Dynamic word insertion..
February 4th, 2008 adminAd text is one of those things that some people involved in search get extremely focused on. You can either write compelling ad text or you can’t, it’s a difficult art to squeeze the best performance out of so few words. I’ve found that literary devices such as alliteration and rhyming led to small, barely noticeable increases in click-through rate whilst calls to action can lead to slightly larger ones. This small changes are probably due to the psychology of reading ad text, firstly the user reads the strap line, then scans the destination URL for a recognisable brand name.
When I started working in PPC, I was taught a few “golden” rules. One of these is that dynamic keyword insertion is good and would lead to higher performance through better click through rates. More bold is better, I was told. I never bothered to put this to the test until a rather large account landed on my doorstep - thousands of ad groups, hundreds of thousands of keywords. By taking a selection of the best performing and highest trafficked terms, I isolated them into a single ad group and created five ads -
1. Control advert, the same in every ad group.
2 & 3. Advert with dynamic keyword insertion, with default keyword text as the same as 4 & 5.
4 & 5. Full copy ad text, no keyword insertion.
My findings were fairly interesting, generally the control advert performed in the middle. With similar click through rates across the board although not great, as the advert was highly generic basically discussing brand term rather than product keyword. The full copy ad text performed the best - significantly higher click through rates than either 1 2 or 3 across the board. The adverts that used dynamic keyword insertion performed the worst.
Now clearly it is possible that my dynamic adverts were set up poorly, but further investigation analyzing the search terms showed there were fairly optimized. There are presently no figures in this entry as I don’t have the data with me; however I will repeat the experiment over the course of a weekend during March and post the results here. I plan to run similar tests using literary devices over the coming weeks in order to pull together some numbers for them and see how they perform compared with the “golden rules” that I was taught originally.
If anyone has similar or opposite findings - please don’t hesitate to let me know.