From falling sales to 100 million: what does it really take? #110

A very warm welcome to a new episode of Zero to One Hundred. With me as usual is Fredrik Ankarsköld. Welcome back. Thank you. Great to see you again on a Monday like this.

Yes, we have talked about lots of exciting things. New week, new start. I am always just as hyped to come back, meet you and talk about what is happening in e-commerce. So much is happening, something new every week. Think how much easier life is for us than for people who lived two hundred years ago, but also how quickly things are moving for us to keep up with.

Our brains have not evolved for this rapid pace. We essentially have the same primitive brain we have had for thousands of years, while developments are moving incredibly fast. If you look day by day or week by week, it is easy to think that not much has happened.

A little more happens month by month. Then it starts to show. But zoom out one or two decades. Two decades ago, we did not have smartphones. Now we have a computer in our pocket, on our eyes if we want with smart glasses, and on our wrist.

We have computers in everything at home, in smart homes. Whoop taking measurements. You have everything, really a complete picture of your body through measurements. That is just two decades.

Now we have agentic shopping and things happening in the background while we sleep. This morning when I woke up, I received notes for the podcast. They were not the best in the world, so I scrapped everything. It really was poor research. But even so, the fact that it just did it...

...then I wake up with all my emails prepared as drafts and only have to click. Soon it will just be AI talking to another AI, a little creepy. But in some cases I prefer that. I am building an AI company on the side myself and am programming a node system. You can configure it so Grok talks to Gemini, which talks to Claude. In what I am building, you give them different roles: someone is a critic, someone works on security...

...then they can discuss things back and forth until they agree. Then you get their joint conclusion, rather than just the first gibberish they spit out. Does the quality improve when they agree?

Incredibly much better; you cannot even compare them. Because you ask it to do... We will not get into that now.

No, but what an amazing thing to do. I am just like: wow, so much is happening. Apart from that, a lot is happening on the e-commerce front.

A lot is happening around Meta and AI, where they are releasing news about more agentic media buying. The question is when manual media buying, as we have seen it over the past decade, will disappear entirely. I think it will be completely gone within two years. Already this year, many startups and scaleups will switch from manual buying to agents buying media for them. Thumbs up or thumbs down, what do you think? We can start by defining it: what do you mean by media buying?

A person buys visibility for a company. That is the media buyer's role, and it should be done as cost-effectively as possible. Driving cheap clicks, quality traffic and, ideally, highly converting traffic in our industry.

That is a good media buyer. So what do you think AI's role will be for Facebook? AI can already analyze more e-commerce parameters than a human, and do it around the clock, all year round. It is always on, just like the ads. Ads on Meta are live around the clock, even while you sleep or do not manage to wake up before a news story has changed the media landscape.

To me, AI replacing that is a no-brainer, and the question is how soon. Right now, at Porobic Group, we use our own AI bots that my colleague Ivan built. He named it Sigrid, a fun media-buying robot. For one thing, it uploads ads that have been approved in a library so they are sitting there as drafts when he wakes up. For another, with certain rules it can raise and lower budgets or duplicate an ad between campaigns. You look at hook rate, hold rate, click-through ratio and CPM, and whether the parameters are met. Just as a human has guidelines around them, you can predefine the same parameters in the agent Sigrid and...

...so when I wake up, it has made changes in the account. Of course I hold Ivan accountable if something goes wrong, because he is responsible. I do not see that as far off; it is almost as though Sigrid is our full-fledged media buyer, while Ivan has moved from hands-on media buying to being purely a strategist.

That is fine. Do you mean that as a strategist he looks at the data with it and plans upcoming ads to be filmed and edited? What is the role?

No, more strategically bouncing ideas around with AI and seeing whether there is a new market to enter. How is the weekend's product launch looking? Last Sunday, for example, we released a new watch at Mockberg. A really beautiful watch that is more like a piece of jewelry: a stylish bracelet where you choose gold or stainless steel, with a small timepiece that obviously works as a watch. But it looks more like jewelry. It is interesting to see: how is it doing in different markets? How is it doing with different demographics in those markets? What will the return rate be? For now, it is difficult to put all the parameters into...

...AI and ask it to make the strategic decision. At the moment, Ivan wears that hat. But in six months or a year, I think he ought to be able to bounce ideas around with AI, which ought to make the decision for him. I had an idea about the node system I mentioned I am building. You can set it up: this is an AI, and then you draw a line to the next node. How should I explain it? You make connections: it should connect to the AI with that role, and then to another one with a different role.

Like a diagram? Yes. If you take Sigrid, you could have one node for data collection. The next checks whether the data has been collected correctly.

Then you can assign different roles and have the same discussion between them that you would have between all the actual people involved.

After that, a presenter who shows the conclusion, how they reached it, the discussion behind it and why. Reasonable. The only question is that the discussion burns more tokens. Does it become an endless loop if they disagree with each other?

No, I have set up rules. If they discuss things back and forth and never agree, they stop after a certain number of attempts. Otherwise, they could work all night and...

...murder the account. You wake up and see that twenty thousand kronor has gone on discussing the American presidential election. Yes, I have made mistakes. That is why I have learned these guardrails. There was something where I thought...

