Some Tuesday Morning Math: Average CPC vs Ad Position

1. Some Tuesday Morning Math: Average CPC vs Ad Position
So I'm trying to build a metric that represents the rise in Average Cost per Click in our Google campaigns, and marries that to our average ad position (we shoot to be top 3 of course, for important key phrases).

Two challenges:

I'm not sure I know exactly what that number would look like, and in the end, what it would mean, so I'm struggling getting off the ground in building a formula.

Anyone else tried to draw a mathematical correlation between these two statistics?

2. Ugh. Where are my math majors and former engineers this morning. I'm going to pop a blood vessel.

3. I'm not quite sure what you're trying to compute. I think you're trying to show that the CPC for a certain position for a specific group of keywords has increased over time?

The problem is that you don't get enough exact information to make fair estimations at any level other than "day and keyword," and even then comparisons might be unfair if you've added negative keywords or changed match types (or changed geographic or language targeting, or experimented with day-parting). Any of these "optimization" strategies would tend to reduce CPC and increase ROI/ROAS, while also increasing CTR and position and reducing volume.

I think that for your goal, the best strategy would be to choose a few specific keywords (probably your highest-traffic keywords), and run keyword performance reports for those ad groups, and instead of generating a "summary," generate a "daily" or "weekly" or "monthly" report, so you get the progression of your bids over time.

Added: I just tried to run a historic report for a specific keyword for one client, but discovered that "all time" was only a few months for that particular ad group -- the keyword was previously in another ad group, probably with a different ad, and perhaps different match-type choices.

Added: And now that I've run a larger report, another issue pops up: over time, the client has added many more-specific variations of their main keyword (for example, in addition to bidding on their primary term "widget," they've added ad groups for "[adjective] widget" (e.g. "blue widget" and "large widget") and "[price-adjective] widget" (e.g. "discount widget" and "cheap widget"). This has drawn traffic away from the primary keyword, thus changing the performance independent of bid rate.

4. Originally Posted by Kevin
Ugh. Where are my math majors and former engineers this morning. I'm going to pop a blood vessel.
This study might help it pop. Richard Stokes is into that kind of analysis.

5. Thanks for the input, Mark. I believe you are right in it being perhaps a "bad way" of looking at success, or lack thereof.

Perhaps even moreso in our case. Here's an example:

Product Group A: We spend around \$2.50 per click to acheive the Top 3 Placement.

Product Group B: We spend roughly \$1.40 per click to get those results.

What I was looking at was 3 years worth of this data, and trying to show that our AVG CPC was not directly correlated to position. Other players have entered the marketplace, and our terms are VERY broad, with multiple contextual meanings.

On top of that, we only deal with CUSTOM variations of these broad terms. So the competition for ranking spots is not only from way outside our industry, but also from retail outlets that sell off the shelf versions of these goods.

We also know that being in the top 3 dramatically increases CTR, which, it's been stated, lowers your CPC for those ads.

So I guess what I'm trying to mathematically represent is that even though our AVG CPC is up 60% or so over 3 years, it doesn't directly marry into position. (We've gone from an average ad spot of 3.2 to 2.8 in that time.)

6. Ah, thanks for that link, John. I'll eyeball that PDF....

7. In an intelligently-managed account, I think there might simply be too many variables to consider, to be able to create a workable formula from "real" data.

I just tried to do a quick analysis using one client's quarterly data (Q4 2006 to Q4 2008), but I quickly discovered that even when I looked only at a single keyword with a single match type, I could not get data that presented any reasonable result. In fact, I couldn't even see a correlation between bid amount and position (because position was impacted by so many other "optimization" factors which were altered over the same time).

If I "sit back and reflect," I'm not certain that there has been an increase in the bid amount for long-time keywords that I've seen, but that's probably because I manage for ROI/ROAS and not for position. I certainly have seen average position decline over time, when the bid remained reasonably constant -- but since my bid rates are set for specific ROI/ROAS goals, I don't "chase position."

8. Do you buy into the theory that there's a dramatic drop off in CTR if you fall to the sidebar as opposed to above the SERPS (Top 3)?

Adwords/Analytics certainly reports that there is.

And the other intangible I can't account for is branding. Being in the top 3 with a well written, well branded ad serves me further purpose, since we deal in custom solutions. It's a B2B long cycle sale, and customers tend to do their homework.

9. I haven't done a recent study on the position break-off -- a long time ago, I found that the largest drop-off was from position #1 to position #2 (I think I came up with a drop in the range of 40% to 60%, depending on keyword), and the CTR didn't decline nearly as fast with the move from position #2 to #3, or #3 to #4.

There is a "somewhat inverse" relationship for ROAS -- if the ad appears in position #1, ROAS is lower than if it's in position #2, but subsequent position drops don't improve ROAS by very much. As I often explain, what I concluded was that measured by ROAS, for a particular keyword one of my clients could bid 42 cents for position #2 or worse, but only 32 cents for position #1.

Again, I don't care about position -- I only bid for ROI/ROAS, not position. Bidding for position is an ego game. If the goal is for the company to make money from sales, then you bid up to the rate that provides the appropriate ROAS, and stop there. If that puts you in position #5, that's fine; if that puts you in position #2, that's great.

Of course, this doesn't mean that you should just accept being in position #4 if there is something you can do about it. If you can find ways to increase ROAS (for example, by increasing conversion rate or average order size) then you'll be able to bid higher, which should improve position. Strategies for this include optimizing the landing pages to increase quality score, and of course streamlining and optimizing the merchant site to improve conversions and transactions sizes.