ON-DEMAND RECORDING
Outdated on Arrival: How Fressnapf Moved from Static Marketing Measurement to Continuous Omnichannel Growth Steering
💡 Read the blog post with key takeaways: From One-Off MMM Studies to Always-On Budget Decisions: 7 Lessons from Fressnapf.
Stay tuned for our upcoming events
Webinars, masterclasses, and monthly measurement insights, delivered straight to your inbox.
Fressnapf's first marketing mix model was a one-off study. The team exported historical data, sent it to an external vendor, went through several rounds of validation and, months later, received a 130-slide PowerPoint. The analysis was valuable, but campaigns, budgets and even the organization had changed in the meantime. In this session, Maike von Runkel (Fressnapf) explains how the retailer moved to an always-on model, Friederike Thackwell (Google) sets out why finance needs different proof from marketing, and Juha Nuutinen (Sellforte) shows what the approach looks like in the numbers. They discuss:
Why timing matters as much as the model
- Friederike's weather app: useful because it tells you about today, not because it hands you a climate report from six months ago.
- Why finance treats marketing as a cost unless it gets proof it recognizes: online clicks and last-click conversions are the visible part, while the baseline that years of marketing built sits below the surface and is often assumed to be free.
- Why to optimize for the range where each additional euro still returns more than one, not for the highest average ROAS, and to combine marketing mix modeling, attribution and incrementality testing.
Fressnapf's journey from a one-off study to an always-on model
- Where they started: dashboards with campaign KPIs, brand tracking for awareness campaigns and last-click ecommerce conversions. Teams planned different parts of the customer journey with different KPIs, so budgets couldn't simply move from one area to another.
- The one-off study gave a broader view across channels, online and offline, for the first time. It also arrived after the business had moved on: the print leaflet it analyzed was already shifting to digital. "The analysis wasn't wrong, but it was answering yesterday's question."
- What an always-on model takes: senior management and finance on board, a data foundation with dedicated resources, and the people who run search, brand and store marketing, plus the agencies, involved from the start. "A new KPI doesn't automatically become the new truth just because you put it on a dashboard."
- Three views of one model: ecommerce, stores and omnichannel. A team with an ecommerce target keeps it and also sees what its investment does in stores.
The first decisions the model changed
- Budget moved between performance marketing channels, based on where the next euro generates more incremental revenue.
- Generic search, stopped because last click made it look unprofitable, was reactivated and given more budget.
- Brand search showed a more limited incremental ecommerce impact than last click suggested, and a much stronger impact on store sales than expected.
- The approach started in one market, moved to Germany and is being looked at for further markets. Maike was clear that not every team uses incremental KPIs yet: "The measurement is starting to influence what we actually do."
What it looks like in the numbers
- One currency for every channel: incremental sales and profit, online and in stores, so search, YouTube, TV and email sit in the same budget discussion.
- Start from numbers finance already agrees with. Juha's illustration: analytics sees 75 of ecommerce demand, actual ecommerce sales are 100, and with store sales the total is 300.
- Include the effect on stores: even campaigns that send people to the online shop drive a large share of their sales in stores.
- Look at the next euro, not the average: a channel with a high average return can be saturated, while one with a lower average still has room.
- Monday morning: incremental and marginal figures sit next to platform and analytics reporting. Scale while the next euro returns more than one euro of profit, make the change, check the following week, keep it or go back.
Q&A: adoption, timelines, finance and AI
- How to use a continuous model day to day: it doesn't mean changing the media mix every day. Platform data still runs day-to-day optimization; the model adds the incremental and omnichannel view for planning and for testing new things.
- How to get senior management on board: start from a real business problem and show real decisions and their results, not the tool or the model.
- How to drive adoption in the teams: involve them early, make results relevant to their daily work, and let them explore and challenge their own numbers.
- How long until the first useful numbers: "more months than weeks". Most of the time went into bringing ecommerce, point-of-sale, promotion and leaflet data together and agreeing what each number means. The next market is expected to be much faster.
- How it changed the relationship with finance: "We have a relationship now. Maybe that's the biggest change." The advice: involve finance early.
- How AI is influencing the work: Fressnapf is exploring it, and Juha described where Sellforte is heading: customers reaching their results from their own AI assistant.
