Posted August 15, 2006 9:47 pm by with 2 comments

Tweet about this on TwitterShare on LinkedInShare on Google+Share on FacebookBuffer this page

ClickMuse has an interesting new offering – Adwords Optimizer. The service claims to scan your AdWords data and makes recommendations on what ads to update. ClickMuse says that those who have used it find it cuts down their “tweaking” time by 75%.

So how does it work?

The AdWords Optimizer will send you an alert as soon as one of your Ad Groups is ready to be optimized. So now you know exactly when it is time to take action and keep boosting the CTR of your Ad Groups.

You will no longer have to go through all of your Ad Groups to manually check and calculate if one ad is outperforming another. All of this will be done automatically by the AdWords Optimzer saving you hours per day.

If you’ve tried it, leave your thoughts in the comments section or add a link to a review on your blog.

  • EGM

    OK Andy – I’ve been playing around with this tool for several weeks now. While it does exactly what it advertises, there is a pretty big potential weakness in its use. The program only looks at click-through rate (CTR) and completely ignores conversion rate (if you have conversion tracking set up, of course).

    Here is a real situation I encountered:

    “Winning Ad”
    CTR: 3.76%
    Conversion rate: 3.7%
    Cost per conversion: $74.30

    “Losing Ad”
    CTR: 1.31%
    Conversion Rate: 25.0%
    Cost per conversion: $9.74

    Digging into the numbers, what you find is that the “winning ad” only had one converter and the “losing ad” had 4 converters (this campaign has only been active for a short time and is extremely narrowly focused so the low absolute number of converters is to be expected).

    If I were to go with this program’s suggestions blindly, I’d turn off the ad that is generating more actual customers. The AdWords Optimizer only looks at clicks and that would seem to be its major limitation.

    I’d love to see them upgrade the tool to integrate conversion data, though I’m sure that it would be a trickier algorithm.

    That’s my two cents, since you asked. 🙂

  • Yikes, that’s not a good thing. Thanks for sharing!