Latest / Inside Commerce: Ecommerce Strategy, CX and Technology Podcast / EP296: Adding Intelligence to Your Marketing Mix; Balancing Sales Today With Future Growth & Profit, With DEMA Co-Founder Henrik Hoffman Kraft
Transcript
- 0:00And that is really where the marketing mix modelling is being
- 0:02that sober guidance on your shoulder and giving you the
- 0:07number to discuss with the Chief financial officer, that is the
- 0:12person that you need to talk to, right?
- 0:15The CFO needs to see the numbers.
- 0:18The CFO in Indeed to see retail world is not a person who don't
- 0:23believe in brand because then they then they would leave that
- 0:26whole business, right? But if they don't see any proof
- 0:30of, you know, your marketing spend actually providing any
- 0:33value, they're just going to say cut it.
- 0:37You listen to Inside Commerce, a weekly E Comm strategy,
- 0:40technology and CX focus podcast hosted by experienced digital
- 0:43consultants and re platform specialist James Goett and Paul
- 0:45Rogers. Get insights into E Comm
- 0:47strategy, leading brands, hear from industry thought leaders
- 0:49and leading tech vendors with practical advice to help you
- 0:51improve your online CX. This episode is sponsored by
- 0:57Shopify. Another important topic today.
- 0:59We are talking about adding intelligence into your marketing
- 1:02mix, balancing the sales today view with future growth and
- 1:04profit. And really happy to have
- 1:07somebody who's got true expertise in this topic matter.
- 1:10I'll let him do his founder story in a minute, but warm
- 1:13welcome to Henrik, who's the founder of Dima e-commerce
- 1:15intelligence platform. How you doing Henrik?
- 1:17Thank you. Yeah, I'm doing fine.
- 1:19Sun is shining in Stockholm today.
- 1:22Nice nice thank you for joining us from Sunny Sweden and for
- 1:25those listening who haven't come across demo before, I'm not
- 1:30quite sure where it fits into the E com tech stack.
- 1:33I'll I'll let Henrik explain his background and what the platform
- 1:37does in a minute. But the context of what we're
- 1:38talking about today is I think a lot of brands can resonate with
- 1:41this. It's there is constant trade
- 1:43impression, especially the last couple of years, it's quite
- 1:45intense. We've got all these
- 1:46macroeconomic factors causing loads and loads of turbulence.
- 1:50There's there's cost squeezes going on left, right and centre.
- 1:54There's massive pressure to deliver on sales.
- 1:56And that brings a sales today mentality.
- 1:57I was speaking to a few of my clients recently who are really
- 2:01frustrated of how they're stuck in a like discounting cycle
- 2:04where they're constantly going after bottom funnel, but it's
- 2:06compromising their, their kind of other programs like their
- 2:09loyalty strategies. And it's also making them worry
- 2:12about brand positioning and, and brand growth.
- 2:15And, and how do they distinguish from ADC point of view if
- 2:18they're constant discounts and whilst all their retail channels
- 2:20are etcetera. And then how do you balance the
- 2:22need to deliver numbers with that more long term piece?
- 2:25So we're going to come on to that and that's why we want to
- 2:26talk about it because the big concept stays around marketing,
- 2:29mixed modelling and what role and value that plays and how you
- 2:34can really approach that to balance those twos and start
- 2:37delivering results through through intelligent data.
- 2:40So that's my Henry, that's my kind of woefully introduction.
- 2:43Let's go to to you because I think your founder story is
- 2:45really interesting. You want to give our listeners A
- 2:47flavour for who you are and how you got to launch in Daemon.
- 2:51Cool. Yeah, Thanks.
- 2:53Happy, happy to do that. And thanks for for for the
- 2:57intro. So yeah, my name is Henrik
- 3:00Hoffmancraft, 36 year old from a small town outside of Stockholm.
- 3:06My parents were both entrepreneurs, mom and pop type
- 3:12of companies always been like in that mindset of kind of being
- 3:19the business. And that is really following all
- 3:22the way to where I'm at, where I am today and kind of what we do
- 3:27at the Emma, to be frank, I studied business and after a
- 3:31while I, I, I landed in the, in the e-commerce group.
- 3:37We were actually we're having, we had like 16 stores as well,
- 3:40but we saw ourselves as an e-commerce group.
- 3:43We were around €60 million in revenue when I joined.
- 3:49We were like 1 and €20 million in revenue when I left.
- 3:53I joined as management trainee, worked side by side with one of
- 3:57the Co founders of that group that is now my Co founder today,
- 4:01Marcus. I did everything.
- 4:05We moved warehouse from like a manually pick and pack to super
- 4:10like automated, the best in breed type of thing.
- 4:14We did all kind of nice projects back then like acquiring
- 4:19companies, etcetera, etcetera. Then I started really being part
- 4:26of the scaling journey of the baby shop group as heading the
- 4:29retention and I ended up doing a lot of general growth
- 4:34initiatives in the group as well.
- 4:38But then I left after a while, went to the dark side of being
- 4:43an agent, like a as a consultant at the marketing agency company.
