What are Responsive Search Ads?
Responsive search ads are a new ad format released by Google which is currently in beta. The format allows you to choose up to fifteen 30-character ad headlines and four 90-character descriptions. Google then serves different combinations to customers and their machine learning assesses which combinations perform best together and shows the top performing combinations above the more poorly performing combinations. In theory responsive search is designed to put an end to traditional A/B testing.
The Benefits of Responsive Search Ads
One benefit of responsive search ads is that they allow you to work faster. Simply think of up to 15 headlines and 4 descriptions, and Google will arrange the combinations together for you. There are thousands of combinations that can be created, which saves you time creating separate ads and A/B testing them yourself to find which combinations work best.
Google tests the different combinations and prioritises the best performing combinations automatically. Google is able to take into consideration multiple signals such as search query, device, past browsing behaviour etc. to try and assure the final ad will be the most relevant to the user.
Google also reports that customers using responsive search ads see between 5-15 percent uplift in clicks compared to standard search ads.
At the time of first testing responsive search, the increased real estate was also a major benefit. However, now that Google have introduced 3 headlines and 90-character descriptions for normal expanded text ads, are responsive search ads still the way forward?
The results I have seen personally haven’t been particularly ground-breaking. Although there have been some ad groups where responsive search ads improved the conversion rate by 3-4%, on the whole the performance was very similar to expanded text ads. One interesting thing I found is that not only did responsive search ads not improve the click-through rate; the overall click-through rate actually fell from 5.8% to 3.13% on one account. The area responsive search ads did improve was conversion rate; admittedly this was a very small increase from 9.19% to 10.20%, however in this specific account that extra 1.1% could mean an extra 200 conversions a week.
Another finding from my test was that the responsive search ad stats improved over time, and also improved when we included more headlines for the machine learning to create more combinations with. This suggests that Google’s machine learning used in responsive search does really work but it needs to be given time. To put this into perspective, in our first week of being live we achieved a click-through rate of 2.62% and a conversion rate of 8.06%. With the latest weekly data, after 5 weeks of being live, this rose to a click-through rate of 3.13% and a conversion rate of 10.20%. I think, when considering responsive search ads, the question has to be would this save time on your traditional A/B testing method?
One benefit of using responsive search is that you can check the combinations which have been served to customers along with the impressions they received. Although there is no direct way to see individual metrics such as conversion rate for the best performing ad, you can presume that the headlines and descriptions receiving the most impressions are the top-performers, and Google’s machine learning will prioritise these combinations. This was helpful because it revealed that some of the most successful combinations were in an order that I usually wouldn’t have considered. The ad had a question intended to invoke customer consideration as headline 2 with an informational headline as headline 1. I would have always logically put the question headline first; however the results showed that this didn’t work as effectively. This is now a factor I can bring in to my future ad copy for the account.
My best piece of advice for RSA would be to use all of the space you are given. You can have 15 headlines, so use them! We first tested RSA with 5 headlines and 4 descriptions and the results were disappointing. After we increased the headlines to 15, we did notice an uplift in performance, as Google Ads had more combinations to choose from and test.
Another piece of advice when writing the headlines would be to change it up; it would be useless to have 15 very similar headlines. Try to have an aim for each headline you create and make them highlight something unique in each headline.
Another top tip is to check which combinations are being shown; you can do this easily by clicking on view asset details and then clicking on combinations. It is important to make sure your ads actually make sense and although Google’s machine learning gets it right the vast majority of the time, there has been a few instances where this hasn’t been the case. This can easily be fixed by pinning a headline or description to a particular position; for example, if there is a headline that is designed to grab the reader’s attention it may be best practice to pin this into position 1.
In conclusion, I would say that there are positives to using responsive search ads; however I wouldn’t adopt the strategy thinking that it is going to drastically improve an account. The changes we saw were very minimal and they both positively and negatively affected performance. Saying that, I would also say that responsive search is definitely worth testing. Every account is different, and just because we didn’t quite see the 5-15% uplift in clicks that Google have seen, doesn’t mean you won’t experience it in your account. It is also worth bearing in mind that responsive search ads are still in beta, so although in this case responsive search ads didn’t perform as well as Google suggested, it doesn’t mean that we won’t see this performance once the full version is launched.