This article has also been published on Marketingfacts.nl
Marco Frighetto will be speaking at the anniversary edition of Friends of Search on March 23rd in de Kromhouthal in Amsterdam. Tickets available at shop.friendsofsearch.com.
- What does a day as the Head of MarTech at your agency, Booster Box, look like?
It can be difficult to describe an average day since they tend to vary, but there are some commonalities among them. Most of the time is invested in four main areas.
Firstly, there are internal meetings to discuss the progress or status of specific projects and to try to remove any roadblocks. While most of what we do has already been standardised and proven, we might encounter roadblocks that depend on a particular type of innovation or solution that we are trying to implement. Therefore, we need to seek the best approach possible together with the rest of the team.
Another important part of the job involves meeting with clients to propose new solutions. A lot of the work here involves trying to explain complex technology in a simple and direct manner and emphasising the business advantages.
The third aspect of the job is checking on the project roadmap. Much has already been decided, projects are moving forward, and we need to keep track of everything. We have a strategy at the beginning of the year and run weekly and monthly check-ups to ensure that we are on track with delivery.
Lastly, it’s important to keep up to date with all the news in the industry. This involves keeping up with the zillions of sources available and establishing a network along the way. This is probably one of the most important aspects of the job.
- What ‘SUPER SCI-FI TECH’ that you have been working on are you most proud of?
The ecommerce package related tech is the one I’m most proud of, and it has delivered a lot of value to our clients. It consists of three market products:
The first is classically defined as profit bidding and internally we call it MOAS (Margin on Advertising Spend). Our technology is actually capable of tracking the profit and the margin of each item, and then importing this back into the advertising platform. The point here is not just the technology per se, but also the procedure. The goal is to change the north star metrics for a business from what was before the ROAS, to now MOAS.
Then, the second point in this package is the product clustering, what we call “Galileo”. It has massive impacts in terms of performance because it allows you to dynamically move around and locate different products across different campaigns.
Lastly, there is also our internally developed feed management tool that gives us the possibility to manage huge inventory and to create specific feeds for a client.
- What do you think has changed in PPC management over time, and what has become the most important part?
I think there have been a few important changes over the years, but the main driver has been the shift from automation to machine learning. Let me explain. Five or six years ago, the key to scaling and delivering better work for clients was the introduction of automation in creating and managing campaigns. I still remember the popular SKAG structure – single keyword per ad group. But managing that complexity required a layer of automation.
Today, the focus has shifted. What’s making a difference is having a solid rework of the data shared with the advertising platform and properly modelling relevant business information. The input shared with the platform is what makes the difference. This change has drastically altered our approach over time. We concentrate on providing data insights that change strategy and need to be implemented at the platform level.
This shift brings me to two important points. First, it’s essential to work in compliance with privacy regulations when treating and sharing data with third-party providers. You must be aware of what’s possible and what isn’t, and what the limitations are. Second, it’s critical to collect and work with CRM data, or first-party data more generally. This has been a significant change compared to the PPC scenario of just a few years ago.
- What are the most successful PPC use cases you’ve seen in terms of industry, audiences, creativity, and so on?
I would like to share with you a couple of case studies.
- Satispay, a mobile payment app founded in Italy, aimed to increase brand awareness among Gen Z in Italy and evaluate the impact of TikTok advertising. We proposed using incrementality testing to compare data between a treatment group (with access to TikTok ads) and a control group (without TikTok ads) to determine the difference. The results showed that TikTok generated 434 incremental app registrations with a 50% reduction in cost per action over the target. This also demonstrated the effectiveness of incrementality testing in analysing advertising effort and suggested how alternative measurement frameworks, especially for Top of Funnel activities, can be extremely valuable in assessing the return on marketing effort. (Here)
- The other case study, instead, is coming from the colleagues at Precis Digital. A Danish ecommerce was struggling with declining growth in Search, while their Shopping was doing well. The digital strategist from Precis decided to try a new approach to search campaigns to regain market position and growth by leveraging machine learning. They used data aggregation based on value per click to group keywords, automated ad group and campaign level search query analysis to limit wasteful spend, and tailored campaign-specific audiences to feed the smart-bidding algorithm. This approach proved to double the revenue while keeping a stable ROAS. (Here)
- Have you fully embraced Performance Max campaigns? Why (not)?