00:00:02 – Welcome and housekeeping
Daria: Hello everyone, and welcome. I see a lot of participants joining us today, and I'm really happy to have you all here. My name is Daria, I'm on the marketing team at Sellforte, and I'll be your host for the housekeeping today.
We have a quick tradition at the beginning of every webinar: we'd like to know where you're joining from. Please tell us in the chat. Malmö, Sweden. Germany. Dallas, Texas. Barcelona. Helsinki. Espoo. I'm in Espoo, Finland, myself. Thank you for sharing.
We also checked the sign-ups for today's webinar, and it was amazing to see so many retailers and leading brands join us. Two housekeeping points. This session is recorded, so you won't miss anything, and you'll get the recording by email afterwards. We also have a live Q&A at the end of the session, so please go to the Q&A box and ask your questions as they come up; you don't have to wait until the end. Chat and reactions are open too, so please react to what our speakers are saying.
Some of you might have heard that this session was a blast at DMEXCO this September, and we're happy to host it one more time here. Friederike hosted it back in September, and she's hosting it again today. Over to you, Friederike.
00:03:01 – Why timing matters as much as the model
Friederike: Thank you, Daria, and hello, everyone. Let me start with this. Most of us probably checked the weather app this morning: one glance to see whether you need an umbrella for the day. Imagine if that app gave you a 100-page climate report from six months ago. Climate science is essential for the long term. Now let's move to marketing. Econometric models are the gold standard for our CFOs, to prove our baseline and guide our strategy. The problem isn't the model. It's the timing. You cannot steer this week's marketing or retail campaign with a six-month-old report. That's why we're here today.
I'm Friederike. As Daria said, I'm from Google, and joining me today are Maike von Runkel from Fressnapf and Juha from Sellforte. In case you're wondering why Google is part of this trio: at Google, we know that our campaigns are only as good as the measurement steering them. When retailers only measure online or last click, they miss the physical store impact and cannot steer budgets effectively and efficiently. That's why we work closely with independent measurement partners like Sellforte to help advertisers like Fressnapf turn econometrics into weekly steering, for both online and store sales.
Over the next hour, we want to show you three things. First, how to prove marketing's true commercial value to your CFO. That's my part. Second, Fressnapf's blueprint for connecting digital ads to physical in-store checkouts. And third, Juha will talk about the technology that makes weekly steering work in practice. As Daria said, please feel free to use the Zoom Q&A box anytime as we go.
00:05:19 – Why finance treats marketing as a cost
Friederike: Let's start with where the friction usually begins: your company's P&L. In almost every business, where does marketing sit? Under operating expenses, right next to office rent or electricity. And when times get tough, the CFO treats these expenses the same way: they minimize them, or they cut them. Compare that to your tech teams. When engineering asks for 10 million to migrate to the cloud, for instance, finance calls that strategic capex. They treat it as a long-term asset to scale the business.
Finance treats marketing differently for one simple reason: they need hard proof points. We cannot change accounting rules, so we have to give finance the measurement proof they actually trust.
Right now, we mostly show finance the tip of the iceberg: the short-term online clicks and last-click conversions that we see directly in our Google Ads or analytics dashboards. Underwater sits this massive baseline: brand reputation, customer habit, people walking into the stores. Finance quite often assumes that this baseline is free. I always compare that to riding a bicycle. If you stop pedaling, the bike still coasts forward for a while. It still moves. And if the CFO sees the bike moving and thinks we can stop pedaling because it's free, they forget the years of marketing that built the momentum behind that baseline.
When we analyzed over 80 European retail MMMs for the Effectiveness Equation, which we can also send you afterwards, the data proved it. In retail, 44% of marketing's total sales impact is hidden underwater, in that long-term momentum. If you only measure short-term clicks, you miss nearly half your value.
00:08:13 – Efficiency versus growth: the Autobahn
Friederike: This looks a little bit like complex calculus, but it's really just like driving a car on a highway. If your only goal is maximum fuel efficiency, it's a bit like crawling in the slow lane behind a truck, or driving 80 kilometers per hour on a German Autobahn. For the fuel gauge, it looks great, but you never really get anywhere on time. On the chart, that is the efficiency point: the blue peak at the left edge of the pink box. When you optimize for maximum average ROAS, efficiency looks heroic on paper, but revenue stalls.
