After our recent article, Join or Die: It’s Time to Embrace Google Automation went viral (or at least, as viral as an article about ad tech could go), I received the following LinkedIn message from a product lead at Google:

I had a few long calls with Adam and his colleagues, and they eventually asked me to come and speak at a training at Google’s NYC office about the impact of automation and how AI has changed the role of the agency.

Due to our partner status with Google, we have a team of 5 Googlers who are dedicated to our client’s success. They assist us with strategy, training, access to beta features, advanced reporting, and occasional emotional support when needed. In many cases, they’ve called into or even attended client meetings. They fall under the Agency Development and Account Strategy umbrella at Google and they’re fantastic.

This training session was for their peers, and the audience consisted of Googlers that assist other top-tier agencies. We'll be flying to Google's London offices in April to continue the conversation with a handful of European agencies and their respective Google teams.

Below are a somewhat-organized collection of my prep notes before the Q&A session. I hope you find it valuable.
‍

What were your initial thoughts on bid automation?

Like everyone, I was incredibly skeptical. We’ve all heard the idea that automation is just a way for Google to get you to spend more money on their platform.

When I was just getting started, there were a lot PPC thought leaders that held that opinion and you listened to it.

In reality… that opinion comes from fear. As an agency, you are constantly worried that your clients are going to think that you’re not providing value to them or that automation makes it easy enough to manage your campaigns in-house. If Google creates technology that makes agencies irrelevant, then how are agencies going to survive?

This is no surprise, but there is a serious issue with the perception of automation and it can only be changed by the people in this room. You need to convince agencies that you are not trying to make them irrelevant. You need to say, “not only do I not want to replace you, but I am willing to teach you the skills that will make you irreplaceable.”

And then you need to deliver on that promise by developing better training material and doing everything you can to increase buy-in.

We’ve embraced the shift and have spent the last 18 months ensuring that we are adapting to changes in technology, consumer behavior, and the needs of our clients. It’s an ongoing challenge, but we’re a better agency because of it.

‍

‍

How did you overcome skepticism towards automation with your agency?

It’s tough. If you’ve done this long enough, you’ve definitely had an experience where you’ve been burned by automation in some way. And people have lost a lot of money and a lot of clients from these mishaps.

People are always afraid of making the same mistake twice, especially when it’s something that’s outside of your control. There’s a lot of folks that oppose autonomous vehicles, despite overwhelming evidence that fewer accidents take place as a result. I’ve spoken to people that hold this opinion, and they all seem to say that in an emergency, they want to be the one holding the steering wheel.

If people hold these unwavering, probably-wrong opinions when it comes to the well-being of themselves and their loved ones, then why should we expect it to be any different when it comes to running an ad campaign?

Skepticism can really only be replaced when you see success as a result of automation. So the first challenge is to overcome the willingness to try and try again.

Education was the main driver of this. We’ve been blessed with a great team at Google that has been incredibly helpful in not just training our team, but fostering conversations that take place internally.

But we’ve also benefited from embracing AI and machine learning across the board and taking advantage of training that is available elsewhere. A few years back I attended a handful of training sessions for a programmatic DSP that we were starting to use. Several sessions went in depth about how their bidding algorithms work, and how you can manipulate the algorithm to achieve your goals.

This was my first exposure to the concept of guiding the machine, which helped me understand that there will be a role for humans to fill in the automated world. The roles will be very different from what we are used to, but if you are willing to learn new skills, you will remain relevant and valuable.

The more you learn about the technology, the more you are willing to test it. Increased education also increases the likelihood of seeing success.

And if it works, you tell your colleagues about it, and so on.

Regarding automation, how do you set expectations with your clients?

We have more discussions internally about managing clients expectations than we do any other topic. Expectation management helps develop trust, which is the single most important aspect of a client/agency relationship.

Trust is more important than producing profitable campaigns for your clients. We’ve lost clients that that we’ve knocked it out of the park for, but dropped the ball when it came to expectations or failed to earn their trust. On the flipside, we’ve retained clients with underwhelming performance, but they trust us enough to work through the rough-patches and turn things around. (You obviously need to deliver on that promise and improve performance, but you’ll never have the opportunity to do so if the client does not believe that you have that ability.)

There is always room for significant improvement regarding expectation management.