- 4:53But I struggled so much of being like outside of my clients
- 5:00businesses not having any data at all that was actually driving
- 5:06their long term growth, sustainable growth like building
- 5:13that healthy business. I was just one of those robust
- 5:18hunters, to be frank. I asked for more data like I
- 5:22wanted the inventory data, I wanted the customer data, I
- 5:25wanted the, the fulfilment cost, the COGS, the, the, the full,
- 5:31you know, contribution margin metrics, etcetera, to really
- 5:37align not only with my counterpart, let's say the head
- 5:41of E com, but actually the full company.
- 5:45So not only the way of course, then deliver Rowas, but that was
- 5:50like almost toxic revenue, like we weren't the line that we got
- 5:54kicked out like all agencies like do right.
- 5:59And I I found that so frustrating.
- 6:02And then I met up with Marcus and David, my my current Co
- 6:08founders of, of Dima and they've all, they both also left baby
- 6:13shop group and we did like 2-3 years apart and we met up and we
- 6:19were like, you guys see what I see?
- 6:22And we were like in different words explaining the same thing
- 6:25because what we did at Baby Shop Group the last years were before
- 6:29we left was actually building really solid data warehouse,
- 6:34like great ETL like structure. We were profit bidding on a
- 6:38contribution margin and profit LTV.
- 6:40Like we were really advanced in the end.
- 6:44So, so we just like, OK, guys, we we need to build something
- 6:49here. It's obviously a need for this
- 6:51in the market to build something that is it's never plug and
- 6:57play, but kind of plug and play full solution for steering
- 7:04towards sustainable, profitable growth where you're aligned, as
- 7:08you were talking about, James, like both aligning today's
- 7:12today's sales target with the contribution or whatever bottom
- 7:16line targets together with the inventory and the customer
- 7:20acquisition. And the inventory connection is
- 7:24really so important because then you can optimize for the, the
- 7:31full season targets. Because if you just sell out
- 7:36today, the best sellers, like you're just going to suffer a
- 7:39week from now or two weeks from now, right?
- 7:42But that is really like that. That is really like why we why
- 7:46we started demon and and ground? What do Mattel, Victoria,
- 7:52Beckham, Sennheiser and Glossier all have in common?
- 7:57They power their businesses with Shopify.
- 8:01Shopify is the most innovative and scaled commerce platform on
- 8:04the planet. It also happens to have the best
- 8:08converting check out on the planet Shopify Powering the
- 8:13future of enterprise commerce. Find out more at
- 8:16shopify.com/enterprise. Yeah, it's interesting because I
- 8:22definitely see us where because you know, global supply chains
- 8:25are still challenging and people going to have short term stock
- 8:29outs that were unpredicted and that happens very, very quickly.
- 8:34And not having the intelligence behind it to, to see, to be able
- 8:38to marry the, the planned stock versus the actual available with
- 8:42the sales velocity to know that, do you know what you're this P
- 8:46OS not landing, you're going to sell out three weeks quicker
- 8:48than you expected. It's really challenging.
- 8:51So I'd love to let's answer that one first, because I think
- 8:54that's really interesting standpoint.
- 8:55Then we'll get into the MMM stuff.
- 8:56So let's talk about how do you marry that up?
- 8:58Because a lot of the business I work with, they reach a point
- 9:02of, I think data fatigue where it's either they don't have the
- 9:07expertise to do it themselves, but they know what they want to
- 9:10achieve and the the resource isn't there or they're not
- 9:12working with the right tools. And therefore it's we can't do
- 9:15that. So we'll carry on.
- 9:16So how do you tie that up and make it work accurately?
- 9:22Yeah. So like, I'm not the data
- 9:25engineer here, but we had 20 of them.
- 9:32Almost no retailers have that. It's so much down to the nitty
- 9:40gritty details of the data models to have everything really
- 9:45connected when it comes to the inventory, the each product,
- 9:50each variant, each size, but also connecting that.
- 9:54And that is where I see from an operator perspective, the most
- 10:02low hanging fruits with all D2C brands jumping on the DEMA
- 10:06platform, really seeing their inventory.
- 10:10And we have not only your current inventory, but we also
- 10:13predict your inventory. So really like marrying that
- 10:19with your marketing spend on each product per day and really
- 10:24see what's going on there. Because it's not only, although
- 10:29we're going to focus on that now the stock outs and everything,
- 10:32but it's not only the the risk of stock outs, it's also your
- 10:36brand perception, right? And to not go into all the
- 10:41details of how let's call them feed based ads, format the the
- 10:48shopping ads, performance Max, the catalog ads, etcetera,
- 10:52right? But how they work is you ask
- 10:56them for a great robots, they will optimize for a great
- 11:01robots, right? They will sell more and more of
- 11:05what is working. They will move more and more
- 11:08spend to what is working for each day that that specific
- 11:12product is working right. It's not uncommon to see 80 plus
- 11:17percent of the spend being down to like 20, not percent,
- 11:24sometimes 20 products. And the extreme products, like
- 11:30the top five of the products are having an insane part of that
- 11:35spend and that can go fast and you need to be fast there.
- 11:39So although I believe that Meta and Google's algorithms are one
- 11:44of the best in the world's, they're just going to like work
- 11:51towards the goal, use you, you tell them to work right.
- 11:55And and the problem is that you can all of a sudden have, let's
- 11:58say that you're a clothing brand, right?
