Google, with its promise to organize the world’s information and make it universally accessible, speaks the language of transparency – yet it doesn’t always practice what it preaches. But I’m betting there will be continued pressure on it to do so.
In a WikiLeak-y world where citizen revolts are fueled by Twitter hashtags, there’s only so much you can paper over with a clever wrapper. When it comes to your product, people want to know: does it work? How does it work?
As Don Tapscott has noted, “If you’re going to be naked, you’d better be buff.” Google plugs in well in this regard. It’s often had better fundamentals than its competitors, so it has less to hide. In the advertising program, Google generally offers more segmentation, access to detailed reporting, and opt-outs than its competitors.
So it’s a little disappointing when Google holds back on the release of some new features and reports. It’s also become fond of inventing new framing mechanisms in reporting; taken too far, the experience for advertisers can become Orwellian. Experienced advertisers may experience revulsion when faced with interface innovations that offer, for example, automated suggestions for cramming more irrelevant keywords into an account. “Lost Impression Share” seems clever, but the implied remedy for this is merely a higher bid or broader keyword match types.
Sophisticated advertisers would be appreciative if Google made a number of changes to the AdWords program in 2011. Here’s a sampling of seven key ones:
- Quality scores in the Display Network. Google happily outlines the rough parameters of how Quality Score is calculated on the search and display sides of your account. But it only displays the actual scores for your search keywords. We know our display ads have Quality Scores…so where are they?
- Session-based search ads. Now that Google has begun arbitrarily tinkering with a feature that re-shows your ad to users even when they’ve gone onto searches you’re not willing to pay to show up on, where’s the opt-out? Sure, in the aggregate, this form of retargeting might actually convert OK. But we’d rather be the judges of that.
- Easier-to-use negative matching. Keyword exclusions come with match types. The problem is, when you’re poring over the “See all search terms” report to find out if any inappropriate matches are costing you money, Google makes it by far and away more convenient to use only the exact match type for negative keywords. When you spy a problem keyword in a phrase, though, you’d usually want a broader match type for the negative keyword. This basically makes it harder to use that report in an actionable way, since in most cases the exact phrases of the poor matches are unlikely to recur very often.
- How does remarketing work? AdWords remarketing is great, but it’s a black box. We can specify the audience – such as “visited site in past 30 days without purchasing” – with precision. Then, the ads are deployed by selecting and bidding on that audience in an eligible ad group that has the Display Network turned on. But how is relevancy a part of the equation that determines who sees your ad how often? In general, how does the bloomin’ thing work? We’re in the dark.
- Separate bidding on the search partners. Unfortunately, we’re still asking for this one. You can opt out of non-Google search partner traffic. Problem: you don’t want to opt out. Search traffic converting at a higher CPA is still good traffic as advertising goes. You just want the option of bidding it down by 20 percent, let’s say, in part to punish the purveyors of lower-converting traffic, like you’d do anywhere else in your account.
- Keyword tool monkeyshines. Back in April 2010, Google announced that it would soon be discontinuing the “Search-based keyword tool” in favor of the consolidated keyword tool. The consolidation makes sense. Google’s keyword research functionality has moved around quite a bit, and you would rather things stayed put. This has unfortunately come with a loss of significant information. Google admitted:
“It’s important to note that due to these changes, we’re removing some columns that you had previously been able to add to your Keyword Tool results: Estimated Avg. CPC, Estimated CTR, Estimated Ad Position, Estimated Impressions, Estimated Clicks, and Estimated Cost. While you won’t be able to access these columns in the updated version, you’ll still be able to see much of this information in the context of an ad group.”
In a transparent world, phrases like “much of this information” and “in the context of an ad group” sound mealy-mouthed, though the latter actually speaks to the likelihood of more accurate group-wide projections. But in the same post, Google is also still recommending you use the Traffic Estimator tool. In most of my dealings with that tool, it’s told me I’ll get next to no traffic unless I bid ridiculously high. After all these years playing with keyword tools, I can say with some certainty that the best keyword research of all is actually building, launching, testing, running, and iterating a live account.
- Review approaches to statistics like “first page bid.” “First page bid” is neither an accurate nor helpful figure in many cases. It’s tied to how the current Quality Score works. Keywords with low Quality Scores require a high bid to gain any impressions on the first page of search results; “first page bid” purports to estimate that amount. But the number feels more like an ideological bludgeon, and what we’ve often referred to as an “inflationary” mechanism, than true information. Put it together with other forms of rhetoric in the AdWords interface, and you’ve created two classes of advertisers: those who take the communications at face value, and those who’ve adopted the appropriate decoder ring.
At some point along the line, it’s easy to cross that imperceptible line between “providing appropriate information to stakeholders” to “waging a concerted campaign of disinformation.” Hopefully, we’re keepin’ ’em honest.
In a future column I’ll look on the bright side, at amazing nuggets of information that are made available.