Regarding automation, all clients are different. Some come to us having run automation themselves and are very open to it, and others have been burned in the past or have preconceived notions and stick to them.

It’s not about getting the green-light to introduce automation, it’s about whether or not client trusts you to make the right decision. Everything we do regarding expectation-setting is carefully orchestrated to help develop that trust.

When you ask the right questions from day one, and then explain why you are interested in that information, you are more likely going to earn your client’s trust. Why do we care so much about micro-conversions on your site? Because it allows us to feed better data into Google so that we can leverage machine learning.

At that point, if they push back and say that they’ve experimented with automation in the past and didn’t see results, then you can simply suggest that it wasn’t carried out in the right way.

Within your agency, how much time have you saved by implementing automation?

Conservatively, I’d say that automation has increased account management efficiency by about 15%. And that is exclusive to the physical work that is done inside accounts and does not have an impact on the time spent on client communication and reporting.

There are two scenarios that take up much more physical time than the average account: New clients, and clients that are struggling with performance.

New clients will always take up a ton of time. There are so many checkboxes you need to hit at the early stages of the relationship to ensure a smooth onboarding. It also takes time to learn the client’s business, industry, and develop rapport with the actual people you will be working with.

Automation has not really impacted this much, if at all.

However, automation has helped us improve the performance of client’s campaigns. If we’re able to turn profit from a larger percentage of our campaigns, then fewer clients are struggling.

So that’s one instance where automation has had an impact on time-savings.

But there are indirect benefits of this as well. We’ve worked extremely hard over the last few years to improve client retention. There are three themes that we’ve been focusing on to help improve retention:

1. The quality of clients that we bring on (and the willingness to turn down potentially bad clients, even if it does seem like a lucrative decision in the short term).
   
   
2. Expectation setting, and establishing trust from day one.
   
   
3. Performance and profitability. As mentioned, automation has had a positive impact in this area.
   

If automation improves performance, and if performance improves retention, then automation has an impact on client retention.

And if we are able to retain more clients, then we have the privilege of being more selective with the clients we bring on, which helps improve bullet #1 above and therefore supports our retention.

We bring on much, much fewer clients now than we used to. As I stated earlier, new clients take up a significant amount of time, and so that’s additional time savings.

So… yes, automation has had a significant impact on our business on multiple levels.

How did you re-allocate time savings gained through automation?

We are now able to dive deeper into client’s businesses and increasingly serve as marketing consultants to our clients. We now spend a lot of time running Conversion Rate experiments, developing and testing landing pages, listening in on sales calls, and working on higher-level strategic planning that improves the efficacy of client’s ad spend.

We recently flew out to spend 3 days with one client and their sales team. We had been tracking deep insights across their sales team to find trends of diminishing returns as it related to the quantity of leads that each salesperson should optimally receive each day.

The hypothesis was that if there are fewer leads, the salespeople will work harder to close each opportunity. If there are unlimited leads, salespeople will likely pass-through valuable leads in search of easier deals. The latter scenario might result in more closed-deals, but is much, much more expensive and less profitable from an advertising perspective.

We then helped develop a new incentive structure for their salespeople that would allow them to increase their overall productivity and close rates. The results allowed us to scale the campaigns in a way that earned more net profit for the company and more income for the salespeople.

That project had a tremendous impact on a lot of people, and that’s the kind of work we weren’t doing 3 years ago.

One Googler asked a follow up question about pushback from agency team members that aren’t willing to change, or enjoy the seemingly mindless tasks that have been replaced by automation. My response to this was along the lines of:

If you have team members that are not willing to change their habits and keep up with the direction of the industry, then you need to find new team members. To quote Jim Collins, author of Good To Great: Step 1 is to get the right people on the bus - Step 2 is to get the wrong people off the bus.

Nobody in our agency really wants to sit around all day performing mundane tasks inside of a Google Ads account. I’ve done that, and I don’t find it very fulfilling.

What we do enjoy is solving complex problems, running unique experiments, and coming up with extremely creative campaigns. We enjoy gaining a 360-degree view of our client’s business and industry, and truthfully knowing that our work is having a meaningful impact on the client’s bottom line. That’s the kind of work our team finds fulfilling.

What impact did Automation have on your team? Junior-level, senior-level staff, etc?