- 12:00But you have some accessories and I've seen this first hand 10
- 12:04plus times and how we can really like on during onboarding.
- 12:08We're like Chase's course correct here fast.
- 12:12But like a clothing brand all of a sudden, sometimes they for
- 12:17whatever reason, have like nice socks in a market that might not
- 12:23be their like first market, right?
- 12:26So you have a head office. They're not like more than human
- 12:31beings, right? All of a sudden, like that
- 12:35third, 4th, 5th market spends 80% of your marketing spend the
- 12:41last days on socks. That is the brand perception of
- 12:45your brand in that market. It's not only going to drive
- 12:51great, robust, horrible contribution margin because
- 12:54those socks cannot defend the fulfillment cost right and the
- 13:00shipping to that market, but it's also all of a sudden you're
- 13:03a sock brand and this happens all the time.
- 13:09And then you sell out of those socks, those socks that were
- 13:13supposed to be amazing. I don't know how you say it in
- 13:16English, but the, the extra, you know, thing you put in the in
- 13:19the basket when you've already picked the sweater, right, that
- 13:22type of product, you know, so like you're, you're out of that
- 13:27extra contribution margin possibility or potential with
- 13:31the socks, but you're also a sock brand in that market right
- 13:34now. So it's like what we do then is
- 13:38really like, we let you automate it of course, but like, but we
- 13:42also let you see it in front of your eyes connected, put it in a
- 13:47graph. If you're a graph person, put it
- 13:49in the table if you're a table person.
- 13:50But like spend sales contribution, margin, LTV
- 13:56contribution, new customer per product, per day per ad, what
- 14:00have you, right? That is really like where you
- 14:03need to get to to do this in an efficient way 2025 because it's
- 14:11so fast now. And if you will, because every
- 14:15people will listen who who are using multi mix models and they
- 14:20understand it, but maybe they've got challenge around how they
- 14:22optimise and they prove it and others who don't actually really
- 14:25know what it is and what the the goals are.
- 14:29So you've talked about one element, which is about like
- 14:32brand protection and that makes a lot of sense about how your
- 14:35brand is perceived based on the ads and the products that people
- 14:39are seeing and what, And for people who aren't used to you
- 14:41knowing what you do sell and don't sell what, what else
- 14:45commercial when you're working with with retailers, what are
- 14:47the fundamental goals and drivers of using?
- 14:51MMM. You want to know what actually
- 14:57works, period. And MMM is not biased towards
- 15:06the clicks. It's not going to be like it's
- 15:12privacy safe. Like we don't care about you,
- 15:16James. If you clicked an ad, if you saw
- 15:18the ad, like actually we don't care about you.
- 15:23Although we're customer centric, but in that marketing spend,
- 15:26like we're not going to be like over obsessing with that buyer
- 15:30journey. We're just going to measure in
- 15:33an MMM investments in different buckets and I say buckets and
- 15:40not channels because of a reason, buckets of ads,
- 15:45campaigns, what have you. It could be a channel that we a
- 15:50bit wrongly called out of home and we just put everything in
- 15:54that bucket. But it could also be top of
- 15:57funnel campaign in Miata or retargeting in Miata or
- 16:04performance Max, very like free performance Max in Google Ads.
- 16:11Like it could be so many different buckets.
- 16:14What you measure in an MMM is investments in these buckets in
- 16:19pounds, EUR, dollars, whatever. And then you see fluctuations in
- 16:28those investments and you measure the output of sales that
- 16:36is typical MMM or actually kind of like that is top line MMM.
- 16:45The thing that we do that is extremely important for
- 16:47e-commerce and I, I, I argue for for brick and mortar retailers
- 16:52as well is that we actually have the, the, the time variance
- 17:00effects becomes a bit technical here, but like the seasonality
- 17:06in there as well. We have spend on a daily basis
- 17:11here and we see your seasonality and measure it over time.
- 17:16And that means that in part of the season, top of funnel
- 17:21initiatives might give you more even right away kind of and part
- 17:27of the season, it might be better to capture the demand in
- 17:32a Google, let's say white sweater like that might be more
- 17:38bang for the buck in one part of the season, but not in the other
- 17:42part of the system. And MMM is also measuring the ad
- 17:46stock effect. That's a, that's a term worth
- 17:49Googling so you can read it in proper English and not have my
- 17:53broken Swedish English. But like the ad stock effect is
- 17:57the effect that the money you spend today actually have for
- 18:02the sales in a week or two or three or whatever, right?
- 18:07And typically bottom funnel investments have a very, very
- 18:11short ad stock effect. And the further up you go, it's
- 18:16going to be longer and longer, let's say a top of funnel
- 18:19awareness campaign in Miata has significant for some brands in
- 18:27some markets because it's always different, right?
- 18:30And for some parts of the season, that awareness campaign
- 18:34can have like a 10A week ad stock effect.
- 18:38And that is something that you need to measure and you get that
- 18:41like in a, in a good MMM, like for example D mass.
- 18:45That, you know, that's a bit that's really interesting
- 18:47because as I said to you before, we we did this recording.