The other extreme is flooring the gas pedal on the Autobahn until you drain the tank: 200 kilometers per hour. On the chart, that's the saturation point, to the right of the pink box, where the marginal return line drops below 1.
Smart drivers do what I usually do: I drive 130 kilometers per hour. You cruise in the fast lane at a steady highway speed. That's the pink contribution zone in the middle. Your average ROAS dips slightly from the peak, but every marginal euro still returns more than one, so you still capture maximum top-line revenue growth.
And what do you do when you reach the limit of that curve? You don't just throw more budget at it. You have to shift the entire curve upwards, and you can use two levers for that. The first is channel synergy. In our Google world, that means YouTube and Search working together: video builds the demand that Search captures, both online and in stores. The second is creative excellence: making ads people actually stop and watch. We saw in our research that great creative alone gives more than 70% higher return with the exact same budget.
00:10:41 – Where most advertisers are on measurement maturity
Friederike: Do you think many brands are actually doing this? Do they have the measurement set up to steer like this? The fact is, only very few. According to a BCG study, 76% of advertisers are still in the puppy phase of measurement maturity. Most are stuck in the emerging bucket you see here, with 42%. That's really just a polite way of saying we have a lot of dashboards, but very few answers. Only 9% are really leading. Those 9% are three times more likely to deliver top-quartile profit growth, because they do four things differently. They elevate measurement to the C-level. They tie learnings directly to their P&L goals. They test and experiment to isolate real value. Some teams, especially in retail, are often quite skeptical about modeling and models in general. But here's the secret in testing: when you run real-world experiments, you feed the results back into your models, and they get smarter and more accurate with every test. That's an important part that Juha will also speak about later. And finally, they calibrate their day-to-day tools against their findings.
That was a lot of theory. To show you what it really takes to build this in practice, let's move on to how Maike and Fressnapf did it in real life. Over to you, Maike.
00:12:38 – Fressnapf's starting point
Maike: Thank you. You said 76% of advertisers are still in the puppy stage of measurement maturity. I have to admit, when we started our journey at Fressnapf | Maxi Zoo, we weren't exactly skipping that stage either. So I'm not here to show you a perfect measurement setup today. I'm here to show you what the journey actually looked like, and still looks like.
Our omnichannel marketing measurement setup consisted of dashboards with campaign KPIs, brand tracking KPIs for awareness campaigns, and ecommerce conversions attributed using last click. Personally, for me as a performance marketer, that feels wonderfully simple. Someone clicked, someone bought. Great, we know who to thank. But obviously, that is not the full story. We couldn't properly answer whether these sales would have happened anyway, because we knew that platform attribution naturally favors lower-funnel interactions. We also had very limited visibility into the relationship between offline media and sales, and into how our online investments were contributing to what was happening in our stores.
And then there was the organizational problem. Different teams were planning different parts of the customer journey with different KPIs and different views of what success actually meant. Budgets often had different objectives attached to them. You couldn't simply move a euro from one area to another, because you weren't necessarily comparing the same thing. We were essentially optimizing what we could measure rather than what was actually driving the business.
That was the reason we started looking into media mix modeling, MMM, in the first place. We needed to move beyond the question of who got credit for a conversion and towards the question of what actually contributed to demand.
00:15:10 – The one-off study: answering yesterday's question
Maike: Our first MMM study was a significant step forward. For the first time, the results gave us a much broader perspective across channels and across online and offline effects. In principle, we had found a much better way of looking at marketing impact. The problem was that the business was moving faster than our measurement.
What we had conducted were one-off, static MMM studies. That means we exported historical data, sent it to an external vendor, went through several rounds of validation, waited for several months, and eventually received a very detailed analysis: literally a 130-slide PowerPoint. To be fair, it was very valuable. It helped us challenge assumptions that had developed from platform reporting and last-click attribution. But the business doesn't wait for the measurement cycle. By the time we got the recommendations, campaigns had changed, budgets had changed, new tactics had been introduced, and customer behavior had moved on. And just to make things even more interesting, during the analysis we also went through a major strategic reorganization. We had new leadership, new strategic priorities, and a significantly changed organizational setup. So while the MMM was analyzing our old reality, we were busy building a new one.
