personalization elements adding complexity. Since Quality Score is reputedly calculated on the fly for each query, the reporting you see in your account is not “the” number, but rather an average. (For Google’s ever-changing summary of how Quality Score is calculated, go here.)

“Other relevancy factors” round out the CTR factor. These may involve semantics; display and destination URL (Google can tinker with how much users, and Google, trust your company’s main identifying factor); and the vagaries of how many ads Google wishes to show on a page.

“Landing page experience” is another component of Quality Score. It’s probably exaggerated by third-party pundits today. Note the word “experience.” User experience is best measured by user behavior patterns, not solely based on some arbitrary formula about which keywords match which landing page elements, etc.

Further to the landing page question: recall that Google started out by banning a narrow range of user experience violations, such as pop-ups. Later, it extended the policy to a wide range of trust-eroding practices. It’s important to scrutinize both the letter and intent of Google’s Landing Page and Site Policies to understand if there is something you’re doing wrong. Google is trying to protect users from scams and bad user experiences, and it does so through a combination of automated and editorial means. Hobby horses such as landing page load times enter the mix at various times, sending some advertisers scrambling to overreact to those stats for all the wrong reasons. (Speeding up your site is always a good idea, but you have no idea how much AdWords Quality Score is penalizing you for having a slow one, if at all.)

There are too many moving parts to user experiences for Google to be effective in policing them with human and bot oversight (though a quick read of the guidelines implies that human oversight and manual scoring shouldn’t be ruled out as elements of Google’s practices). Rather, proxies for bad experiences may be used as Google’s models become better and better at confirming bad patterns statistically. (Do horrible bounce rates factor in? Well, they should, but then, why does Google let you keep spending so heavily on pages with horrible bounce rates? Assume nothing.) And it might be easiest for Google to do relatively little on this front unless real humans take a real dislike to your ads or business model. Now, as ever, Google does not like “thin” affiliate sites, click arbitrage, fake comparison sites, banned pharma products, and so on.

Make no mistake: “crappy” pages often have little difficulty being associated with keyword Quality Scores of 10. If the site is good enough to get the job done, and the whole campaign does a good job of matching up users with related commercial intent, then Google isn’t going to throw up roadblocks needlessly.

It’s not a good idea to obsess over Quality Score. Advertisers are doing a bad enough job settling on the correct metrics to manage campaigns to; testing ads methodically; understanding statistical significance; understanding campaign settings; unraveling attribution puzzles; and so on, that they’re likely to fail in attempts to test cause and effect in Quality Score engineering.

That being said, the system isn’t bulletproof. We can still prevail over competitors if we follow strong hunches about the vulnerabilities of Quality Score and the effects of the overall formula on rank and CPC. Consider the following tips:

• In addition to keyword-level calculations, Google may apply an account-wide calculation that impacts your ad positions and CPCs. Many advertisers get paranoid, therefore, about low Quality Score keywords, and race to pause them. But they may be overreacting. Low Quality Score keywords are probably diffuse or wrong in intent, and need to be addressed. But if they’re not high volume, they probably don’t hurt that much. What is more interesting is the opportunity you might have to lock in a higher overall account Quality Score by continuing to hammer hard at your high volume “10” keywords. Maybe you’re bidding a bit higher than you would like on some of these. But more impressions for 10-Quality Score keywords that are working OK for you economically can’t hurt your account Quality Score. You’re laying down all of this positive history. Maxing that as a proportion of your spend in the account may be a benefit from a Quality Score perspective.
• Google has kept “Display URL” in the mix as a kind of wild card in Quality Score calculations. I believe it is largely a “rich get richer” shortcut for Google to emphasize trust and quality in the results. Think broadly about your business strategy. Everything about the way you build your brand and conduct your business should be aligned with improving that brand’s reputation. “Throwaway domains,” generic microsites, and quick shifts in strategy won’t be aligned with Quality Score health, long term. This also means that Quality Score benefits recognizable brands and well-liked micro-brands. The Johnny-come-lately with a weak offering and a stop-and-start ad spend is not going to garner the same responses as a business with a consistent marketing strategy.
• Units like Sitelinks, designations like Trusted Stores, captivating visuals like Seller Ratings Extensions: various ad innovations of these types tend to goose up CTRs, all else being equal.
• In a Quality Score world, trying to finesse your spend by not buying your brand keywords is a counterproductive strategy. You need to own these and you need most variations of these to become 10s. Combine premium placement in the large ad unit with Sitelinks (for example), and your CTRs may go through the roof. As such, you’re laying down statistical evidence that you’re significantly more loved and trusted than Google’s models would predict (i.e., what your competitors can muster).
• If you’re managing only to CPA and failing to push your ad tests harder to find potential mutant ads that deliver great ROI along with “less bad” CTR, you’re not optimizing your account fully. Reconsider the advantages of more clicks on ads.
• Geo-target if that helps you push up response rates to ads and user engagement on-site.

With these ways to legitimately “trick” Quality Score, you’ll have enough on your plate. Meanwhile, many of your competitors will be frantically moving their chess pieces into clichéd or trivial positions, until they finally realize you’ve been thinking 10 moves ahead. Checkmate.