- 18:49It's this is not my expertise. I don't don't do digital
- 18:52marketing day in day out. But that's, I guess that's the
- 18:55that's a real good benefit that I hadn't thought about is that
- 18:59is being able to future project and really help to measure the
- 19:05impact of upper funnel when people really struggle in sales
- 19:09based numbers based cycles. I've been able to hive out small
- 19:14budgets initially to demonstrate the impact to the sales rate
- 19:19later on, specifically from those campaigns.
- 19:22That is really, really interesting actually.
- 19:25Yeah. And I mean, to be frank,
- 19:28everyone who has been in any type of it's called marketing,
- 19:32but it's so much more sales to be frank, when you're like
- 19:35working e-commerce performance marketing, right?
- 19:37So anyone who has been in these kind of roles have been wanting
- 19:42to spend upper funnel, but have no signals and can show can
- 19:47never show that that is actually giving effect.
- 19:50So whenever sales goes down and the the reporting up to
- 19:57management, it is not so, so to speak like welcome with two
- 20:02thumbs up, meaning like sales is down, marketing spend is flat,
- 20:07meaning your cost of sales is horrible.
- 20:09If you hopefully measure contribution margin, maybe that
- 20:13is also horrible like you need to do something, you need to cut
- 20:16spend. It's not wrong to cut spend, but
- 20:20we're going to spend this where you can't show numbers to
- 20:23support that spend. That it will always be top of
- 20:28funnel versus bottom funnel, everything else equal, right?
- 20:32And MMM protects you from that. And that is so important because
- 20:37what you said in the intro for for for this episode, which I
- 20:44really emphasize with or not maybe the wrong the wrong
- 20:50English word there, but but I, I really feel empathy for the
- 20:53teams that are pushed so hard to meet sales targets and they're
- 20:58pushed down bottom funnel. The thing that they might need
- 21:02the most for the sales right now this week might be to stay,
- 21:07maybe not top of funnel, think out of home, sure.
- 21:12And you can't really make those decisions on a daily basis.
- 21:16But the more higher up me at the ads, TikTok ads, Google, let's
- 21:22say YouTube, like whatever, that might drive better sales this
- 21:27week than the very bottom funnel things.
- 21:32Although the very bottom funnel things will always show a better
- 21:36robust. So you keep them because your
- 21:41fear of missing out. And that is really where the
- 21:44marketing mix modelling is, being that sober guidance on
- 21:49your shoulder and telling you what everyone understands and
- 21:53giving you the number to discuss with, not the CMO.
- 21:58Yeah, the Chief financial officer, that is the person that
- 22:03you need to talk to, right. The CFO needs to see the
- 22:08numbers. The CFO in indeed to see retail
- 22:11world is not a person who don't believe in brand because then
- 22:15they then they would leave that whole business, right.
- 22:18But if they don't see any proof of, you know, your marketing
- 22:21spend actually providing any value to you, like they're just
- 22:26going to say cut it, Yeah. Yeah, I think it's the, the, the
- 22:32challenge that ECOM teams have is that they want to be able to
- 22:35demonstrate the impact of short termism.
- 22:38But if you only look at it from a ROAS point of view, short
- 22:41termism pays because it's delivering sales and it's highly
- 22:45measurable and it's like great, we, we increased sales by 15,000
- 22:48last week and OK, it was, it was only a 2 to 1 ROAS, but still it
- 22:53made profit because of our margins.
- 22:54Great finance team happy. But then if you flip the long
- 22:58term view and look at like the retention piece, the lifetime
- 23:01value and the the the like margin over time of those
- 23:05customers and are able to demonstrate that short term
- 23:09promotional activity might be good in initial sales.
- 23:12So if you've got a discrepancy, you've got to pick up great.
- 23:14However, sustained focus on that actually compromises long term
- 23:18margin. That's the evidence ECOM teams
- 23:21need to be able to fight their corner and be able to get those
- 23:24buckets of investment in like the upper mid funnel.
- 23:28And I think that one, one of the big problem here is really the
- 23:32whole siloed, not only because that's kind of like what we're
- 23:37solving and why we are around not only the data silos, but the
- 23:42org silos here. The E com team fighting for like
- 23:46if it should like you're right, but it shouldn't be the E
- 23:50comments to fight for their right to spend and their share
- 23:55of budget or whatever. Like it might.
- 24:00It might sound sound silly, but like you should be one team,
- 24:04right. You should be one business.
- 24:07You need to fight together for what's best for the company and
- 24:11then you need to have all your data connected and and that is,
- 24:17and I hear myself pitching what we have too much, but you you
- 24:21could in theory at least try. You can build it yourself, but I
- 24:27mean, you really need to connect online and offline, meaning like
- 24:32the stores as well, right? So the MMM needs to be Omni
- 24:35channel. Of course, the MMM can't only
- 24:39look at the top line. Let's say you, you discount too
- 24:42heavily. Like the MMM should reflect that
- 24:46and say that you're actually not bringing contribution dollars
- 24:50to, to the, to the brand right now because you really have.
- 24:55I see it as and I don't know how that translates to English, but
- 24:58you you have two gas pedals as an E com team, you have the
- 25:03marketing spend on the discount and and you need to find the
- 25:08balance. And sometimes it's not
- 25:11necessarily binary, meaning you you can't do discounts and that
- 25:17will just ruin the whole brand. But it most likely.
