Thursday, August 28, 2008
With another major change to Google's Quality Score formula and reporting comes much uncertainty for advertisers. The changes: fixed minimum bids have been eliminated, and Quality Score is calculated in real time, at the time of the query.
I can come at this from a few angles, and probably will, in the coming months. Today, just one piece of the puzzle to address: efficiency for both the advertiser and Google.
First, some background assumptions:
1. New advertisers (and unevenly-engaged advertisers returning to refresh their memories) do keep pouring into the space, especially internationally. The optics of high minimum bids don't look good. They're alarming and off-putting to newbies.
2. Google likes its black box, and likes to avoid black-white distinctions. Building very flexible (read: confounding) architecture helps Google achieve a number of goals. And even those goals are subject to change.
3. Yet Google faces pressure for additional disclosure. So for every layer of complexity they build in, they try to offer up at least an equivalent step forward in terms of disclosure.
4. At Google AdWords, CTR is king. Clicks drive revenue, and continue to be a reasonable proxy for relevance. This is the biggest constant since 2002.
5. The platform as it stood at version 2.6 (my nomenclature), contained pockets of inefficiency. It did a good job of ramping up the "quality" bar, to the delight of users, but as even Sergey sheepishly admitted to investors, they might have "overtightened" the calibration of the platform, showing too few ads for advertisers', Google's, and investors' taste. The new release is intended to offer Google the ability to "untighten" selectively, without giving anyone the satisfaction of being able to point definitively as to exactly how that is being achieved.
The "pockets of inefficiency" buried in the fixed minimum bid regime were evidently ferreted out by smart Google engineers who realized that fixed minimum bids for keywords were too rigid. Rather than determining that a keyword in a given account should be "all on" or "all off" no matter what the context, Google has designed the new system to give keywords a fighting chance to show ads in all cases. (The official explanation is that no keyword is ever technically inactive for search.)
Quality Score is now determined in real time, per query. But wait. Don't think that means the only negative thing that happens to a Low-Quality-Score keyword is that it's relegated to a Very Low Ad Position. No, it can still be inactive at query time if it fails to meet what Google is calling a Bid Requirement. (Among other things, this gives Google an excuse to charge high prices for clicks in some instances, even if no other ads are showing up on the page.) What's different in this version is that the same keyword is eligible to be re-evaluated for the next query, and the one after that. So like the parrot in the sketch, it's not dead, just resting.
Let's be especially clear about this much: keyword quality (whose formula is outlined in Google help files, but clearly rests on measures of CTR history as well as predicted relevancy, especially for newer accounts with less data history) will determine both where your ad ranks for a given query, and whether it is eligible to show up or not at query time. So "fighting chance" and the "chance you might show up" even on a low quality keyword, some of the time, doesn't mean the same as "free for all." Advertisers aren't being encouraged to "go to town" with unrelated keyword experiments just to "see what sticks" -- in fact, that tactic is as bad as ever, because this can hurt account-wide quality.
Through all of the complexity of the explanations you'll read, then, and the potentially excessive focus on some tweaks in reporting (scale of 1 to 10 transparency for Quality Scores) and projections (First Page Bid estimates), the efficiency angle, and Google "shading" its quality initiative to make it more flexible and subtle, is the main story here.
And not a moment too soon, I'm guessing. By Google standards, the months of July and August were likely slower than they or investors would like. By notching up revenue in September, Google can turn in a respectable Q3. (Especially internationally, I'd expect to see click arbitrageurs given some respite, and getting a chance at more clicks - raising Google's revenue, but lowering overall search quality.) By the time things are off to the races for Q4, Google can always tighten things up slightly from a quality standpoint, and raise prices at the same time. If you asked, I'd predict steady-looking year-over-year revenue growth in Q4, with a bump in profit margins. Investors will cheer, and Google (GOOG) stock will head back to $700.
Labels: goog, google adwords, quality score
View Posts by Category