Let me give you one simple example: our leaflet strategy. Historically, the leaflet was a classic print vehicle. During the period our MMM was analyzing, we were already testing a shift towards digital. We were changing the format, the frequency, the geographical setup, and the partners we were working with. So by the end of the analysis, the MMM could tell us quite precisely what had worked for the old print setup. The slightly awkward part: we weren't really doing that anymore. The reaction from our marketing manager was basically, "Good to know, but what am I supposed to do with this now?" And that was the real issue. The analysis wasn't wrong, but it was answering yesterday's question. We had a measurement system that was very good at explaining the past, but not at helping us steer the future.
00:18:02 – From a study to an operating model
Maike: That changed what we were looking for. We didn't want to buy another MMM study. We needed measurement to become part of the ongoing marketing operating model. Instead of measuring, analyzing, and reporting once, we wanted to continuously learn from new data, use those insights to make decisions, and then measure again. In other words, measurement should become part of how we run the business, not a reporting exercise.
Saying that now sounds fairly straightforward, which of course it wasn't. This is probably one of the biggest things we underestimated at the beginning. Continuous MMM is not simply a technology implementation. It is not a marketing project. The model itself is only one part.
First, you need senior management and finance on board, because if measurement is supposed to influence budget allocation, people need to trust it as a business steering mechanism and not as another marketing report. Second, it is a data project. You will need data engineering, analytics, and a solid data foundation, and you need dedicated resources for that. Trust me on that. Third, and this is maybe the most important one, you need the people who actually run marketing to be involved. Your SEA manager, brand marketing, your store marketing team, and your agencies all bring knowledge that a statistical model simply doesn't have. Marketing experts know when there was a tracking issue, why there was a sudden spike, when a campaign was unusually strong, or when something happened in the business that the data alone can show but not explain. They know the customer, and they know the business context.
But you also need to get these experts to trust a statistical model, especially when that model sometimes tells them something that goes against what they have learned from years of platform reporting and last-click optimization. You can't introduce a new KPI and expect everyone to celebrate. If someone has optimized a channel for years based on platform ROAS, and you suddenly tell them that incremental ROAS is the new KPI, that changes how their performance is evaluated. A new KPI doesn't automatically become the new truth just because you put it on a dashboard. So we had to treat this as a business transformation, not as buying a new tool. Ultimately, it's about creating one common view of marketing impact across the whole organization.
00:21:15 – Starting with existing decisions: three views of one model
Maike: Once we understood that, we also realized that we didn't have to change everything at once. We started with the decisions we already make today. We didn't try to transform all of marketing overnight. Instead, we took existing planning and optimization processes and gradually added a different perspective: the incremental business impact.
There is another important part of how we approached this. We didn't immediately say, "From now on, we are only optimizing omnichannel." That would have been too far away from how our business and our budgets are actually managed today. Instead, we built three perspectives: ecommerce, POS (stores), and omnichannel. That means, for example, that if a team has an ecommerce target today, we can still look at ecommerce impact while at the same time seeing what that investment is doing for our stores. Over time, that helps us move towards a truly omnichannel way of steering the business, rather than trying to force that change overnight.
00:22:48 – The first decisions the model changed
Maike: These are some of the practical examples we started with in Poland, our MVP: using the decisions we already make, and adding incremental revenue and incremental ROAS to the equation.
First, budget allocation. We can now look at our performance marketing investments and ask: where can the next euro generate more incremental revenue, and where can we reduce spend if the incremental impact is low? That has already led to budget shifts between performance marketing channels.
Second, challenging assumptions and testing again. Generic search is a good example. The team had initially stopped generic search because last click made it look unprofitable. But the model showed that it was actually generating incremental demand, which led us to reactivate campaigns and gradually allocate even more budget to it.
Third, seeing impact that we simply couldn't see before. SEA brand is a good example here. Historically, the assumption was often that brand search is mainly online navigation: people already want us. And the incremental ecommerce impact was indeed much more limited than last-click attribution suggests, but the impact on POS sales was much stronger than we expected. So suddenly we had a completely different picture of the role of brand search: understanding what that investment contributes to the total business, and that it shouldn't be seen as an ecommerce performance marketing instrument only.
None of these are huge transformations of the entire marketing organization, and that's exactly the point. We started in Poland, we followed up by bringing the approach to Germany, our biggest market, and we are now looking at further markets. We are also very much still in the process. We haven't magically broken down all our silos. Not everyone is using incremental KPIs yet, and we still have work to do to get all relevant teams fully involved. But the important shift has happened. The measurement is starting to influence what we actually do. It is no longer the destination; it is part of our growth loop. Sellforte has been a big part of this journey with us, so over to you, Juha.
