In another product update that strengthens Google’s stance on automation and machine learning, the tech giant has decided to reduce visibility for up to 33% of search terms within Google Ads whether or not they led to an impression, click, or conversion.
Many pay-per-click advertisers are upset by this development, because it’s a move away from transparency and control. PPC account managers have long relied on the search terms report for feedback about which keywords are most valuable to bid on, and which ones need to be blocked to prevent the wasteful spending of a clients budget.
Read more from Seer Interactive in their wonderful roundup post.
So why would Google make this move, and why now? In their official statements, they’ve claimed that it was in order to protect the privacy of their users. However, we’ve seen the rise of smart campaigns that span all networks, smart bidding strategies, and the deterioration of match types. Each of these products launches was a harbinger for a future where the primary work in managing a Google Ads account is more strategic and creative.
Regardless of your feelings on the matter, there is a reality to deal with, and we’re all left with some of the same questions: in light of the changes, how do we respond in order to best optimize our accounts to maximize results?
“The Great Filter”
In the new world of paid search, display, shopping, and social, data is the the most precious commodity for advertisers because machine learning algorithms require sufficient “trials” in order to optimize effectively toward a particular goal.
If you receive sufficient volume of “events”, meaning whatever your campaign is optimizing for on a daily basis, the search terms change may not be so important if you’re committed to staying in the game long term.
You should take care to preemptively add negative keywords that you would never want ads to show for, but day-to-day semantic matches of queries are probably less important if you have Responsive Search Ads, smart bidding, and lots of impressions for thematically-relevant keywords.
A really interesting question to ask yourself in this circumstance is: what if I did the opposite of trying to control my account?
For example, how might you go broader in targeting, test maximize conversions, smart campaigns or dynamic search ads (DSA)?
Suppose you run a lower volume account where you don’t receive enough events presently for machine learning to be very effective. In this instance, you’ll want to either increase the number of events or not play the “smart” game at all.
For the first case, you can try to increase the number of events by doing what I call “optimizing up the stack”. This means changing what goal your campaign is optimizing for in favor of one with more overall volume:
If you don’t get enough purchases, optimize for add to cart.
If you don’t get enough add to carts, optimize for product views.
Then landing page views.
Then clicks.
- Then impressions.
Additionally, you can pull other levers to bring more individual campaigns’ data together:
- – Implement a shared budget
- – Use a portfolio bid strategy
- – Broaden match types
- – Add more keywords
- – Consolidate ad set/ad group targeting (within a given campaign)
- In the second case, it may be impossible for these strategies to provide enough results. That may be due to limited budget, small geographic targeting, or poor performance. Here, you will want to maximize control and lower costs as aggressively as possible, striving for maximum profitability.
4 Ways to Get More Data in the Meantime
If we define the principal “problem” with this change as receiving ad clicks for search terms that are irrelevant and NOT knowing about it, leading to wasted budgets, the below are avenues still open to us to reveal more of the map that’s currently shrouded.
- Shift dollars to Microsoft Ads because they haven’t yet made a search terms change to their platform. If you’re heavy up here, you can build more robust negative keyword lists. Review Search Console for long-tail organic terms, especially ones you may not rank or convert well for but conceivably could match with in paid search. Use keyword research tools to mine out long tail from related searches and preemptively block irrelevant root words. NGRAM your historical data as a housecleaning measure, and perform an intensive negative keyword workup