Jul 302013

AdWords Brand CPCs on the Rise!

Several of our analysts at RKG noticed that brand CPCs on Google have been on a pretty steep upward trend. Individually they looked to see if there were affiliates or competitors bidding up their brand keywords, or if the brand terms were being served on non-brand queries. Finding no big changes there, we recognized that this wasn't isolated to a few clients, it was widespread. That prompted us to dig in.

Let's look at data aggregated across a basket of long standing RKG clients.

Over the last 12 months, Brand CPCs on Google have increased 80%.

Looking at YOY comparisons to take out seasonal effects shows an even more dramatic picture of what's happening.

Google's Explanation

We raised these concerns with our Google rep and got a very credible explanation. Let's start this explanation with more data.

If we back up our time window we see a much more complete picture of what's happening. The increase over the last 12 months is just reversing a slide that had happened for the 18 months before then.

Again, looking at YOY changes makes the two trends quite clear.

Brand CPCs are therefore, in Google's view, getting back to where they belong. From about February of 2011 to December of 2012 brand CPCs were falling, and it is only since then that they've started to rebound to previous levels.

So what drove these trends? Interestingly RKG's very own Mark Ballard touched on the answer in last summer's 2012-Q2 Digital Marketing Report: All those extensions, site links, bigger site links, seller ratings, etc. had unintended consequences for Google. By increasing the real estate of brand ads, click through rates increased therefore Quality Score increased therefore CPCs declined.

CTR has always been far and away the largest component of QS. We've said it, and Fred Vallaeys confirmed that in a fine post a couple weeks back. But anticipated CTR has to be measured carefully because of dependencies. Google wants to reward the ads that consumers like best, but page layout can distort its view.

Positional dependencies are the most obvious. Suppose you have ten ads to show on a page and you don't know which ones users will like best. You order them based on bid as a starting point with the highest bid at the top. Lo and behold: you find that the ad you put at the top has a much higher click through rate than the one below it, the second ad has a higher CTR than the third on down the line. You seem to have guessed the ranking right the first time! Amazing! Amazing but wrong. The ads at the top have a huge advantage of prominence over the ads at the bottom on the right rail that has nothing to do with whether users prefer one ad over the other. To really determine which ads get the best user response Google has to normalize CTR by position on the page to take prominence out of the equation.

Google has always normalized CTR for position so that they can determine the optimal arrangement of ads to maximize their revenue per impression -- ahem, er, to maximize utility to the user and advertiser.

However, as they started creating extensions, sitelinks, seller ratings etc which applied only to the ad in top position they did NOT normalize the CTR for the enhanced prominence given to those ads. Those ads got a significant bump in QS and therefore a big reduction in CPC. It isn't clear whether this was an oversight, or part of the original plan which they've decided to reverse. It is clear that they've reversed course on this and that it has a clear impact on brand CPCs.

Brand ads are the most obvious and easiest to research. It's possible that this has impacted other ads that are virtually always at the top of the page and regularly qualify for additional features.

As we've always said, averages and aggregates lie. There are clients for which brand CPCs have been low for a very long time and only now have increased significantly. Others see much less dramatic swings. The key question is: what are you seeing?