00:25:45 – What Fressnapf does well: five things
Juha: Thank you so much, Maike. Nice to meet you, everybody. My name is Juha Nuutinen. I'm the CEO and one of the co-founders of Sellforte. Sellforte helps advertisers measure the incremental impact of marketing and then turn those insights into budget and campaign decisions. By the way, Friederike, I really loved your Autobahn example. It's very dangerous to drive 80 kilometers per hour on a German Autobahn.
Friederike: That's true. That is more dangerous than 200.
Juha: I think so. And Maike, the examples on generic search, bringing it back, and understanding the real role of brand search were really amazing. Thank you so much for sharing these practical examples.
Maike explained the journey. I'd like to make it a bit more concrete and show, from Sellforte's perspective, what Fressnapf is doing extremely well. I think it's five main things. First, measuring all of the channels together in one unified measurement solution. Second, using sales and profit numbers that finance agrees with. Third, including the impact on physical stores. For an omnichannel retailer, this is huge. Fourth, understanding the return on the next euro, or the last euro, you spend on paid media, not the average. Bringing marginal incrementality into the discussion is very important. And finally, turning all of this into recommendations quickly enough, not with a six-month delay, so they become part of the active steering of marketing, instead of confirming something we believed was true six months ago, when all the changes have already been made and we cannot affect those decisions anymore. Really making it relevant for day-to-day decision-making.
00:28:17 – One currency for every channel
Juha: Starting with number one. Very often, performance marketing is measured with conversions or conversion value, brand marketing with reach, and offline media sometimes with reach, sometimes with something else entirely. So how do you move budget between these channels? They are not comparable. If you don't have a single currency where you can actually compare these channels, you cannot confidently move budgets between them. You need one common currency to value them.
The top row is "before", where most retailers are today: only digital performance marketing is measured, against ecommerce sales. The bottom row is where Fressnapf is today and where Fressnapf aspires to go: estimating the incremental sales and profit generated by each channel, both online and in stores. Every channel is included and made comparable by one unified metric. We want search, YouTube, TV, and email to sit in the same budget discussion.
00:29:50 – Start from numbers finance agrees with
Juha: Second, Fressnapf did amazingly well by starting from sales and profit numbers that the finance team already agrees with and recognizes. This is an illustration. On the left, GA4 sees only 75 euros of demand, but the actual ecommerce sales, including the people who did not consent to cookies, are 100. That is the consent and tracking gap between the 75 and the 100. Then, when we add the physical store sales, which for an omnichannel retailer can be two or three times as much as ecommerce sales, you come to 300. So how are we able to steer our marketing budget if the marketing team only sees this 75 of last-click revenue? It's not a good place to be. It's simply not a good metric to steer the marketing budget with, and we should not steer with it.
Our recommendation to our customers is always to start with the company's actual total sales and profit, across ecommerce, stores, and all the other channels, so that the starting point is numbers that finance and controlling already agree with. And then estimate how much marketing contributes to it.
00:31:20 – Include the effect on stores
Juha: The next slide shows why this is important. Especially for an omnichannel retailer, the halo effect on stores is really big. Our Shopping and Performance Max campaigns, for example, direct customers to buy online, and we measure the conversion value from ecommerce purchases. But even though they direct people online, still something like one-third, plus 30%, sometimes plus 50% of the sales happens in our stores. So if we only look at GA4 and ecommerce sales, we are blind to something like one-third of the impact of those channels. And that is Shopping and PMax. For generic Google Search, the halo effect on stores is even bigger. For paid video, YouTube, the majority of the effect very often comes from stores, from offline sales. If we optimize our Google marketing and digital media only on ecommerce sales, we are blind to so much of the effect, and we might be making many wrong decisions.
An example for Performance Max, which most people think of as a performance channel targeting ecommerce sales. We start from the number reported by the ad platform, 7.5. You might see in Google Ads that the ROAS of PMax is 7.5. GA last click might be telling you, no, it's 5. And when you model only the incremental sales in your ecommerce sales channel, you might end up with 4. So 4 is the incremental ROAS of your PMax, based on your incrementality testing and your MMM. But even this misses the important point of offline store sales. In our experience, on average, if your incremental ROAS in ecommerce is 4, you need to add 1.6 on top to get the real incremental ROAS on the overall level, including your store sales.
