Hi all,
I am curious on the meaning of the Click Accuracy values. Can some one provide some sort of description on the value? How does it relate to changing competative environemnts? Does a budget shift increas the inaccuracy of the moddel due to changes in the impression level?
Looking forward to your response,
Regars,
Nana
Solved! Go to Solution.
Click accuracy is the difference between forecasted clicks versus actual clicks. It is not tied to impression share; rather, it is related to the clicks metric that is logged in AdWords and then passed on to Ad Cloud via the API.
A budget shift does not necessarily increase the inaccuracy of the model. If your model accuracy is strong, then shifting to a new budget level, especially if it's a spend level that's been included in the simulation you will have run, should carry on with a similar level of accuracy.
Click accuracy is the difference between forecasted clicks versus actual clicks. It is not tied to impression share; rather, it is related to the clicks metric that is logged in AdWords and then passed on to Ad Cloud via the API.
A budget shift does not necessarily increase the inaccuracy of the model. If your model accuracy is strong, then shifting to a new budget level, especially if it's a spend level that's been included in the simulation you will have run, should carry on with a similar level of accuracy.
Adding to Meghan's response if due to increase in Competition the CPC's inflate resulting into lower number of clicks than predicted this may cause a temporary click inaccuracy. However due to half lives setting of the portfolio it is gradually modeled into the system and the bidding pattern so generated takes care of this as well.
Adding to above answers,
Yes, competition can impact click accuracy, because the click forecast is based on historical data collected by AMO and due to competition the click volume is different from ongoing trend, it reflects as click inaccuracy.
A massive budget shift can impact this accuracy too as the forecast remains based on the historic data.
In case you experience major inaccuracy for some days it is recommended to exclude these dates for Data collection bucket, such that the forecast is not impacted by these outliers.
And in case it is major budget change, the impact on accuracy will settle down as mentioned by Ritika.
Further if there are constantly changing trend in your data, it is recommended to keep a shorter Half life for the model, thus allowing it to respond to these changes more accurately.
Views
Likes
Replies