- 25:23And that effect is extremely hard to measure and be certain
- 25:27about, but it will take down the willingness of paying full
- 25:31price, right. But we can just yeah,
- 25:34philosophically think about that, but it cannot come to the
- 25:39cost of you know, your discount and your push marketing spend,
- 25:44meaning you will have a great robust, but absolutely terrible
- 25:48contribution, Mario, because you know, the basket can't protect
- 25:52the the fulfillment cost, right. So that needs to be on the
- 25:57finger tops like here or how do you say it like easy to to get
- 26:03ECOM teams and the CFO or the whole financial controlling team
- 26:08merchandisers, etcetera. They need also to have this data
- 26:11right from their point of view. Yeah.
- 26:14And the, the delayed stock impact, I I like that idea of
- 26:18because if you can model and it's never 100% accurate because
- 26:21things change over time and they're changed by market and by
- 26:24what's happening with a competitive etcetera.
- 26:26But if you know that a particular channel, whether it's
- 26:29Meta or Google of upper funnel like brand marketing has a 10
- 26:35week impact cycle, the ability to plan in that marketing around
- 26:40your promotion cycles where you know, a, a, a budget has said
- 26:45right in April, we need 30% year on year growth.
- 26:48Well, actually we've got to have those up funnel campaigns
- 26:50specific to those products or promotions live -10 weeks.
- 26:55The ability to improve your your content and promotion calendar
- 26:59off the back of that and then keep tweaking it is really
- 27:02powerful. Yeah.
- 27:03And another one that I sat working on a lot at baby shop
- 27:08when we, when we scaled up business, we had, I keep
- 27:13forgetting I'm, I'm not good at remembering exact numbers, but
- 27:16it was it four or five web shops and we were shipping to 80
- 27:20countries on like a significant volume.
- 27:23I mean, you can just imagine the mess, right?
- 27:25But when we were planning for let's say Q4, I knew that from a
- 27:31new customer perspective, I need X number of new customers,
- 27:36right, that are willing to buy again.
- 27:38Let's for, for for lack of other words, let's say having a good
- 27:45predicted Ltd, right? Just that before that specific
- 27:51period that you were talking to talking about as well, right?
- 27:54And connect that to products. I see that as a really low
- 27:58hanging fruit for anyone that are able to connect their data
- 28:03internally, like understand if not a daily basis, weekly basis,
- 28:10at least early on in the season, what products in what markets,
- 28:16Dr. new customers that are profitable.
- 28:20So, you know, as you say now, James, like if I want growth
- 28:25late season, it could be worth pushing, pushing a certain group
- 28:32of products harder in some markets, because in these
- 28:36markets, these products give me a lot of profitable customers,
- 28:40right? And that profitable customer are
- 28:44more likely to come again and buy again later in the season or
- 28:49like what have you, right. There are so many metrics for
- 28:51all these kind of things, but in in more of a conceptual way.
- 28:56So there's so many things that I'm like, I'm like, and maybe
- 29:01you can hear me being frustrated, but I'm, I'm
- 29:04honestly frustrated. That's so many of my like
- 29:07favorite brands out there. When I talk to them like it's,
- 29:10it's a privileged position. I have like I can personally
- 29:14reach out to my like privately favorite brands and be like,
- 29:20Hey, you should have my, you should have my toe.
- 29:25But then I talk to them. I'm like, oh God, I wish you so
- 29:29much better in in terms of like data, like you should know, you
- 29:32should have this data. Like I, I love the brand, but
- 29:35you don't have this data and it and that is also why I see you
- 29:38discounting too heavily now when we're entering June, July,
- 29:42right? Yeah, the, all the brands I work
- 29:46with, what are the the biggest gaps I have in data is, and I
- 29:50know Paul posted something linked to him this week about
- 29:54the disconnect and the inaccuracy in analytics data.
- 29:58So where people don't have large teams who've got the ability to
- 30:00set up proper data warehouses and they're pulling all the data
- 30:03through and doing all of the, the data storage manipulation
- 30:06and views there, where people are on and on GA and people have
- 30:09moved from the old UAE to to GA 4.
- 30:12There are still so many partial implementations where people
- 30:17were either the data views aren't as they need if they
- 30:19don't have the event data to trigger the analysis or the data
- 30:22accuracy is not correct because they haven't set up like the the
- 30:26DTM correctly and the tags aren't firing correctly.
- 30:28And the duplication issues with how the the platforms are firing
- 30:32tags. And you have a lot of what what
- 30:35is your advice? Because you could have a really
- 30:38cool brand come to you and say tightly, Henrik, amazing, but
- 30:43our data's screwed. Like what is your?
- 30:45How do you advise people? Because you you can't fix
- 30:48underlying broken data. The data's the data's got to be
- 30:51clean to help inform models. You're absolutely right there.
- 30:57And given everything going on in privacy land, and I feel like
- 31:05the regulation is just one part of it.
- 31:08And then you have like the Big Apple, the Apple being so
- 31:13aggressive on. And I personally, I, I have no
- 31:17problem with being not, not tracked that much, to be frank.
