"SKAG"(Single Keyword Ad Groups), "Alpha-Beta". These are all examples of popular jargon used to describe various Google Ads account structures. One of the latest additions to this list is the recently found "Hagakure'' structure. This structure is being advertised often by Googlers as being a future-proof method which utilizes the features of Machine Learning.

Should I use Hagakure?

Smart Campaigns and Smart Bidding have boomed in 2020. Plenty of studies have shown that Machine Learning is now simply better (whether this is fair or not is a different question) than Manual Bidding in reaching set objectives. This explosive rise in everything "Smart", has changed the way that campaigns are to be structured. Where account managers previously relied on very granular ad groups (in order to serve the most relevant ad per query), now ad groups need to contain more data in order to optimize. We need to consider:

  • More conversion data is needed within the campaign to allow A/I driven optimization.
  • More data is needed on the Ad and Ad Group level to optimize ads.

As Smart Ads (Responsive and Dynamic) adjust Headlines based on the keyword and query, there is less need to match the ad with the keyword. Rather the challenge is feeding enough data to the ad and ad group for assets to optimize efficiently. This is why the Hagakure method was proposed.

  • Rather than taking the keyword as a base for the ad, Hagakure takes the destination URL as the base. Ad groups are split per URL rather than keyword group. 
  • The keyword match types in Hagakure are generally in the Broad Match Modifier match type rather than Exact and Phrase, to capture sufficient volume.
  • Dynamic Search ads are used for URLs which don't have enough volume to split into separate ad groups.
  • IF-statements are used to make ads more relevant to the user.

The structure is very easy to manage across a large variety of products, which is why it is commonly used for eCommerce clients. It embraces Machine Learning to achieve the best results. As A/I is improving over time, Hagakure ensures that your campaigns are "future-proof".

So far so good, However, there are pitfalls to keep in mind, which are inherent to reliance on going "Smart" and to Hagakure. These are some of the main ones:

  1. Not enough data?

Hagakure recommends to split out ad groups when a destination URL has at least 3000 impressions. This will ensure that the algorithm has enough data to learn from and stay optimized. However, it may happen quite frequently that there just isn't that much data. Even larger webshops may have issues getting this many impressions as the impressions are divided across sometimes hundreds of products. While the total number of impressions may be high, single URL impressions may be low.

The impression requirement makes it difficult to start off an account with Hagakure, as there hasn't been any historic data to validate the 3000 impression requirement. This leaves Hagakure mostly as an option for a full account restructure. Which is a large undertaking and in cases where campaigns are already profitable. Often the choice is made to just optimize current campaigns rather than go for a full restructure.

  1. IF-functions in ads can't be split tested

Ads with IF-functions, Dynamic ads and Conditionally formatted ads suffer from an important issue, they can't be split tested. As the exact content of the ad is only determined at the moment of the impression, we can't simply compare ad A vs ad B, as we don't know which parameter made certain ad perform better than the other. It becomes increasingly difficult to extract meaningful conclusions from the ad-level data, not only because the ad content is determined in real-time, but also because there is a wide variety of keywords triggering the ad. In ad groups set up with Hagakure, it's unwise to use ETA ads as the content may easily mismatch with the triggering keyword. Regarding ad content management, we are thus left in the hands of the A/I.

  1. Dynamic Search Ads 

Hagakure proposes to use Dynamic Search Ads in order to serve the bulk of the traffic that doesn't reach at least 3000 impressions. However, there are reasons why professional Google Ads account managers often hesitate to rely on them. 

In Hagakure, the targeted URLs in the DSA campaign need to be continuously updated as URLs pass or drop below the 3000 impression threshold, meaning that they would need to be split out. When setting the campaign simply to target "all webpages'', you risk overserving the same products in several campaigns.

Another major issue is that Dynamic Search Ads will primarily work well for companies that already have good SEO. As the headlines are often pulled from the meta tags in the html, poor title and description tags will produce poorer ads. When title tags are too long, or not properly set, then ad texts may be truncated. Besides that, we once again lose the overview of which exact texts are served and once again need to rely on the algorithm to produce results.

Conclusion

Even though Hagakure is leveraging the power of machine learning (which is clearly the direction in which Google Ads is going), there are quite a few things to be careful for. It may be a smart idea to switch campaigns to Hagakure one at the time to evaluate if the change really led to better results and before even starting, you better be sure to: have enough conversion data, have strong title and description tags on all pages and make sure you are comfortable in losing control over your creative ad copy.

Artificial intelligence makes our work much easier and frees up time that we can spend on more strategic tasks. In this article, we discuss whether marketers should be afraid of losing their jobs to  A/I.

At eLama, we help you set up your Google Ads campaigns appropriately as our ad creation tool helps to set up your campaigns along the way throughout the campaign structure and creation and build the strongest ads with our custom ad templates. 

Start your free trial today!
 

Start your free trial today