Whenever you take over new campaigns or want to restructure outdated campaigns, you are facing a potential drop in traffic due to the loss of the performance history of the old campaigns. The new campaigns will need time to build up history and for Quality Score to improve, but how long does this take? We performed a test to estimate the loss in traffic over the first month of a new campaigns life. In order to do this we chose some high traffic campaigns and duplicated them into a new account. To avoid cannibalisation keywords day parting was set to an alternating two hour schedule. Geo-Targeting and device settings were kept but RLSA lists were paused during the testing period to rule out any influence on performance.
As the test campaign was setup in a completely new account that didn’t have any performance history which could aid in pushing performance, we expected Quality Scores (QS) to be lower while Average Positions and CPCs would be higher than in the original campaigns. We estimated these initial performance differences to be especially strong during the first few days but to level out in time. What we still were unclear about was the exact amount of traffic we would lose out on and how much influence the QS had on the volume of traffic.
Quality Scores Quickly Caught Up to the Original Level
After starting at the standard level of 6 on day one, the QS improved quickly and within one week even exceeded the average of the original campaign:
After lingering just below 8 for a few days QS jumped across the campaigns on day 5, surpassing the QS of the original campaign.
When we compare the time period before QS improved and after we can see that while QS was about 8% lower in the new campaign in the beginning before increasing by one point, exceeding the QS in the original campaign over the rest of the testing period.
Performance Picked Up Immediately After QS Improved
Did the improved QS have an impact on the overall performance and traffic volume?
During the first four days of the test the overall traffic of the new campaign reached only about 70% of the volume in the original one, though due to a slightly higher click rate the loss of Clicks was slightly lower than the loss of Impressions.
The numbers above indicate that, as expected, the lower Quality Scores due to the missing account history decrease the odds during an AdWords auction, which in turn leads to higher positions and CPCs in addition to a lower Impression volume in general.
Looking at the rest of the testing period performance is much more evenly distributed between the two campaigns. Impression and click volumes are almost identical now, as are CPCs. Only the Avg. Position is still better in the original campaign, the difference is still at 14%. This is hinting that the performance of the new campaign could be even better at a similar Position.
If we compare the four days before QS picked up with the period after it looks like the improved QS improved traffic significantly. Both Clicks and Impressions already reached a similar level just 5 days after the setup. CPCs and CTR, too, both only changed very slightly during the following three weeks.
What Does This Mean?
While we lost about 32% of Impression volume and 29% of Click volume during the first four days of the test, we only lost 3% of Impressions and 5% of Clicks over the rest of the test period. Of course this still adds up to a significant total difference in the end and shows that even after a month, traffic has not reached its original quantity.
Of course the QS is not the only factor influencing performance in the early stages of a campaign. One other factor for lower traffic could also be that the number of keywords receiving impressions at all is lower. In our case during the first four days the new campaign had 17% less keywords that received impressions than the original campaign while in the end the number was more even, with the new campaign having 5% more kw with traffic:
When Does It Make Sense to Shift Traffic?
The results of the test suggest a transition period between upload and traffic shift. It makes sense to upload the new campaigns and let them run alongside the existing ones for about a week, using a similar scheduling model as described above. Only after Quality Scores have improved, more traffic should be shifted towards the new campaigns until finally the original campaigns can be paused.
This way the transition can be performed smoothly without a loss of traffic.