- 31:22But you know, it's so many data points you can't trust out there
- 31:26because all of a sudden Apple puts in something new in their
- 31:29iPhones and then, you know, they triangulate out Miata and Google
- 31:34even more. Like there's so much data in the
- 31:37visits and everything that is not in Miata or in in, so to
- 31:43speak, in Google, like the ad platform data that you can't
- 31:48really trust because then you land on the website and UT Ms.
- 31:53are like taken away or you have cookie consent, deny all or your
- 31:58like ad blockers is not a problem these days.
- 32:00It's Apple like and so on. So like I've never trusted those
- 32:07metrics less, but also I've never seen a better reason to
- 32:14look at them less because you have so much possibility in
- 32:20looking at metrics that can't be wrong that actually should steer
- 32:25you. Like MMM in is one way of
- 32:28looking at it, but but think of it investments in initiatives
- 32:35and output. The output is an order.
- 32:39It's a new customer or it's a returning customer.
- 32:43It's a returning customer for the 4th time.
- 32:46It's a profitable customer for the 4th time.
- 32:49It's a product sold to a country where it's constantly delivering
- 32:58a negative contribution margin. Like there's so much in the data
- 33:01that is not failing. And yes, I, I advise every brand
- 33:09out there having huge GA four issues to have something like of
- 33:16a fixed shore, like fix whatever.
- 33:19But then year over year data, like you, you, you cannot follow
- 33:23that, right? You cannot be like, oh, now
- 33:25we've fixed it and then start reporting up, I don't know,
- 33:31engage sessions or what have you, right, or users on the
- 33:35website. And then you look at year over
- 33:36year data and and you're like, what, what, what you just fix
- 33:40something and that's going to change everything.
- 33:42So you can't look at every year data, right?
- 33:44So there's so much reasons in in start looking at what really
- 33:51drives sustainable growth in your business and what really
- 33:58matters because a visit is just a vanity metric.
- 34:03It's the number of the number of sweaters that are out there used
- 34:10by our customers, you know, and the, the ability that we drive
- 34:15back customers, everything else like you could just like take
- 34:21away. So in one way, like an MMM for a
- 34:25company that doesn't have an MMM is of course to like look at
- 34:28your income statement, to be frank and correlate that and be
- 34:35like follow the, the different spend patterns and, and really
- 34:40see it, you know, model it in one way or another down to your
- 34:43like income statement, to be frank.
- 34:47Yeah, the the whole vanity sanity metric debate.
- 34:50I mean, there's things like conversion rate.
- 34:51I drives me nuts because conversion rate is a completely
- 34:55useless metric it with no context around it and other
- 34:59metrics to understand the levers where I want to talk about
- 35:02incrementality testing. It's a hot topic.
- 35:05People use it all the time, but equally.
- 35:09And I know there'll be people listening who will will have
- 35:11been jaded about this. There are too many what I would
- 35:14call suspect case studies in the market saying you doing this
- 35:19triples your revenue conversion, whatever we're actually, it
- 35:24isn't proven incrementality. There's probably a lot of
- 35:26substitutional sales. What is your about?
- 35:28How do you genuinely measure incrementality in marketing?
- 35:33Yeah. So that's a, that's a good
- 35:34question. It's an important topic and like
- 35:44incrementality testing, it's often called year lift testing
- 35:47in the ad platforms. There are other type of
- 35:50incrementality testings going on, right.
- 35:54But how, how we work with it, which I think is obviously I
- 36:00think is the best way, is that incrementality test is let's,
- 36:05let's use a term that everyone remembers from being in the
- 36:09industry 10 plus years. An AB test.
- 36:13I think people can grasp it a little bit better when you talk
- 36:17about an AB test and it's more of a scientific AB test or at
- 36:22scale in an ad platform. But you run an AB test, you get
- 36:29causality. That is very important.
- 36:33You get causality in this in in this test.
- 36:38What how we do it, we feed the MMM, the MMM is talking to the
- 36:43attribution model. That is how I see it how you
- 36:46should work with it. But incrementality tests is so
- 36:51easy to do these days and everyone should do it.
- 36:57What you should test is top of funnel effect because it not
- 37:04taking care taking into account the attribution right?
- 37:09But you should also look at the bottom funnel things and do
- 37:14incrementality tests there. Easiest most low hanging fruit.
- 37:19Branded search, the branded search campaign.
- 37:24Does it add any value at all? Any value whatsoever?
- 37:29Most likely limited value. But the problem with the
- 37:33incrementality tests are mainly 2 things as far as we look at
- 37:38it. 1, you do it first in Meta and then you do it in Go Google.
- 37:45Let's make it simple and take out all the other ad platforms.
- 37:49And it's really putting Google versus Meta.
- 37:52And worst case, you have one internally working with Google
- 37:56ads and then you have one with Meta ads.
- 37:57So then it becomes a fight internally.
- 37:59And it's like, I should prove you this.
- 38:01I should prove you that the problem is if you do it in AD,
- 38:05in platform incrementality tests, you don't have a unified
- 38:10way like a measurement framework for this.
- 38:14Meaning it's actually comparing apple to pear.
- 38:19What we do is very shortly like we make it apple to apple.