Junior level team members generally embrace automation more easily than senior-level staff that have been burned by automation in the past (or at least have constantly been hearing about how automation is bad from their colleagues). But everyone on our team has been exposed to enough education and positive results that we are all on-board.

Not to brag but… we bring on high level people that just get it. As part of our hiring process, we ask all candidates (whether or not they have specific experience in managing Google Ad accounts before), to put together their own presentation about how the Google Ad auction works, what Quality Score is, how to calculate Actual CPC, and a handful of other questions that gives us a sense of their ability to understand supply/demand, logic, and the basics of behavioral economics. And we grill them with loads of follow up questions that challenge them to think through complicated topics on the spot.

By the time someone joins the team, they’ve already been challenged to think extremely critical about these topics.

Expectation management is not just something that is important for client relationships… It’s also a critical practice when it comes to onboarding new team members. New employees walk in on day one knowing that they will be challenged and will need to consistently develop new skills.

Automation, and how technology changes and impacts our client’s accounts, is something we are constantly discussing in the office. Diversity of opinion and experience drives these conversations, and we pride ourselves on having the type of culture that encourages open questions and debate from everyone on the team.

Can you walk through how you’ll eventually roll out automated ads + formats (SDC, etc)

To be completely honest, introducing new ads is not a top priority for anyone on my team. It’s time consuming and yields marginal or no value to our performance. I would much rather spend that time testing new copy on a landing page, developing display or video ads, searching for trends in Google Analytics, or doing just about anything else.

And with increased automation, it has even less value. We spend a lot of time writing ad copy for Responsive Search Ads, including multiple variations of each component, and we’re not able to see which of those headlines or descriptions work best (which would be helpful when testing copy on landing pages).

We have been testing new creative elements across the board and we have not seen a significant difference in performance... yet. There are certain instances where we see the smart-creative outperform the control, but my opinion is that we are still a few months away from these features really being powerful enough to make it worth 100% adoption.

We understand that smart-creative is the next frontier for Google, and I am willing to accept these changes with open arms. But we are not rushing to implement this across the board until there is more data and value that comes from this.

What can our team do to make your life even easier?

Well… since you’ve asked...

‍

More transparency with ad creatives, placements and search terms. Even if smart-campaigns outperform traditional campaigns, certain clients will push back if we are not completely transparent about these elements.

We also want more data about our users, ideally in the form of pivot tables where we can uncover insights about customer segments. These are the insights that trigger creative ideas and campaigns, which result in additional experimentation. Most clients would love the opportunity to spend an additional 10-15% if we uncovered an opportunity within one customer segment.

‍

More education about the technology. As mentioned, this was the most important factor that lead to our agency to adopting automation across the board, but most of the education was not originally provided by Google (it was either learned by other platforms or elsewhere).

Some topics that should be explored:

• How do the various algorithms work and what exactly are the reasons why we are migrating this way?
• What is the role of conversion data, and what are creative solutions that can be offered to accounts that just don’t see a ton of conversions (and please be more thoughtful than suggesting content-downloads as a micro-conversion).

As an aside, here are some resources that I personally recommend if you are interested in learning more about automation:

• Join or Die: It’s time to embrace Google automation (shameless plug)
• Artificial Intelligence for Marketing by Jim Sterne (buy it here)
• The Master Algorithm by Pedro Domingos (buy it here)

More insight into attribution and more training materials about attribution. Specifically, we need more data about time-lag, conversion paths and customer lifetime value.

It is also concerning that display / video campaign interactions are not factored into attribution reporting. It’s nearly impossible for client’s to understand the value of their prospecting campaigns (or remarketing campaigns at that) if we cannot easily draw this picture for them.

Admitting flaws and making recommendations to overcome flaws. These tools are not only imperfect, but they need certain conditions to exist in order to make them effective. If an account does not have a lot of conversion data, smart-bidding just won’t work. Google needs to be willing to work with advertisers to find creative solutions that help overcome these imperfections.

We’ve been experimenting with Smart Display campaigns for about a year, specifically the Pay-Per-Conversion feature. To be honest, I have never seen it work until recently, when our team came up with a really unique loophole that has allowed us to profitably run these kinds of campaigns for a handful of clients.

This revelation was conceptualized and developed in-house and without any instruction from Google. I’m really proud of this breakthrough (and others that our rockstar team has come up with over the years), but Google should be more proactive about offering these types of creative solutions.