Traffic estimator inaccuracy (or deception?)

[arikgub]

Google traffic estimator inaccuracy has been mentioned many times here. I just want to put your attention to the following.

While I was playing with the "estimator" I decided to check its forcast for the actual keyword that is running in one of my campaigns. The actual keyword's performance is: 50-80 clicks/day at average CPC of $0.04 and max CPC set to $0.05. Nevertheless, for max CPC of $0.05 the estimator reported me the forcast of 0.4 clicks/day. I don't know how the hell the estimator algorithm works if not by taking the actual statistics from my (and others) account?

In order to get my 50 clicks/day (that I'm in fact getting at $0.04 per click) I had to increase the CPC in estimator to >$1.00! I tried another keyword with similar results. Well, I am not a big fan of conspiracy theories I read here from time to time, but it makes me think that the estimator results are biased to make us pay higher CPC. I mean, I can believe the estimator is inaccurate but at least in my case this inacuracy is consistently biased in one direction - pay higher CPC to get the same number of clicks.

Or may be for you guys it is different?

[dbhatta]

It's bias in both ways. I get estimates for 20 clicks and get none, it says I'll have position 7 but I end up with 58 all the time.

Although it seems there is a conspiracy going on, the estimator does not discrimiate. It shows crappy result regardless, equally crappy both ways.

Deepak

[toddb]

I thought the estimator just assumed a 1% CTR. So all of the underlying math would cause a higher CPC.

[inferno]

i ignore it, i dont think it has ever been corrent, not even once for me.

[shorebreak]

If you are an AdWords advertiser and are relying on G's Traffic Estimator, you are never going to find the New World. If you use your own traffic data, then unless you have serious modelling jockeys inhouse you will never know your optimal operating point for search - you may *think* you do, but IMO you don't.

If you're working with an SEM firm one of the first questions should be "How do you model impressions/clicks/cost/revenues/margin on AdWords? If they can't provide a good answer odds are they have no clue how to manage spend on G.

-Shorebreak

[arikgub]

toddb,

If the estimator assumed strictly 1% CTR, we could easily figure out, based on the clicks projection, the number of impressions (searches) per month. But this is exactly the information everybody is looking for and G is hiding.

This tool, however, is not merely useless, it is misleading. Better not to use it at all.

arikgub

[grigoroo]

I think the estimator uses a random number generator to determine position. The last time I was relatively open to its results I was told that a minimal bid would get me 1st position and tons of impressions. I ended up about ten pages in with few impressions.