- 38:25So we look at all your data, all the orders, all the ads spend
- 38:29that we can actually from a more unbiased way, we can say that
- 38:34this is the test. This is the way you should test
- 38:37and if you want to test your performance marketing
- 38:40performance Max campaign or your top of funneling me after like
- 38:44this is how you can test it and you can you can actually compare
- 38:48Apple to Apple. I think that is extremely
- 38:50important. But then another thing that is
- 38:53important to know, like it's not that you should not do it, but
- 38:57when you do it in platform, you don't know maybe the two most
- 39:04important parts of the incrementality and that is the
- 39:09new customer acquisition because Mia and Google, they cannot know
- 39:16if it's a new customer or not. We do, for example, we have all
- 39:21your orders, right? So we can actually model that
- 39:24in. So you have your kind of like
- 39:26incremental CAC, right? And you can look at also the
- 39:31contribution And we're so obsessed with that because it's
- 39:35very often that if you have incremental Rowas, you might go
- 39:41after the most low margin customers good at finding
- 39:48discounts or what have you, right?
- 39:50It's not the row as you want. So you want that incremental
- 39:56like sustainable robots that their their profit robots, so to
- 40:00speak, right. Yeah.
- 40:02So it's it's it's really about like again, and that that is our
- 40:06our whole belief. Like you cannot isolate being in
- 40:10one data silo. I think that incrementality is
- 40:14is really to tell you like building trust in the media mix,
- 40:20right. But it it's.
- 40:23And it's something that you should run and if you can only
- 40:27run it in AD platforms, the incrementality test there do it
- 40:32by all means. But what you need is really like
- 40:36the same measurement framework for for all ad platforms.
- 40:40And I think that incrementality, what it can actually serve you
- 40:45as an head of E com, head of marketing, whatever is, is also
- 40:51a way of trusting the media mix that you have, right?
- 40:54But it's a little bit too in inefficient to run all the time
- 40:58to stare at the media mix, right?
- 41:00So you need then like an MMM to not fall back in that full like,
- 41:06oh, I'm fully dependent on attribution, like cookie based
- 41:10attribution, right? I.
- 41:12Think it's. It's interesting because.
- 41:15The. If you go back maybe three
- 41:17years, everyone was still very, very much obsessed about
- 41:20attribution models. And now I've not heard people
- 41:25talk about attribution models, how to do what's the right one
- 41:28to use and for for quite a while.
- 41:31Whereas incrementality and actually what what people want
- 41:34to know is what is the impact of my action and is that impact
- 41:39reliable? Is that impact consistent
- 41:43therefore I can model of it? Or is it an outlier where it
- 41:46happened but we don't really know why and therefore I can't
- 41:49with confidence repeat that. And that's to me why
- 41:52incrementality testing is more powerful.
- 41:55Yeah, because it enables you to isolate specific elements and do
- 41:58that. And I'm not surprised that this
- 42:00is a where the technology is headed, but be especially with
- 42:03all of the data blocking issues out there, but be where ECOM
- 42:08teams are focused. Yeah, but the important part
- 42:14where where incrementality tests fall short is the whole, you
- 42:19know, you need something always running, always giving you
- 42:25feedback. Like you can't have this holdout
- 42:27group where you have 40% less spend or no spend at all the
- 42:32entire season and then all of a sudden you have the seasonality
- 42:37changes, right? I think that you should have
- 42:40more of more or less like an always on testing agenda in
- 42:46general, whatever company you are, like you should always be
- 42:49experimenting and testing. But that incrementality tests
- 42:54can't go on forever. All, all markets, all the time,
- 42:58all channels, right. So it, it's, it's going to serve
- 43:02you with a guiding light. But then all of a sudden you
- 43:05have discounts, you have weather like in some, in some countries,
- 43:11right? They differ a lot and the
- 43:13citizens and what have you, right.
- 43:15So yeah. Now it's important, and
- 43:18nothing's perfect, you say. Look, I've got 1 we had loads of
- 43:21potential questions before we start this, but I've really
- 43:22enjoyed that the, the, the conversation has been really
- 43:25quite open and intriguing listening to you and I know
- 43:27listeners with a lot of value. I've got one more key question I
- 43:29do want to get through though you, which is that one of the
- 43:32one of the key things in in Denmark is that you calculate
- 43:36the money brands are left on the table.
- 43:38Where people are wanting to see tools evolve is the automation
- 43:43around the intelligence and surfacing intelligence rather
- 43:46than expecting you to be the proactive one in finding it.
- 43:49So I'd love you to talk people through like what signals are
- 43:51you using to help? I like identify where there
- 43:55could be an issue as in you, you missed out Southern X or you
- 43:59know, you're you could have got better margin of why.
- 44:03Could you talk people through a couple of examples maybe of how
- 44:05you're doing that? Yeah, no, but it's, it's, it's
- 44:10again, like when you have all data connected, you're not
- 44:16forced to be a victim of averages all the time, right?
- 44:22So this sweater is not the same in all my markets, right?
- 44:27You leave money on the table if you push this sweater sweater
- 44:33the same way in all markets, meaning, oh, the contribution
- 44:37margin on this sweater is X. So I should treat it equally in
- 44:44UK, Sweden, Germany and the US. No, that's not the case, right?
- 44:50So then you need to automate all these things.