The really dangerous thing in reporting ad platform numbers is that your finance team might believe them. They see that your Google Search brand is generating a ROAS of 30, and they ask, "Why are we doing generic Google Search if the ROAS is only 7.5?" And you might get your generic search cut altogether. I don't even know which is worse: that finance doesn't believe your numbers, or that finance believes your numbers. But trust me, the best thing is that you show finance numbers that are true, and finance believes them. That is where you want to get to.
00:35:17 – Look at the next euro, not the average
Juha: We were talking about incremental ROAS just now, but what you could argue is even more important is the marginal return on the last euro you are investing. It's the saturation Friederike mentioned with the German Autobahn at 200 kilometers per hour: even if you push the pedal to the floor, you are not going much faster, but you are burning gasoline at a very high rate.
Again, these numbers are completely imaginary, just examples. Brand search can have a very strong average return, let's say 6, because it also drives people to stores. But if you already win 90% of your brand searches, measured by impression share or click share, there is no room left to win more, even if you double or triple the budget. On the other hand, you might have YouTube, paid video, on the right. It doesn't necessarily give you the best average incremental ROAS, but it might still be the channel where you get the best return on the last euro invested, because you have not yet saturated it. You have not yet reached all of your customers, and you might still have a lot of very cost-effective reach left in your paid video channels. That's why you need to look at the marginal return on the last euro invested, not the average return.
As a summary: digital media drives online sales and store sales. Offline media also drives online sales and store sales. So there are four effects. Click attribution shows you only the bottom-left box: the effect of digital media on ecommerce. For a typical omnichannel retailer, that is only one-fourth of the whole picture. The unified measurement that Fressnapf has adopted covers all four. You want the whole picture visible before spending a single euro.
00:37:57 – What changes on Monday morning
Juha: My last slide. What changes for a marketer on Monday morning? First of all, it's a change altogether if you get the weather report for the current week on Monday morning. That's a positive change already. Next to ad platform ROAS and GA4 ROAS, you also get incremental ROAS and marginal incremental ROAS for all of your channels. It might be that the channel that looked like one of the worst based on ad platform ROAS, Demand Gen for example, is actually the best channel when you measure it on marginal incremental profit on ad spend. And the other way around: brand search might look like the winner, the star, if you only look at ad platform ROAS or GA4 last click, but based on marginal incremental ROAS or marginal incremental profit on ad spend, it might give you the lowest return on the last euro spent.
So this is in addition to, not a replacement for, ad platform ROAS and GA4. They are still amazing tools. But we also want to make incrementality and marginal incremental reporting part of everyday decision-making. The rule is simple: scale as long as the next euro brings back more than one euro of profit. Then activate the change in the ad platform and check next week whether the forecast held. If it didn't, go back. If it did, keep the change and continue learning. That's the message I wanted to give: measurement is a tool for learning, and the faster we learn, the better. Thank you so much for listening. Back to you, Friederike.
00:40:25 – Three things to take home
Friederike: Thank you, Juha. I think this Monday morning table says it all. Once you include physical store sales, your upper-funnel campaigns, like Demand Gen here, but also video campaigns, look completely different. We also tested that with Sellforte across 35 major retailers. For every euro spent on YouTube, retailers generated an incredible €5.70 in incremental sales, and €4.40 of that happened in physical stores, at the physical checkout. So if you only measure online clicks, you miss nearly 80% of what video does for your business in this case. That's also why Google has an interest in this.
We're almost at the end of our webinar and heading over to the Q&A, but there are three things we want you to take home today. Use the numbers your CFO actually believes. Juha mentioned that at the very end, and I think it's a really important message, because we are all used to talking in marketing KPIs and marketing language. But if we don't talk to the CFO in the numbers they believe, we will never be able to move marketing towards being a profit engine. Include your physical stores, in the case of Fressnapf and other omnichannel retailers. And update those numbers fast enough to steer Monday morning's campaigns.
We now have 10 to 15 minutes for Q&A, so please feel free to use the Q&A box.
00:42:31 – Q&A: How do you use a continuous model day to day?
