Competitive Edge

Issue 16 – AdWords Arithmetic: 4=1, 3=1, 3=2

There are more AdWords rules of thumbs than most of us have fingers. But in the quest for the golden rule, I’ve come to realise that a healthy amount of scepticism can be a very good thing when looking at an AdWords account.

When you log into your account, you’re presented with an astonishing amount of data. Some of it is of critical importance, a lot of it is interesting, certain sections are irrelevant, much of it is meaningless, and a significant quantity is downright misleading.

One of my personal favourites in the latter category is that of the average position.

At a most basic level, what does average position mean?

According to Google, “An average position of ‘1.7’ means your ad usually appears in positions 1 or 2, and it may appear more often in higher positions than an ad with an estimated average position of ‘1.8.’”

Okay so far. Aside from that last little eyebrow raiser – “may”? But there’s more:

“Values may contain decimals because the Traffic Estimator displays estimates as averages-not whole numbers-based on dynamic keyword activity among advertisers. Also, average ad positions are not fixed; they may vary depending on various performance factors.”

Let’s consider a basic example. A company has a very low-volume account, with one campaign, set to display ads in the US and the UK only. In the UK, the ads are generally displayed around 1st to 2nd place, while in the US the ads are generally displayed around 9th to 10th position.

The average position, assuming that the number of impressions in the two countries is more or less the same, should be around 5.5. Which is utterly meaningless and totally irrelevant.

Let’s add a little more realism here. Let’s assume that some of the competitors are only displaying their ads between certain hours, some have a low budget stretched thinly across each day, and some are panicking and adjusting their bids on a daily basis. All of which may affect the company’s average position.

I know that this is a pitfall in the very concept of averaging, but the fact is that the information is presented as being accurate (one decimal point reinforces this) and useful when in fact it is neither.

And it gets a whole lot less accurate than that.

By now I’m sure you’ve noticed that searching for certain keywords will sometimes (but not always) display a number of ads above the natural search results.

This is problematic on a number of levels.

First of all, you, the paying advertiser, have no say in whether your ads will be displayed there.

Second, this information isn’t recorded anywhere.

Third, these ads are 100% certain to have different behavioural patterns to those that are displayed on the right hand side. A well run AdWords account is based on data analysis, so taking this information away dilutes the accuracy of any conclusions.

Fourth, aside from the fact that you don’t even know if or when your ad may be displayed above the search results, you’ll have no way of knowing how many other ads occupied this position. For any search there may be between zero and three ads above the search results.

Fifth, don’t even bother looking for how Google choose which ads and/or how many of them are displayed above the search results, you’ll only get a very vague answer for your troubles.

Sixth, even by checking your keywords manually you still won’t know whether your ads are displayed in this position or not. What you see will be different from other searches at other times of the day and in different countries.

I’m aware that by now a number of readers of this article may be wondering whether my annoyance at such an apparently trivial fact is symptomatic of other problems in my life. I can only assure you that this is not the case.

The reason this irritates me so much is that I spend many hours every week analysing various datasets from our clients’ accounts. And because too much of the AdWords system is built on hazy mathematics, it’s only possible to go so far with any assumptions drawn.

Consider this example.

In a hypothetical ad group, my ads are running around the fourth position mark, with an average position of 4.2. In an attempt to improve this, I write three new ads with well-chosen keywords and excellent copy.

As a result of my efforts, the ads move up to being around 3rd place, with an average position of 3.1.

But when I go to check on my new ads a week later, I see that the click through rate has fallen from 8% to 3%. It wouldn’t be unreasonable to assume that the new ads are not performing as well, and when I delete the new ads the CTR increases.

What I didn’t realise, however, was that the majority of my keywords are generating three ads above the search engine results. When I was in fourth position, I actually occupied the top spot in the ads displayed on the right hand side. When there are no ads above the search results, this is the same spot given to ads in first place.

By moving up to third place, I’ve lost this apparent top spot, and have effectively been relegated to the bottom slot of the ads above the search results, which I would hazard a guess generally performs far worse than the top spot on the right hand side.

And why do I, such a hard-data obsessed preacher of numbers, have to hazard a guess instead of being able to prove it? Because there’s no way of identifying where the ads were displayed. The numbers are rendered meaningless.

Following on from this, Google’s Position Preferences are instantly rendered between misleading and meaningless. In fact so much data analysis relies on the average position that this inaccuracy effectively spreads its data mangling across the whole account.

And let’s take it one step further. We know that Google have far more access to ad data than we do, which is quite legitimate as it’s their system, ultimately built for themselves.

It’s safe to assume that Google will know whether third position above the search results has lower performance than the top position on the right hand side. So wouldn’t they then reorder the positions, and place better performing ads in a different order? It may sound a little paranoid, but if they realised that when three ads are displayed above the search results, the best positions are in fourth, first, fifth, sixth, second and third, wouldn’t they then place the better performing ads accordingly?

After all, doing so would be in their interests, and why should the advertiser care? They still have their token (even if meaningless) average position to fall back on.

And what can advertisers do about this? Nothing. Nothing at all. But I’m a firm believer in knowledge being power, so be aware of the issue, and factor it into all of your decision making.

Be seen, be aware, be sold.

Competitive Edge


The Competitive Edge newsletter is a monthly in-depth look at the issues faced by independent software developers today.

If you’d like to comment on any of the information within the newsletter, please email Dave Collins directly atdave@softwarepromotions.com.

Common Sense Disclaimer: The advice, views, experiences, opinions and other information shared in the Competitive Edge newsletter are researched to the best of our ability. At the time of publication, we believe them to be correct.

SoftwarePromotions can not and will not guarantee that any of the information here is accurate, and/or will work for your company. In plain English, use this information at your own risk.

All information herein is offered “as is” and without warranty of any kind.

To the greatest extent allowed by law, SoftwarePromotions, nor its employees nor contributors, are not responsible for any loss, injury, or damage, direct or consequential, resulting from your choosing to use any of the information and/or advice presented here.

SoftwarePromotions
http://www.SoftwarePromotions.com