- 44:53So what we do is we, we speak to not technically, but like
- 44:59conceptually, like we speak to the feed, whatever, like feed
- 45:02ads you, you have out there, right?
- 45:04So you can make this automated into and treated differently
- 45:13depending on the market and and and is leaving money on the
- 45:19table today versus in the future.
- 45:20You, you are very like much talking about the future before
- 45:24and, and, and, and, and, and so you, you were, you were saying
- 45:34that either we, we, we, we get sales now or we get a better
- 45:41brand or whatever in, in the future, right?
- 45:43I'm, I'm, I'm oversimplifying, but what, what you, what you
- 45:47actually can do is understanding that some products today are
- 45:53better sold in one market versus the other, right?
- 45:57And that is also like, if you connect that to your potential
- 46:00stock outs coming up. So this is let's let's call it
- 46:05a, a best seller. We predict that this is going to
- 46:07be sold out. I, I think you mentioned like 3
- 46:10weeks from now. OK.
- 46:11So in three weeks now if we continue as we do this one is
- 46:14going to be sold out. And let's say that that is going
- 46:18to be mid season. Like that is not good.
- 46:20We want a very balanced inventory throughout the season,
- 46:26especially the the keep the best sellers and maximize them over
- 46:31the season, right? Because that is going to drive
- 46:33more efficient sales on, on the free and organic traffic
- 46:37etcetera, right? So what you should do with this
- 46:40sweater might not be to cut all marketing spend because it's
- 46:46kind of like wasted marketing spend because it's going to sell
- 46:49out anyway. What you might want to do is and
- 46:53you can automate all this, right, you want to cut it in the
- 46:58countries where it brings least new amount of new customers,
- 47:06worse margin or where that market is less dependent on this
- 47:13sweater for the the total saves. Kind of like it could be so many
- 47:17of those things and it's money left on the table in like the
- 47:23long tail is so long, right? There's so much money in the
- 47:28long tail that no one gets to when they have dating silos and
- 47:33not able to like automate those things.
- 47:37But that's also like where where the MMM comes in really left
- 47:42money on the table is where you overspend or underspend in a
- 47:48bucket of of of ads, right? Where you can if when you can
- 47:53efficiently scale that bucket and you don't, you leave money
- 48:00on the table. Yeah, you don't scale it right
- 48:04now because your attribution model is not letting you to
- 48:09let's say it doesn't look that efficient now for XY said reason
- 48:13in GA 4, etcetera, right. But an MMM could tell you with
- 48:18the ad stock effect and everything like this is
- 48:21extremely good. You should spend more here.
- 48:28And that is also why we believe that an MMM should not be on top
- 48:34line. It should be on the contribution
- 48:37line. Because then for for all the MMM
- 48:40people out there that see saturation curves in their head,
- 48:44when I say it like a saturation curve with contribution margin
- 48:49is actually going down to 0 again, meaning there's an
- 48:53equilibrium of exactly how much you should spend to optimize for
- 48:58a Max for for like maximizing contribution margin per per per
- 49:05per channel or bucket of of categories.
- 49:08So that is actually always helping you to know where you're
- 49:13overspending and and underspending.
- 49:16And you don't need to think about like when should I stop
- 49:19spending? Because that is very difficult
- 49:21to understand in a in a typical MMM.
- 49:24Interesting. I could carry on going like all
- 49:26these episodes. I could carry on going for ages.
- 49:28But I will I will stop there. Also, want to leave, leave the
- 49:33door open for people listening who might want to speak to you
- 49:36more about that. They e-mail the platform on how
- 49:38it can solve some of their challenges.
- 49:40So if someone listening either wants to challenge you on
- 49:43something, push you further on how or how it really works, or
- 49:46even look at the the platform, how do they reach out?
- 49:50So yeah, my name is Henrik Hoffman Craft.
- 49:54I think I'm the only one with that name on LinkedIn.
- 49:59So LinkedIn is a perfect place. Also, if you want to discuss
- 50:03stuff I I love discussions on on LinkedIn, obviously you're more
- 50:11than welcome to e-mail me as well.
- 50:13That's Henrik at dima dot AIDEMA dot AI.
- 50:18Amazing look Henri, always love to chat and see you find it
- 50:21fascinating. Really enjoy seeing the success
- 50:23story of Dima and thanks so much for coming on the podcast.
- 50:28Thank you so much, James I, I really appreciated this
- 50:31discussion. Thanks as always everyone for
- 50:32listening. Any questions do reach out.
- 50:34Do contact Enrique if you want to learn more and hopefully
- 50:38catch you next week. You're listening to the Inside
- 50:42Commerce podcast, hosted by experienced consultants James
- 50:45Gerd and Paul Rogers. Inside Commerce is your fast
- 50:48track to digital knowledge, featuring interviews with
- 50:50leading e-commerce retailers, industry thought leads and the
- 50:53C-Suite, key tech companies to explore the latest digital
- 50:56strategy, customer experience and tech.
- 50:58Discover what the latest trends really mean and get practical
- 51:01advice based on the experience to help fine tune your
- 51:03e-commerce strategy and tech stack.
- 51:06Follow us on LinkedIn and reach out to James and Paul if you'd
- 51:08like to discuss an e-commerce project.
- 51:10We hope you enjoy the latest episode.