Friederike: There is a question in there; I'll read it out loud. "The day-to-day MMM outputs wouldn't change drastically compared to the attribution models. Then how do you use continuous MMM for daily inputs?" We had a similar question ourselves: what happens on a Monday morning when the model says something different from what a channel manager sees in the ad platform? Juha, Maike, do you want to take that question?
Maike: Maybe I can start. I think the point is that using continuous MMM doesn't mean that we need to change our media mix every single day. The value is that we regularly get an updated view of incremental impact and omnichannel impact, rather than relying on last click or platform attribution data alone. For day-to-day optimization, we still use platform data and the usual KPIs. But the MMM adds the broader business perspective, and it helps us make better-informed planning decisions, especially when testing new things or when something out of the ordinary happens.
Juha: Completely agree. Like I tried to say, we don't want to replace what the ad platform is telling you. We don't want to replace what GA4 or Adobe last click is telling you. We want to bring additional data. We want to incrementality-correct what the ad platform reports and what your last-click measurement is telling you. The change we want is that you don't need to give away the granularity and the fast-paced updates of what your ad platform or your measurement told you yesterday. We want to incrementality-correct it, so you can make the right decisions based on your most granular platform reporting and last-click measurement. We want to amend your current measurement, not overhaul it.
Friederike: Thank you both. From a Google perspective, I'm not sure if you've all seen it, but we have this measurement trifecta. We cannot rely on attribution alone, and since GDPR even less: attribution breaks quite often and is not able to track everything. So we recommend this measurement trifecta, or triangle, of MMM, attribution, and incrementality testing, to get a holistic picture of your measurement activities.
00:45:50 – Q&A: How do you get senior management on board?
Friederike: The second question: "If Fressnapf's success was driven by treating MMM as a business transformation, what can you do to drive this kind of mindset in the organization? How do you get this level of buy-in from the C-level?"
Maike: I would focus on the business value, not the technology. Show how better measurement can support better investment decisions and ultimately drive growth. Start with a real business problem, get the stakeholders involved, and demonstrate the value. We're actually right in the middle of that. Show real decisions and KPIs and what the outcomes were, not just the tool, and don't just talk about the model.
00:46:45 – Q&A: How do you drive adoption across the organization?
Friederike: There's a related question just coming in: "How do you drive this kind of measurement and new thinking adoption across the organization? Any tips on where to start to get stakeholder commitment?" I think that's a very good question, because there are usually so many stakeholders involved.
Juha: I'm totally with Maike on this one. I have this shirt for a reason: I love impact. I want the measurement to provide a concrete, measurable increase in incremental sales and profit after taking these decisions. From my perspective, we need to show the real money to the business stakeholders. If we can concretely show that we made these decisions and, as a result, not only did our incremental ROAS increase, but we got a concrete increase in October and November total sales and profit, that's the best proof you can show. And that gives you the mandate for the next level of changes. So I would say start small. Make the changes where you feel 100% sure, then measure the effect and prove to the business stakeholders and the C-level what you achieved, so you have the mandate to continue to your next level of changes.
The big difference with continuous measurement is that you can make the change, measure the effect, and learn from it. The quicker you can do it, the more you can learn, and the more you can learn, the more impact you can make. This is a massive difference compared to annual marketing mix modeling PowerPoints that come six months later, once a year. The pace of learning is too slow.
Maike: Yes, slow is a good word for it. Moving on from the C-level to the actual teams that have to work with the tool, I think it also helps to involve the local teams, or the teams in general, early, and make the results relevant to their day-to-day challenges. Let them explore their own numbers, understand the model and the insights, and then let them challenge what they see. That creates much more ownership than simply presenting them with a new tool and the results. I would also identify a few concrete use cases where the MMM can challenge and improve budget decisions. That's true for all teams. Once people see the value in practice, they will adopt the tool.
00:50:08 – Q&A: How long until the first useful numbers?
Friederike: A quick question for Maike that we also got from our DMEXCO audience: how long did it take you to get the first numbers? That's often a hesitation I hear. MMM is known as this heavy thing that takes so long to set up. From your experience, how long did it take to get actually useful numbers?
