Help With Split Testing Bing Ads

5 replies
  • PPC/SEM
  • |
Hey warriors!

I would just like your help on something as I'm new to Bing ads.

I understand that it's common practice to keep split testing your ads and keywords to try to improve CTR.

What I would like to know is what data to use. So for eg if you have been testing an ad or keyword for say 1500 impressions and then decide to pause an ad or keyword that has been out performed, would you then start at 0 when split testing with the new ads or keywords?

Bing has an option to select different time scales, so I guess what I'm asking is do you always use the "Entire time" option or the relevant day you have added a new ad/keyword when checking your figures.

I hope this makes sense!
#ads #bing #split
  • Profile picture of the author ZMorgan
    Banned
    [DELETED]
    {{ DiscussionBoard.errors[10448981].message }}
    • Profile picture of the author cooler1
      Originally Posted by ZMorgan View Post

      I like Bing, I have a 9.28 CTR, but the bid is increasing like crazy lately.
      That seems to be true.

      I created a campaign and set the bid to $0.30 but it got very few impressions and no clicks even though the ads are relevant to the keywords. So I had to increase it to about $0.60 to start getting impressions and clicks.

      People put a lot of emphasis on the importance of the quality score to get cheaper clicks, but in my experience even if a keyword has a quality score of 10/10 the CPC is still high, the same as if a keyword has a quality score of 7/10.
      Signature

      {{ DiscussionBoard.errors[10450003].message }}
  • Profile picture of the author Bright Future
    Hey, panditmarketing

    I would suggest always comparing the statistics for the time period in which both ads were active and working.

    It makes the split test more objective. Mostly because many other factors that change over time can impact the effectiveness of your ads, for example, seasonality, specific events, competition etc.

    Also, don't just optimize ads fot CTR. Compare the conversion rates of different ads too.
    {{ DiscussionBoard.errors[10449320].message }}
    • Profile picture of the author panditmarketing
      Originally Posted by Bright Future View Post

      Hey, panditmarketing

      I would suggest always comparing the statistics for the time period in which both ads were active and working.

      It makes the split test more objective. Mostly because many other factors that change over time can impact the effectiveness of your ads, for example, seasonality, specific events, competition etc.

      Also, don't just optimize ads fot CTR. Compare the conversion rates of different ads too.
      Thanks for your advice.
      Signature
      Why You And 98% Of Affiliates Are Failing Online Go Here!
      {{ DiscussionBoard.errors[10449961].message }}
  • When testing - anything, whether ads or manufactured parts - you need to have the same reference. So you need to compare Ad1 to Ad2 in the same time frame and same parameters then Ad1 to Ad3. If you look at Ad1's all-time stats to Ad3's, you are not exactly comparing apples to apples and you may come to wrong conclusions. In your words, start at 0, make comparisons at the time you made a change.

    So definitely compare the same time frame. Also, any change to your ad group or campaign must be taken into consideration. That includes changes to keywords, including negatives, bids, using ad extensions and everything else, including changes to landing pages. You are essentially restarting the test every time you make a change somewhere because again, you need the same reference to properly compare. Not doing so really invalidates your test results which can lead you making wrong or poorer decisions.

    As for data, there are two main metrics: CTR and conversion rates. The two together would determine your profits so you have to balance these out. You also need to take position into account when comparing CTRs. A 5% CTR does not have the same value in fourth position as it does in 3rd. Ad position must be about the same or again this can lead to poor decisions.

    You also need enough data. The important part is not the impressions, it's the clicks for CTR and conversions for conversion rate. It means little if one ad has 5% CTR and the other 10% if they each only have 5 clicks. You need dozens of clicks to make the test statistically significant. Same for conversions.

    Finally, never edit an ad. This deletes it and creates a new one. Instead, pause it and create a new one. You never know when you'll need that ad again. Plus, you can test it under different circumstances if you make some other change. You'd be surprised how often an ad you thought was no good later proves to be better than thought just because it's running under different parameters.
    {{ DiscussionBoard.errors[10449918].message }}
    • Profile picture of the author panditmarketing
      Originally Posted by LucidWebMarketing View Post

      When testing - anything, whether ads or manufactured parts - you need to have the same reference. So you need to compare Ad1 to Ad2 in the same time frame and same parameters then Ad1 to Ad3. If you look at Ad1's all-time stats to Ad3's, you are not exactly comparing apples to apples and you may come to wrong conclusions. In your words, start at 0, make comparisons at the time you made a change.

      So definitely compare the same time frame. Also, any change to your ad group or campaign must be taken into consideration. That includes changes to keywords, including negatives, bids, using ad extensions and everything else, including changes to landing pages. You are essentially restarting the test every time you make a change somewhere because again, you need the same reference to properly compare. Not doing so really invalidates your test results which can lead you making wrong or poorer decisions.

      As for data, there are two main metrics: CTR and conversion rates. The two together would determine your profits so you have to balance these out. You also need to take position into account when comparing CTRs. A 5% CTR does not have the same value in fourth position as it does in 3rd. Ad position must be about the same or again this can lead to poor decisions.

      You also need enough data. The important part is not the impressions, it's the clicks for CTR and conversions for conversion rate. It means little if one ad has 5% CTR and the other 10% if they each only have 5 clicks. You need dozens of clicks to make the test statistically significant. Same for conversions.

      Finally, never edit an ad. This deletes it and creates a new one. Instead, pause it and create a new one. You never know when you'll need that ad again. Plus, you can test it under different circumstances if you make some other change. You'd be surprised how often an ad you thought was no good later proves to be better than thought just because it's running under different parameters.
      Thanks for your advice, this makes sense.
      Signature
      Why You And 98% Of Affiliates Are Failing Online Go Here!
      {{ DiscussionBoard.errors[10449964].message }}

Trending Topics