This is a funny question. Do you really think there is a choice here? No, I’m kidding. I will say yes, this is something that we are using and leveraging to improve performance. However, we are approaching it with a critical mindset and not taking everything that comes from Google for granted. I believe that there is no real choice but to move in the direction that Google is taking. You have two possibilities: either fully challenge it or try to understand what can be done based on the setup and framework that Google is sharing with you. Initially, when the technology was still immature, we did not fully embrace it. But now, after running some tests, we have seen that with a few tweaks and common sense, PMax is delivering good results.
A couple of best practices that I see here, especially for advanced eCommerce, is to limit the delivery of PMax campaigns only to shopping assets, for example. What we are doing is trying to recreate a bit of what used to be the classic smart shopping approach, still with the possibility of more powerful targeting that comes from this type of campaign.
Nevertheless, I still believe that testing is king. The approach that we are following when it comes to this kind of shift is not to fully step in, but to be conscious of the downsides and advantages and define concrete use cases to test this out. This was exactly the case with shopping campaigns on our end.
- Do you have any best practices you can share when it comes to pMax?
Yes, I have a couple of inputs. Probably the most powerful one that we are seeing, especially for ecommerce, is having a specific approach to the way in which you create the structure of the PMax campaigns at the account level. The core idea is to segment products into different buckets and replicate the same structure at the account level by matching those buckets with different campaigns optimised against different performance targets. This way, the performance improves, and you have more granular control over the segmentation.
Another best practice is to tailor your copy and creative for specific asset groups. Of course, the asset group is the first entity that defines the targeting. If you improve the quality of the creatives and copy based on user research and link these elements to the user’s actual search queries, you can potentially drive better performance, increase the CTR, and ultimately, the conversion rate.
Finally, something that we have started using more recently is audience signals. We try to include first-party data, CRM audiences, and match them to the asset group, so the campaign points in a clearer direction.
- Should we be sharing all our first-party data with Google?
“All” is a strong word when it comes to sharing first-party data. Generally speaking, it makes sense to share this data as it can be a point of differentiation for your digital marketing strategy and a competitive advantage. However, there are two potential risks that need to be addressed from the outset.
Secondly, there is a business risk, particularly for big advertisers like Amazon. These companies may consider carefully what they want to share with players like Google or Meta, which may be close to their industry.
Sharing first-party data can boost and differentiate your strategy because only you know your margin and have visibility on LTV. If you can feed the platform with this kind of information, it can also help Google to deliver better results according to your objectives.
- As an advertiser, how can I share my first party data with Google and what are the benefits?
In terms of the possibility of sharing first-party data, I think there are two main ways: offline conversion (and we can distinguish now two different ways of conversions – I’m going to dig into that in a moment) and the classic audience part. For the second, it’s pretty straightforward. You upload audience information, and the essential information is just the email, but you can also add other information like the phone number, geo information, and so on. This is something that you can do directly with the platform. A fun fact is that usually, the match rates tend to be higher if you’re passing only the email address rather than linking it to other data points.
Then there is the offline conversion: the information you share is more related to the conversion value for the specific user’s actions. Until a few months ago, it was only possible to do this via the upload of the gclid code (the click ID code connected to the specific conversion). Now, the so-called enhanced conversion for leads is giving you the possibility to upload that data in a slightly different manner to share visibility with Google on the offline journey of your user. We are going to dive more into this during the speech, so I don’t want to spoil everything here 🙂
In terms of benefits, as I was saying before, first-party data is a real superpower for advertisers in the current time and it’s going to be even more so in the future. To give you an example of how this actually makes a difference, think about what we used to do a few years ago. For example, at Booster Box, we worked a lot on defining the best bidding model possible. We were trying to tweak this kind of input to make the campaigns work better. This method was outperforming until probably three years ago, and then machine learning came in and became better and better. Now what we are focusing on is not just the ability to automate the bidding strategy, but it’s actually the output that you can collect. This sits with the capability of working on your first-party data and correctly sharing this information with the platform, which then does the hard work and starts looking for the users that are more likely to deliver according to the target and the performance that you want. So this is the straightforward benefit. This is the superpower because it differentiates what you do from your competitors. And this is something that cannot be easily replicated by the competition.
- What can people expect from your session at Friends of Search?
As you may have gathered from the rest of the interview, my session will focus heavily on first-party data and its integration. I will offer practical advice and suggestions on the strategies and implementations you can use to leverage those insights, as well as some of the downsides and loopholes we’ve encountered along the way. I will also provide a strategic overview on how to approach this type of data and offer practical examples, such as how to store UTM and click ID information into a CRM using Google Tag Manager. It’s going to be a very nerdy session, so apologies in advance, but at Booster Box we love that kind of stuff! 🙂 I’m excited to join you in Amsterdam!