21 Responses to "AdWords Brand CPCs on the Rise!"
Jon says:
Could First Page Bid have a factor? We've suspected for some time that Google is trying to Raise our branded CPC's by putting a "Floor" bid on showing on the top of page 1. At times, we'll have no active competitive bidding on our brand kw's, yet Google will require a $0.14 bid in order for us to show on our own branded term! and low and behold, if we set our max CPC to say $.08, no ads show on page 1 and ours will have been bumped to page 2, again, with no competition.
Jon says:
to clarify, this used to not be the case where we used to be able to get away with a $.05 bid. We've been seeing this first page bid "floor" have more and more influence on our branded kw's.
Jon, excellent observation. The minimum promoted bid threshold, or first page minimum threshold could absolutely explain the increase. Google has a number of ways to dial up revenue and those are prime examples. I don't think that's what we're seeing because it wouldn't explain why CPCs on brand terms dropped precipitously for most folks prior to 1/1/13. The normalization explanation gives us a single solution to the mystery, which Occam's Razor would suggest is right because it is the simpler of the competing theories. If the brand CPCs exceed previous levels, then we need to look to those thresholds as the better explanation.
Jon says:
I do agree the explanation in this post definitely has a big impact leading to these trends, I was just curious if your team have done any studies on minimum thresholds as yet another factor. Thanks for the response and great blog post!
Thanks for commenting, Jon. We were the first folks to "out" the Google ghost bidder a few years back. It is very difficult to study it and understand its effects separate and distinct from other competitors. We'll noodle around on it though, and if you have ideas on how we could tease out that influence let us know!
John says:
Great article as always George! This explanation makes sense. I also wonder, however, if sketchy affiliates are to blame as well? We've recently noticed a number affiliates (including some of the big ones) are bidding on out client's brand name and using their Display URL (an issue that RKG has written about before). It's odd because on our end it seems as if the practice went away for a while, only to come back with a vengeance the past few months (which would coincide with the recent trends). At any rate, if this practice has become more widespread then I could see it driving up branded CPCs (as the brand would need to increase their bids to have their ad show over the affiliate). Then again, one could argue that it isn't widespread enough to affect the big picture. Still, I could see it as a secondary factor. Thoughts?
John, thanks for the kind words. You know me: I'm always willing to consider the nefarious behavior of affiliates as an explanation for annoying trends. :-) That was our first thought but we couldn't find evidence of affiliates changing behavior. I wouldn't discount the possibility of that being a driver for some folks, but the fact that our data reflects exactly what Google's explanation suggests (they made a change last December/January to start normalizing, and that's exactly when the YOY trends go from negative to positive) makes me think this is a real answer for many if not all.
Ian O says:
Every week, you guys put out the best PPC analysis and content. This post is another example in a long line of excellent posts. I wonder: what would happen if competitors in a specific category got together and decided to take a "bid holiday" and lower bids say 20%? Has anyone tried this? This would apply best to unbranded term(s), so this might not be the *best* post on which to pose this question. It also feels like it *might* be some form of collusion, but in an auction-based, relative ad quality based system, where Google can somewhat arbitrarily set prices, it doesn't feel entirely wrong to consider. Is it against that ToS? I just wonder how much LESS we could all be paying per click on behalf of our clients if we cooperated a bit. Thoughts?
Wijnand says:
Hi George, very interesting read! We see similar trends over here (in Europe) for our clients. I was wondering: do you also see branded CPC's rise when the advertiser is the only one advertising on its brand name? We've also seen CPC's rise in those cases, which makes little sense in theory. So that would mean Google simply raises the bar to be in top position?
Thanks Ian. The notion of a bid holiday or collusion is one that I explored last year. Ultimately, I don't think it would work out because if you think of the top 10 bidders as the ones getting all the traffic, then lowering bids hands that traffic to the next 10 in order and they'll be delighted to take it. The incentive for #10 to cheat and not lower bids so that they get top position at an affordable bid would also be quite keen. I don't think there is a law against a buyer boycott, I just think it would be impossible to coordinate and enforce. That said, I totally understand the sentiment!
Hi Wijnand! Thanks for sharing. Yes, we see this regardless of whether there are other ads showing on the brand term. As you suggest, there is a threshold of AdRank that must be met to show on the first page, and a higher threshold to appear above the natural listings. The notion would be that as the QS improved the CPC needed to exceed that threshold lowered, so advertiser's paid less. As the QS sinks back to where it was pre-extended ad formats the CPC increases. Now, as the earlier commentator, Jon, pointed out: it could also be that Google is tinkering with those threshold values particularly as they're struggling to meet Wall Street expectations, but I find it hard to believe Google would have lowered those values systematically and then changed course. Google's explanation makes more sense to me.
Peter says:
Ahh, that you don't *see* competitors, doesn't mean they aren't bidding on the terms...Remarketing Lists for Search, for example, could be a driver in increased CPCs. It wouldn't be immediately visible and would require Auction Insights etc, but I suspect that it's at least in some way connected.
Peter, thanks for your excellent comment; it sparks all kinds of thoughts which might have to be follow up posts. Your point on RLSAs is a great one! It is always dangerous to generalize as what is true in eCommerce apparel may not be true in B to B tech services. In our experience in retail and travel bidding on competitors' brands is rarely a huge win; might be cost effective but usually very little volume. People who typed in "Walmart" aren't interested in going to "Kmart", that's why they typed in Walmart. That said, in a world of RLSAs Kmart might only show ads on Walmart searches to people they know are also Kmart customers which would make their ads more relevant and attractive to those users. Kmart ads in that context are a much more competitive alternative to Walmart's brand ad. That could absolutely drive up Walmart's brand cpcs. If the Kmart ads aren't visible though it may be that while they're bidding on Walmart's brand, they aren't bidding enough to get over the threshold AdRank. When that's the case it's not Kmart's bid that impacts Walmart's cpc, it's the threshold itself that acts as the next highest competitor. But, to your point what you "see" when you search doesn't actually tell you what's happening when someone else searches. Your point is well made that reality is not well represented by either averages or individual observations. That point is worth exploring further. Thanks for stopping by and for the thought provoking comment.
Craig says:
Might be nitpicking, but it seems somewhat unlikely that big brands had only average quality scores to begin with, and somehow through 2012 they doubled their quality score (and halving their CPC). Isn't it more likely that the quality score of the competitors dropped due to all of the ad enhancements that favored the actual brand? It seems more likely that the collective auction score of the competitors dropped by 50% than the quality score of the actual brand doubled. Regardless, very insightful...thanks for the continued thought provoking posts.
Craig, interesting point and you might be right. We see this drop and then increase even in instances with no other ads showing suggesting competition against a fixed threshold. However, now you've got me thinking about relative QS and what is shown versus what is calculated. Very interesting, thanks for your excellent comment.
Craig, chewing on this more, I think we're both saying the same thing if we understand that QS is just an aggregation of factors: primarily CTR, a touch of 'relevancy', a touch of landing page relevance/quality. CPC is determined by relative ad rank. Competitor A's Bid *(CTR * weighting factors for relevancy and landing page) vs Competitor B's Bid * (CTR * weights for relevancy and landing page). Doubling A's CTR doubles it's AdRank and therefore cuts its CPC in half assuming that nothing changes for B. What shows in the interface for B would in fact be lowered QS for those competitor A brand phrases in B's account. Taking away that CTR boost for A therefore has the opposite effect for B, and all of a sudden the price to compete on a competitor's brand just came back to earth. Turn those affiliate monitoring systems back on!
We're seeing this clearly in Northern Europe as well. With one client I tested keeping branded keyword bids static for August and September. Auction insights show an impression share of 95,55 % in August, but only 82,48 % in September (no change in QS, no competitors bidding, no budget caps). Somebody wants higher bids!
Thanks for the update, Toni!


Check out what others are saying...
[...] AdWords Brand CPCs on the Rise!, RKG Blog [...]
[...] again. After noticing a trend in a couple accounts, RKG’s search team has found that overall, Adwords Brand CPCs are On the Rise. In their article, George Michie and Rachel Schnorr give great background data as well as insights [...]
[...] few months back we noted that CPCs on Google for brand terms had been rising at very high rates and offered that the cause was Google starting to normalize QS for the impact that ad extensions [...]

Leave A Comment