Maike: A bit longer than we had hoped for. But I don't think it was really the model or the tool in this case. It was our data. We had to bring together all these data sources, ecommerce, point of sale, store data, promotions, the leaflet, and then agree with each other on what each number actually means. That sounds easy, but it is not when you're actually doing it. So it's more months than weeks. It depends on the size of your company and how many data sources you have, obviously. The good news is that expanding to the next market is going to be much faster than the first one. The first one is the MVP.
Friederike: That's probably your experience as well, Juha, working with so many companies. The most time-consuming bit is probably the data collection.
Juha: Yes, I think it's typically the sales data, if it's custom. Sometimes just getting the ticket for the sales data integration through the IT organization can take four weeks. Typically I say eight weeks from connecting the data to having a regularly updating MMM that you can really trust. Eight weeks is the number, but it can sometimes be longer if we need to do a lot of data improvements and data gathering for offline spend data like TV, out-of-home, and radio, or a lot of processing for the sales data. Then it can be 12 weeks.
00:52:45 – Q&A: How has this changed the relationship with finance?
Friederike: Another question, for Maike: "Great talks and a very inspiring journey. How has this transformation changed your relationship with finance, and how do you work together? Any tips for marketing and analytics leaders?"
Maike: We have a relationship now. Maybe that's the biggest change.
Friederike: That's a good start.
Maike: It's not about finance telling marketing how to spend its budget, or marketing having to justify every single channel. Maybe "relationship" really is the right word. We're developing a shared understanding, not working against each other, and speaking with the same voice about what drives business results. The advice is definitely to involve finance early. They usually don't take part in a lot of these decisions, but they're very important.
Juha: I always say that finance people are very smart. The controlling people know the business so well. Their knowledge comes from so many years of working, operating, making decisions, and seeing the outcomes. I really admire the finance and controlling people working with marketing teams, and I have a lot of respect for them. My feeling is that when you finally bring them numbers they can use for next year's budget planning, where they can feel comfortable that "if I spend this much on this specific channel, I will get an incremental ROAS of 4 or 5", they are very grateful, because they can build their budgets with trustworthy data. They don't need to correct it: "The marketing team tells me it's 20, so I need to divide this number by four to get something I can feel comfortable with." They are smart, and I feel great working with those people.
00:55:24 – Final question: How is AI influencing the work?
Friederike: We have five minutes left. Thanks for joining us. I just want to say this is a great partnership with Fressnapf and with Sellforte. I personally learned a lot from looking behind the scenes, beyond just Google reporting. Thank you for sharing that with us. Back to you, Daria.
Daria: Thank you very much. It was an amazing session. Thank you, Maike, Friederike, and Juha. Judging by the reactions, the audience really loved it. And thank you especially for the questions. We still have time for a last-minute question, and anything else we can answer in a follow-up. Actually, we have a final one.
Friederike: The final one: "How is AI influencing your MMM transformation at Fressnapf?"
Maike: I think that has more to do with the maturity of using AI at Fressnapf. As a big company, we're probably not the first movers in using AI, especially when it comes to sharing data. But we actually talked about this just this morning: the possibilities of using MCP servers, and of using Claude, Copilot, or ChatGPT as a means to bring the data directly into your AI and then work with it. Sellforte also offers the possibility to use AI, which we are exploring.
Juha: Really great question. My dream is that all of our customers use the insights from Sellforte via their Claude, ChatGPT, or Gemini. They don't need to go to the user interface if they don't want to. To be honest, I don't visit user interfaces very often nowadays, even though they update with daily data. I'm working with my Claude all the time. I want the data to be there, and I don't want to transfer CSVs back and forth. I want my Claude to have access to all the data streams I work with. This is not a paid commercial, but we are giving the MCP for free to all of our current customers, to help them get access to their own marketing data more easily via their Claude. And we are working on it with Maike as well, hopefully getting this working very soon. But I know big companies have diligent processes.
Friederike: I love that. I think it's such a great opportunity, because everybody can ask questions at their own level. Once you have it integrated with AI, there is really no stupid question to ask, and it gives you that predictive power for your own data.
00:59:00 – Wrap-up
Daria: Love it. A last-minute question, but such an important one. Thank you very much again for your questions. If you have any others, just reply to the email you'll receive soon with the recording and the key takeaways. When you leave the webinar, a short survey will pop up; please let us know how you liked the session. We really appreciate your feedback. See you at the upcoming events, thank you again for joining us, and have an amazing rest of the day.
