Google’s Bid Simulator Tool: So Transparent, It’s Deflationary?

Over the years, we’ve counseled advertisers to be wary of certain inflationary tendencies that are built into the very terminology and interface design of paid search. Experienced bidders know better but new advertisers can get tripped up.  Inflationary tendencies in paid search auctions include:

• New accounts being opted into content targeting at the same bid as search
• Bid estimators available at the setup phase
• The temptation to ego-bid as opposed to testing and optimizing
• Impatience in waiting out foolish competitors. (You want the house, but do you want it badly enough to pay $700,000 now, when down the street you’ll get the same one next week for $550,000?)
• The new “first page bid” notation in AdWords for lower-position keywords that suggests how high you should bid to stay on the first page of SERP’s (this could be set conservatively or aggressively. You need to take it case by case).

As market psychology shifts, these mechanisms have been less successful in holding up click prices. Advertisers are, in the end, free to bid lower. And, as we’ve seen recently with economic woes, many advertisers have gone into penny-pinching mode.

How can the bid simulator tool help?

It’s at that precise moment that the question arises: but just how low can I bid on a keyword while maintaining a decent ad position? I don’t really know what others are bidding, after all. Google’s new bid simulator tool is designed to answer this question. The tool obviously doesn’t show other advertisers’ bids, but provides relevant information that effectively answers the question indirectly.

Similarly, if you’re an advertiser seeking growth, and want to know how much extra it’s going to cost to make a significant jump in ad position, the tool has data in graphical form to help you answer that question. If it costs you $5.02 on average to stay in ad position 4, and you want to know what kind of price tag you’re looking at for position 2, the estimator will try to tell you (indirectly). It shows you lists of projected impressions at various price points, and you already know what your current ad position is. So, if you see that a significant increase in impressions will occur above $7.00, that may look like a decent tradeoff. If you see that a breakout in impressions and clicks (indicative of a significantly higher ad position) only occurs at $16.00, you’ll likely say “thank you very much, but I think I’ll stay where I am.”

The projections are based on the last 7 days’ worth of data for that keyword (on my wish list: longer time frames so low volume keywords can be eligible, or the ability to group keywords for projection purposes). Of course, past performance doesn’t predict future results, but by showing factually what would have occurred with a different bid, the prediction is likely to be in the ballpark. Here’s a screen shot from Google’s examples. It shows nothing dramatic, but you can see that jumping from the $2.10 per click range up to $6.36 or even $5.26 would be a pretty steep price to pay to gain a small amount of extra volume. Literally, it’s a steep pricing curve.

Real world examples: how pliable is the auction?

The next example, taken from a client’s account with keyword data disguised, shows the subtle tradeoffs involved in seeking the right bids and positions in a relatively competitive, high-cost auction. By spending 41.2%  more per click, we would have received 12.9% more impressions. But because those impressions would have been out of higher ad positions, the projected clicks rise more dramatically—by 44.4%. There’s no question that nearly any bid increase will lead to more costly CPA’s (lower ROI), but for those seeking increases in volume, the simulation can be helpful in gaining an understanding of just how pliable a particular auction is. And this information is richer than average ad position, in the sense that an average ad position of 1.9 could still be showing you in 4th and 5th spot more often than you realize, so your lower bid is costing you more volume than you might think.

Next, here’s a lower volume example from a less competitive auction. Not knowing what type of drop in volume we’d experience if we went with lower bids, we always kept to a relatively high bid, just to stay safe. The simulator tool tells us there is little buying support well below our current level, so economizing by dropping bids significantly is a good bet. In fact, at $2.21, the current bid looks kind of silly. The fact that the yellow dot, denoting our bid, is vertically on a y-axis above the next green dot in the plot on the graph, basically says we’ll see no volume decrease and no change in position until we drop to $1.54, and there, the drop is only slight. “Dude, just lower your bid to $1.54,” the tool seems to say. Thanks, dude.

Caveat: with low volume, we get an impressions prediction only, not clicks. So clicks could decline more dramatically than expected if you make big changes suddenly.

Finally, there are all too many examples of auctions that show significant opportunities to double and triple click volume, but the bid ranges to achieve that seem to stretch up and up and up, to ever-growing levels of insanity. From our current bid of $3, next stop $6. For a bit more traffic, next stop $8. Next stop after that, $13, and finally, you could really max out clicks by going over $19. All else being equal, that means someone’s probably chasing bids of $15 or more in an auction where we figure a click is worth around $2.50 with major pain starting at $4.00. Yikes! What are they doing? The bid simulator saves a lot of time here too. Experimenting over time to see how much money you’ll need to bid would be a big waste of time. Stand pat, or go up or down slightly, is the solution in many of these cases. There’s no point in chasing rainbows.

If you want to change your bid instantly while viewing the graph, you have the option of clicking on the presets and changing your bid, making either subtle or dramatic changes in positioning strategy. Or, you can enter your own bid if you don’t like what the presets say. I recommend often entering your own bid, because I like gradual bid moves. In general, projector tools don’t do a good job helping with real-world “walking up” and “walking down” methods that many of us have cautiously used for years (we make smaller bid changes over time, for various reasons).

Reasonably accurate projections—geared to the specific auction conditions and real time competitive bidding scenario in each keyword competition—are a great advantage over previous guessing methods. Before, it would simply take more time to gauge the impact of bid changes. You’d need to make the change and monitor the effect closely. With the bid simulator tool, you may be able to save time if you’re aware that there is little “buying or selling resistance” below or above your bid. In specific cases, advertisers can get away with making more dramatic changes.

Simulator beware

Back when I had more time, I used to play with a video flight simulator. Sometimes the simulator indicated I had successfully landed the plane. Even if I didn’t crash, under no circumstances would it have been wise for me to think to myself “say, I can fly a plane!” I’m no pilot.

Looking at a simulation is just that. Changing bids is, in the end, a serious business. Think about the information, and act intelligently on it. Don’t “believe” everything you see.

Is Google shooting itself in the foot?

Industry-watchers never tire of citing the flaws in AdWords. But anytime you go over the wish lists we’ve put forward over the years, you can easily point to the many feature requests where Google not only obliged, but exceeded the asked-for functionality. The bid simulator is a pretty radical step, in that it takes an apparently unfixable shift in the auction methodology (from a straight auction under GoTo, with all bids public) to a proprietary algorithm that includes bids as well as relevancy factors under a sealed-auction scenario, and appears to do a decent job of providing usable competitor information without revealing anyone’s private info.

It’s worth noting, on that front, that this feature first surfaced in a similar simulation offering released by Yahoo Search Marketing with its Panama release. Credit Yahoo with that first attempt at auction transparency.

By being so transparent in a recession, though, isn’t Google practically begging us to fix inefficient bids? If we see limited buying resistance below us, won’t we accelerate our bid decreases? Is Google trying to ruin its Q4?

Google would likely not persist with any feature or auction mechanism that was particularly biased against Google’s profitability in the medium to long term. In fast-moving retail, especially as the recovery unfolds in 2010, many advertisers will probably use the bid simulator to figure out the best way to raise their bids, not lower them.

So like most observers, I won’t worry too much about Google. In the end, only a small percentage of advertisers will actively use the simulator. And as Google diversifies its revenue stream, they know better than you and I whether their next few quarters of growth are looking smooth. Growth in average CPC’s may not be in the cards near term, but overall revenue growth is likely to elevate gradually from “flat” to “steady.” And that is of course from an incredibly high base that’s currently north of $20 billion per annum.