AdWords
4.6K members online now
4.6K members online now
Improve your AdWords performance and boost your ROI, CTR, and Quality Score
Guide Me
star_border
Reply

How long should you allow before measuring impact of Ad changes?

Visitor ✭ ✭ ✭
# 1
Visitor ✭ ✭ ✭

I created ad schedules to bid 20% less at night time (between midnight and 5am) and all day on Saturdays and Sundays on 24/01/17.

 

I've just now run a report from 24/01 to 12/02 compared to the previous period and conversions are significantly lower.

 

Initial data showed that conversions were non-existent at night time and were poor over the weekends hence the ad optimisation.

 

My questions are:

 

1. Is my date / time range sufficient for measuring?

2. Should I undo my changes or ride it out a little while longer?

 

Here's a screenshot of my data

 

Adwords 24-01 to 12-02 compared to previous period.pngConversion data

 

Thanks in advance for your help

1 Expert replyverified_user

How long should you allow before measuring impact of Ad changes?

Follower ✭ ☆ ☆
# 2
Follower ✭ ☆ ☆

Well ... i think is enough. At least for your first row (copy/adgroup?), with 15k impressions,10% CTR and all the 18% of conversions down.

If time schedule was the only change you made, you should treat this row separate.

I must say also, that i optimize time schedule just when all the other parameters are already optimized.

I don't know if this is your case.

 

Re: How long should you allow before measuring impact of Ad changes?

Top Contributor
# 3
Top Contributor

Hi @Emeka O how much time you need to give this experiment really depends upon the time lag in your conversion path.  Put simply, it's common for conversions not to happen on the day of the first click; how big the "lag" is between the first click and the conversion varies significantly from business to business, some will see most of their conversions happening the same day, for others it may take weeks or even months.  This time lag is important because if you have a typical lag of anything more than a few days then the period up to the 12th of February may be "missing" some conversions that will be attributed to a first click in that period at some later date whereas the period before the 24th will have a higher number of these "late" conversions already recorded.  In other words, it's not so much a problem with the length of the period being tested as it is with how recent the data is for the second period, you may be seeing inaccurate (incomplete) conversion data, making the totals look lower than they'll actually be once all attributions are complete.

 

You can take a look at your time lag data under the conversions reporting area of AdWords (see here) but whatever that suggests, I'd recommend leaving at least a week past the end of your second comparison period for testing.

 

If the data turns out to be accurate and you have lost conversions, you'll need to do some digging as to why that might be and again the "path to conversion" could be important.  Presumably you're not seeing any first-click attributions to these night-time periods, which is why you've reduced the bids, but it could be that these night-time clicks are actually part of a multi-click path.  For example, lets say I see your Ad mid-afternoon, go to your website, like what I see but don't buy there and then.  I think about it for a while, then decide in the night to make the purchase; I don't see your Ad, but I see one from your competitor who has a similar price, so I buy from them instead.  It's these paths that make it difficult (and dangerous) to turn your Ads off unless you're completely sure these impressions and clicks are not an important part of the path.

 

Hope this helps.

 

Jon

AdWords Top Contributor Google+ Profile | Partner Profile | AdWords Audits