Quality Score Questions & Answers, Part III

We have a few questions left from last week’s Quality Score webinar with Bryan Eisenberg. The first batch of 10 Quality Score Questions/Answers are here, and the second batch of  10 more Quality Score Questions/Answers are here. Any more? Post ’em in the comments. Q: Does using the keyword in the title and description and URL produce a high Quality Score? A: Repeating a keyword in the text-ad and landing page and yes even the URL (not often possible 🙂 certainly does produce a part of the relevance that delivers a good Quality Score. But remember that CTR is the most important attribute, there are other components too (like load time etc.) and almost certainly the relevance calculation is somehow more complex. So nothing other than doing everything right (with everything being a set of things and measures we don’t know) will guarentee a high Quality Score Q: What is the most optimized landing page in terms of Quality Score? A: One that is perfectly relevant to the keyword, search query, and text in the ad, and gets people to stay (low bounce rate) and probably move forward (another click if not a conversion). Beyond that, nobody knows. Q: Can high impression low Quality Score keywords impact low impression keywords? A: Yes. Their are overall Quality Score for an Adgroup and for the Account components which impact everything. This is why improving or removing low CTR keywords and text-ads is important to overall and broad Quality Score. Q: How many impressions should an ad get before evaluating it’s CTR? A: It depends upon margin of error you’re willing to tolerate. A simple calculation for the margin-of-error in any sample size can be calculated as 1/SquareRoot-of-Sample-Size.  So, with 100 impressions your margin of error is 1/10 or +/-10%. For an ad with a CTR of 7%, this is clearly an unacceptable margin of error. For a brand keyword/ad where CTR is in the 40-50% range, this margin of error is probably acceptable. A good general guideline is that brand ads should probably see something on the order of 100-150 impressions and that general ads should see at least 200-300 impressions before making any meaningful decisions. More impressions yield even greater confidence levels.

aroubtsov

aroubtsov

The First Machine Learning Marketing Platform
Built to Scale Search for Local Resellers & Agencies

Automate, optimize and track more campaigns, more profitably.