Accuracy of traffic estimator

[georgiek50]

To the more experienced of users, a question. When you are trying out new campaigns and you enter a bunch of keywords you believe would do well but Adwords Traffic Estimator recommends 0.5 max CPC and predicts < 1.0 clicks per day, does that deter you from using these phrases or is this the general idea to testing and finding new profitable phrases?

There are plenty of KW's which I would want to bid on but don't really have the budget for, and then there are a bunch which seem great pricewise (through some Overture KW Suggestion Tool research) yet Google predicts no traffic.

[DavidDeprice]

Google traffic estimator can not be accurate by definition. Unless they hire a psychic, they can't predict what CTR you'll have 0.5 or 11% (I have campaigns where CTRs are exactly that). The difference is twentyfold.

[grigoroo]

I wouldn't use the keyword estimator to make any important business decisions. I have learned that it is really only accurate in telling me which keywords in a group are most likely to generate impressions. Run your ads for a week and see what you get. Set your bids and budget for what makes sense for your business model.

[onlineleben]

Regarding trusting the Estimator

Just today I found the following:
When raising the max bid for a keyword from 5 cent to 8 cent, I expected a better position estimated. But the estimate was the other way round. From Position 4.5 the estimator thinks I will get 5.0 in the future.
My thinking was that adrank is a product of max bid and CTR, so a higher max bid (by keeping the CTR) should result in a better rank.
Anyone having a clue why this is?
Is it just a program bug or id I get the math wrong?

[gamb]

all you can do is test it to get traffic numbers, then you might be able to estimate. I usually divide their estimate in half.

[AdWordsAdvisor]

a higher max bid (by keeping the CTR) should result in a better rank.

onlineleben, you are correct. Typically, raising one's Max CPC will result in an improved position (other factors staying the same, which may or may not be the case in the quickly changing competitive landscape).

The Traffic Estimator is doing a really complex job, taking tons of (constantly changing) factors into account, and giving you a snapshot that is accurate for the circumstances in play in the moment you checked. So you can get different results from one moment to the next - and, yes, even results that are counter-intuitive.

All that said, I'm fully undestand that this tool does not have the too many supporters on this Forum, I'd be interested in hearing everyone's comments about what the ideal Traffic Estimator would do (and not do). We're highly interested in making it more useful.

I'm mainly interested in serious suggestions, as opposed to flames though. ;) I've heard most of those already, and I want real solid ideas to pass on the folks behind this tool. I actually have their ear right now, so this is a perfect time for some constructive feedback.

Thanks in advance.

AWA

[jbgilbert]

AdWordAdvisor

All that said, I'm fully undestand that this tool does not have the too many supporters on this Forum, I'd be interested in hearing everyone's comments about what the ideal Traffic Estimator would do (and not do). We're highly interested in making it more useful.

This is not a flame... the estimator tool is just horribly inaccurate. The best possible thing you could do with the tool is make it at least "kinda" accurate.

I had it tell me (many times) that a keyword at 0.05 will get 2nd position, but in reality, in the actual campaign it gets 27th position... If you want examples, I can make them available.

[grigoroo]

The traffic estimator needs to provide a reliable indication of how a given bid will generate impressions, and how many clicks might be expected given the resulting positioning, which it's supposed to be doing now. Additionally, it would be invaluable if it could report in easily presented summary form how an increased bid would improve CPC, or not. Accuracy should be within about 25% for metrics such as impressions, CPC and position. I accept that CTR is somewhat unpredictable. Perhaps a range can be given for the 80th percentile.

If you can't do it all, then just improve the accuracy of what it does now.

[tenerifejim]

The problem with the estimator is that while google can take stats such as keyword impressions, bid-price, etc. but it cannot measure how 'good' you advert text is at getting click throughs.

As for position, this should be accurate, but over time if your CTR drops then so will your advert, again, the success of ad text is the reason for this, so Google cannot count on this into the equation either.

[kona]

How about letting us specify our expected CTR?

This is the only unknown as far as the estimator is concerned, so for a given CTR it should be able to estimate CPC, clicks, and everything else very accurately. We can probably make a much better guess of our expected CTR than the estimator based on our past performance (or we can try a range of CTRs to get a range of the other values).

[mark1111]

Someone in another thread (here or another forum) suggested having it predict impressions rather than clicks, because this could be more accurate.

It appears that the estimator works on the same technology as the "Avg. Pos." column in the ad groups, because I've found those to be inaccurate as well, though not as bad as the estimator.

I assume this is also tied in with the concept of predicting how well keywords will do and having some of them go on hold or in trial as soon as they're put into an ad group, without any impressions. The logic of this is a little shaky, as others have pointed out, because a lot of it depends on the ad and the targeting.

I have a feeling the traffic estimator is sort of a security blanket for new users. From what I've seen, most experienced users don't pay much attention to it--it's just another step we have to go through to get our keywords into the ad group. I'd rather see Google implement power posting by spreadsheet and ad replication with one click and other ways to expedite setting up campaigns than spend their time on a more accurate traffic estimator--which will never be all that accurate.

[good2go]

It is rare that I prefer a feature on Overture to Adwords, however, I believe Yahoo Search Marketing / Overture provides actual number of searches on keywords the previous month. I am then left to my own experience to predict CTR, Clicks, Costs.

This would be a welcome addition to Adwords.

[AdWordsAdvisor]

Hi all,

Amazingly (to me anyway) I am just now getting to WebmasterWorld to post for the first time today, as part of my prep for this weeks Advertiser Feedback Report.

I just wanted to say thanks for all the feedback in this thread, which I'll be passing on in the next few hours. :)

AWA

[StupidScript]

I've never seen the estimator work properly when modifying terms or adding terms to a group. I just ignore it and try to track actual performance.

What I mean is:

- I go to the group display.
- It indicates current position avg at, say 2.3
- I click "Add Keywords"
- I add a keyword and click "Estimate Traffic"
- Existing terms show current avg position of 1.0 (or some other incorrect figure)

I figure at least the estimator could pull the "real" current avg position from the same db and show that, but it never does.

Plus, I see in G search results that my ad is "really" in position 17. Now maybe there's something in the averaging process, but that's a mighty big disparity that I can't account for by using normal math.

Any thoughts? Does anyone else see similar disparities? With that kind of performance, it's not a useful tool for me.

[gamb]

Your overture rep will give you predictions for certain terms based on ranking #1, #2 and #3.

Adwords is all over the map and is really tough to gauge....not all their fault though, I have some campaigns with a 3% ctr, and others where I'm happy with .8%......tough for G to estimate that.