Google’s search terms move will make millions in ad spend invisible to advertisers
This morning, I negated a word that cost a campaign more than $3 for the one click it received in a brand campaign last week. I didn’t add the whole query, just one irrelevant word that triggered a brand keyword. Going forward, I might not ever see that type word or know if it showed up across multiple low-volume queries.
As we reported yesterday, Google has notified advertisers the search terms report will “only include terms that were searched by a significant number of users.” It has given no details about what “significant” means. The company told us the reason for the change is “to maintain our standards of privacy and strengthen our protections around user data.”
Unsurprisingly, the move has angered advertisers.
To be clear, maintaining and respecting user privacy should be a priority for tech companies like Google as well as for marketers. Google should keep personally identifiable information (PII) and otherwise sensitive queries that could be tied to someone out of search terms reports and it’s other platforms (the amount of PII that gets passed through to Analytics is a topic for another day). But, Google’s oblique communication about this change has left the door open for skepticism about its motives.
The “not provided” privacy issue is nearly a decade old
Next fall will be the 10th anniversary of “not provided.” Google started limiting visibility into the search terms that drive organic traffic to websites in 2011 for users logged into Google, again citing privacy. The so-called “not provided” issue (how that traffic appears in Analytics reports) has frustrated site owners ever since. For whatever reason — Google’s never been clear about it — the company has continued to surface that data to advertisers in the search terms report.
Is it hypocritical and self-serving to give this data to advertisers and not to others if privacy is behind the decision? It’s hard to deny. Limiting search term data would have undoubtedly hurt Google’s ad business, though advertisers have had a pretty good argument that they’re paying for the clicks and should have access to be able to eliminate wasted ad spend and optimize their campaigns. When “not provided” happened, keyword buying was the only way to buy search ads, and the way keywords matched to search queries was relatively straightforward.
So why make this change to ad search terms now?
We don’t have clear answers on why now. Google’s not saying.
We can either take Google’s statement at face value:
That there was suddenly a privacy epiphany about this data, or that there were enough specific scenarios detected to raise attention, or that this is a reaction to the increase in data privacy regulation and anti-trust scrutiny. We don’t know if there was a specific trigger. GDPR went into effect in 2018. CCPA went into effect Jan. 1 of this year, enforcement started Jul. 1. We know Google had CCPA on its radar because it introduced its solution for advertisers last November (unlike Facebook which waited until July).
Or we can be more skeptical:
Google isn’t providing responses to follow up questions we’ve asked. Regardless of Google’s intentions, there are millions, if not billions, of dollars in ad spend reflected in this longtail data. It’s no wonder advertisers are looking for more answers and clearer communication.
Last quarter, Google’s Search business was hit hard by advertisers pulling back amid the pandemic. The company generated $21.3 billion from Search and other properties such as Maps, a 10% drop year-over-year. (Total revenues were off 8%, marking the company’s first year-over-year decline.) I’d like to think this isn’t a revenue-incentivized decision, and it is certainly reflective of the trend toward automation (more on that later), but it’s nearly impossible not to view this kind of decision, at least in part, though this lens.
Does this longtail search query data really matter? What’s the big deal if a keyword only ever gets one click?
All that data adds up to big money and important insights for advertisers. Based on what some are seeing, advertisers could lose visibility into the search terms that account for a quarter or more of their ad spend.
Digital agency Seer looked at the initial impact of this change and calculated that Google Ads now “hides search terms for ~28% of paid search budgets and removes search term visibility of 20.4% of PPC clicks.” Meaning:
For every $100K you spend on Google search, you get search term data for $71,000 of it.
For every 100K clicks you get, you see search term data for 77,900 of those clicks.
The irrelevant word that showed up in that $3 search term I negated this morning could be sprinkled across many other low-volume (again, we don’t know what the threshold is) search terms and make a substantial dent in that company’s marketing budget. And, nobody would be the wiser.
The implications for small businesses, in particular, are huge.
Moving away from keywords — All aboard the ML train
Search advertising was built on keyword targeting. Then came Facebook and it’s audience targeting power with pinpoint “people-based marketing” capabilities.
And at the same time there have been massive advancements in machine learning and AI.
Google has subsequently marched toward audience targeting and away from keyword targeting, diluting match types and introducing campaign types that don’t use keywords at all. As I wrote last year, when Google extended same-meaning close variants to phrase and broad match modified keywords:
The days of zero-keyword search campaigns have long been predicted as machine learning has taken over and audiences have come into play. In fact, those days are already here with automated campaign types such as Local campaigns, Smart campaigns and App campaigns.
https://searchengineland.com/is-google-prepping-to-phase-out-etas-339964A more recent example of this shift to machine-learning powered audiences: In July, Google introduced audiences based on new predicted purchase and churn metrics in Google Analytics. (On the ads front, Google is making its automated responsive search ad format the default in a current test.)
This requires more trust than Google has earned
I’m no Luddite when it comes to automation, machine learning and AI. If it means better results and less work, bring it on. But, we aren’t at a point where marketers can take our hands entirely off the steering wheel.
Let’s say we put aside the argument that advertisers should have access to (non-PII) search query data because they are paying Google for those clicks. For advertisers to be remotely comfortable with this change, they need to trust that the algorithms that match query intent to keywords are as good as Google would like us to believe they are.
No doubt, the algorithms are getting better. But, the proof that they are far from perfect is in … those search terms reports.
Google hasn’t said this, but I can imagine the response now: “Our data show that .000001% of advertisers actually act on that data and add negative keywords.”
Whatever the number, it doesn’t matter.
“We’re continuing to invest in new and efficient ways to share insights that enable advertisers to make critical business decisions,” Google said in its statement to us. We’ll have to wait to see what this means.
In the meantime, advertisers will continue to feel like they’re yelling into the void. This move also potentially squanders the goodwill the company garnered by issuing ads credits to businesses hurt by the pandemic. As we head into the critical fourth quarter, more critical than ever given the pandemic, advertisers will be forced to navigate without visibility into a sizeable share of their ad spend — or look elsewhere.
- ^ searched by a significant number of users (searchengineland.com)
- ^ September 3, 2020 (twitter.com)
- ^ 2011 for users logged into Google (searchengineland.com)
- ^ Google’s never been clear about it (marketingland.com)
- ^ continued to surface that data (searchengineland.com)
- ^ hypocritical (searchengineland.com)
- ^ GDPR (martechtoday.com)
- ^ CCPA (marketingland.com)
- ^ solution for advertisers (martechtoday.com)
- ^ unlike Facebook (searchengineland.com)
- ^ @sengineland (twitter.com)
- ^ @GinnyMarvin (twitter.com)
- ^ #ppcchat (twitter.com)
- ^ September 3, 2020 (twitter.com)
- ^ September 2, 2020 (twitter.com)
- ^ Seer looked at the initial impact of this change (www.seerinteractive.com)
- ^ wrote (searchengineland.com)
- ^ Local campaigns (searchengineland.com)
- ^ Smart campaigns (searchengineland.com)
- ^ App campaigns (searchengineland.com)
- ^ predicted purchase (searchengineland.com)
- ^ responsive search ad format the default (searchengineland.com)
- ^ no Luddite (searchengineland.com)
- ^ ads credits to businesses (searchengineland.com)
Powered by WPeMatico