Among the many changes in digital marketing over the last two years, one of, if not the largest shifts in paid search has been the release of modern search, Google’s latest best practice, which was first announced at the start of 2021.
Central to this new approach is a simplification of match type coverage and segmentation, validated by the elimination of broad match modifiers (BMM) in July 2021. Moving from BMM to pure broad match may be scary and counterintuitive for classic search marketers.
While tools like Dynamic Search Ads (DSA), which dynamically match to search terms based on page content, have been around for some time, traditional paid search strategy has largely been about control. This entails utilizing large lists of finely segmented keywords, often segmented by match type. Conversely, a modern search approach relinquishes some control to the “machine,” trusting it to accurately match to and bid on relevant terms at large scale.
Shifting towards modern search
To deploy modern search and effectively use the full broad match type, we need quality conversion data and auctions at scale.
Quality conversion data
SmartBidding strategies need to have data at the most granular level possible, on what actions are important or provide value to your client’s business. This means setting up appropriate conversion goals and weighting them accordingly. While first-party and customer relationship management (CRM) data is often most valuable, thinking beyond just leads, sales and revenue may be necessary if volumes are not significant. Actions taken on site that suggest a higher propensity for a full conversion to take place may be an effective proxy. Track these, weight them appropriately and include that data in the bidding strategy.
Auctions at scale
Broader match types lead to higher query volume, which results in greater accuracy and more auction data for the bid strategies. In other words, when the bidding system has a very high degree of confidence, it can set an appropriate bid for a search term with a 0.01% conversion rate, just as much as it can for a term with a 3% rate. Previously, the 0.01% query may have been paused or blocked, because there were no means to bid according to its value.
This provides previously untapped expansion opportunities. However, the issue with this is that it increases the disconnect between our keyword sets and what queries we’re actually matching to. Essentially, how do we navigate an account when we go from 10,000 queries from 1,000 keywords, to 1,000,000 queries from 100 keywords? How do we interpret this massive data set to make optimization decisions and take advantage of it from a content perspective?
This is where n-gram analysis comes into play. N-grams are essentially the splitting of a search query string into two, three, four or n numbers of grouped terms. For example, if we take the search query “where to find cheap car insurance”, and want to analyze it using a two-gram approach, the output would be:
- “Where to”
- “To find”
- “Find cheap”
- “Cheap car”
- “Car insurance”
Similarly, a three-gram approach would look like this:
- “Where to find”
- “To find cheap”
- “Find cheap car”
- “Cheap car insurance”
When we collate the performance data associated with these n-grams, we can start to understand patterns. When we use this analysis before and after a transition to a different match type (full broad), we begin to get a much clearer picture of how the search term landscape within our account has changed.
N-gram analysis in practice
At Croud, we recently used this approach for a client shifting towards modern search, where we analyzed 100,000 n-grams which resulted in two impressive findings:
- Finding gaps in coverage: The analysis uncovered several mid- to long-tail search query trends that we weren’t covering before. Think of it as the new structured query reporter (SQR).
- Understanding unintuitive search behavior: We were able to observe a huge increase in the type of searches involving search behaviors we had not fully appreciated before. This included a higher number of search terms containing adjectives and ‘near me’.
The n-gram analysis is one example of the methods and tools that we, as digital marketing practitioners, need to develop in order to adapt to a modern search approach. We need to create new ways to understand what is actually going on, to the greatest extent the platforms will allow. This isn’t so we can interfere in the minutiae or to exert control, but to help build trust in the systems and to have a high-level understanding of how to manipulate them to best achieve our clients’ objectives.