Paid-Search Automation Not Ready for Prime Time
January 29, 2019 | Contributed by: Joe Morsello
Back in December I received an email from Google asking to develop a completely automated paid search campaign on behalf of my new digital marketing agency, Rithm Marketing. My guess is that this is part of a larger effort to test the company’s ability to build automated campaigns from end to end (keywords, to bids, to creative, etc.). I was skeptical but gave it a shot as I was also curious to see what the technology could do.
A few weeks later, I received an email notification from Google saying some of the ads were disapproved because they linked to a different domain. When I logged into my account to check, I discovered the campaign was linking to a completely different agency and using copy from that agency’s website (see below). I turned the campaign off immediately.
The idea of completely automated campaigns, based on a business’s website and/or other data is very attractive, especially for small businesses wanting to utilize Google Ads. But if the technology doesn’t work as promised, I would expect many to abandon it just as I did, even if only one component of the campaign failed. If too many of these experiences happen, Google will have an uphill battle getting advertisers to trust that the technology works.
While my experience may or may not be unique, it only takes one failure to cast doubt on the technology behind the company’s automation efforts. Also, it wasn’t like the mistake was a poor keyword choice. The actual linking URL was completely incorrect and has no relationship to my business; it was another digital agency entirely.
The idea of ad automation is great, but the efficacy of it isn’t — yet. Back in the summer of last year, I was originally skeptical of Google’s “responsive search ads” which allowed marketers to create over 15 headlines and multiple ad descriptions and then select the best performing combinations. My issue wasn’t with the technology but was more related to the creative. I wrote:
The emotional nature of consumers and subjectivity that influences actions can’t be accounted for by a machine. The responsive search ads merely improve the odds of landing on something interesting or compelling.
In this scenario, automation has the capability to help marketers learn what the audience cares about, but it doesn’t completely remove the need for a discerning human. Also, as any creative could attest to, inspired creative (copy/images) comes on its own time. Just because there are 15 headline options, doesn’t mean the chances of resonating with an audience are greater, especially if the copywriter isn’t creatively inspired at the time.
The question of automation’s impact on agencies was the focus of a recent LSA webinar. During the webinar, Greg Sterling of LSA, Alex Porter of Location3 and David Mihm of Tidings/ThriveHive discussed the role that automation will play and how it’s going to impact agencies. Though there wasn’t complete agreement, ultimately everyone agreed that automation, when accurate, had the potential to disintermediate agencies by allowing advertisers to buy high performing search ads with ease.
In addition, just last week, Google announced that they would be identifying key changes on behalf of advertisers in order to help them “get more out of ads.” This included restructuring ad groups, modifying keywords, adjusting bids and updating ad text. It appears as though these efforts will come from an actual person, however it is likely the company’s machine learning tech and automation tools will be utilized in order to scale these efforts.
Given my experience, and likely the experience of other test accounts, automation isn’t ready yet. Additionally, even if automation on Google worked as intended, how would any of the small businesses know the ads were effective unless they implemented sophisticated analytics and conversion tracking? So even if the ads are effective, someone (an agency) needs to connect the dots and provide the proof.