What do you call them, guardrails? Yes, think of when you are bowling and the bumpers come up so the ball does not go into the gutter. That is how I set up the rules. They have come from trial and error. There was one thing that burned through twelve thousand, and I wondered why the money had been charged. Oh dear, that was a bit stupid. Then I could sit and discuss with it why the error had occurred.

I wanted a high level of security and had built in cross-discussion. The problem was that if I wanted to change a button from blue to green, they would start discussing the color back and forth. It was like that with everything I wanted to change. Then I had to say...

...you do not need to discuss that; it is simple. I needed to relax the rules and change the criteria. Speaking of some things not needing discussion...

I came here this morning a little frustrated after a call that meant we started the podcast later than we should have. In my head, it was a five-minute call where we would agree on how to proceed.

Instead, it took thirty-five minutes of bouncing things back and forth and arriving at exactly what we said in the first five minutes. I had to go through why, how, why again and a third time, and then we agreed on the original approach. Sometimes I wonder why organizations need such long discussions before they make decisions.

I do not get challenged that much. I love being challenged when there is data. I do not like objections based solely on gut feeling, because my gut feeling is then probably stronger than yours. Then there is a HIPPO in the room. [unclear in the source: Det här jag är det här gäst.] What do I become? The hippo, the hippopotamus.

"Highest income person" or something like that. Your opinion does not trump mine if we both only have a gut feeling. That makes me think of Jon McNeill's book about the algorithm.

He was president of Tesla. In the book, or an interview I heard, he said that when they presented something to Elon Musk, they first had to have clearly defined the problem.

They needed to examine possible causes and think about solutions: "Of these three, we chose this one for this reason," and back it up with data.

Then they could bring him the problem even if they did not know whether the solution was best. As the main decision-maker, that makes things much easier for you than starting by figuring out the problem itself and getting to grips with it. It makes things easier as a manager.

I know. I do not demand that much from the team around me. We are not talking about Porobic Group now, but an outside company that wants us to become co-owners together.

After the call, I thought that if every conversation is going to be like this, I do not have time. Initially, I thought the project required little involvement: higher risk, higher reward. There is a lot of risk, as in every project. But if the risk and reward are high and the time commitment is low, perhaps I can still participate, since I only have a little time to give.

I can have a team I have worked with for many years take care of a lot, and build the strategy myself. But if the strategy is challenged on day one or has to be explained at every step, I would rather pull out than work with the team in that way.

Could that not just be a natural part of what you need to do in the first few weeks? No, you should have a high level of trust in each other from the first few weeks and throughout the project. My strong opinion is that in a startup, you should divide up the roles and communicate that clearly in advance. Who is responsible for what, and why that particular person? Of course, they must have experience in the role. If you bring me in, I want the final say and responsibility for e-commerce and marketing. I do not want discussions about everything outside that.

For example, a new product formula. I am not the best in the world at product, nor at customer communication in customer service. Then you hand the responsibility to someone else?

Completely. That person can, like Elon Musk's team, come with a proposal and two or three reasons: this is what we should go with, this is what we have tested and this is what we have rejected. Just so you know. I say nice. I can be in an information thread where I only need to receive the information, not take part in the discussion. Do you understand?

Is it not a hundred times more fun when everyone has defined roles and can run at two hundred kilometers an hour in their area, instead of jointly discussing everything in some kind of consensus process?

One thing I am thinking about: you mentioned that a new hire with less experience and perhaps a lower salary can work their way up. You might expect it to take two years before the person knows everything in the role. Compare that with someone who comes in and says they should have a sky-high salary. Then you expect the person to be independent. But in this case, it is an outside company that wants you to invest.

Yes, an investment case where I do not get a majority stake, which creates reasonable expectations. I do not get a majority to make decisions with, and that is fine. I am okay with a minority as long as it still has a veto in the area I am good at.

Do you not think it is good to be challenged sometimes? Sometimes, but I do not have the time. I have to run five, ten, fifteen, twenty different projects. At a strategic level, I cannot sit and get into certain details. Challenging things is great, but with the right data, or at least some data. Not "I feel that yellow color is off-brand" about the site upgrade. It takes energy and it is completely unnecessary to involve me in that discussion. I have sped life up so much. You know when you listen to a podcast, say ours...

...at 1.0 speed. It plays normally. If you increase it to 1.2, you do not feel any difference, but it is playing twenty percent faster. Suddenly, you can get through five episodes instead of four in the same time. If you then go back to 1.0, you notice what slow motion it is.

That is exactly what life is like for me. [laughter] But I see increased speed as learning. One shift for me was that learning is really about changing behavior, not the amount of data I take in.

I have two phases. If I first need to get into a new field, perhaps the volume matters before I can do anything, because I need an overview.

You may have taken in a lot of data at high speed, but if you do not convert it into actual actions, increasing the speed is not positive either.

No, not always. But this is not about learning. There is a difference in level: one person might not know e-commerce at all and have never done it, while the other has worked in it their whole life.

How would you want it to be? I did not go to the doctor two years ago and tell them how to operate on me. Admittedly, I do that now, but back then...

...I trusted that the doctor knew their job. Today, I do not, given my medical history, and have to read up on things. But then at least I come informed, with data, practically with printed research papers that I put on the table.

I still do not walk into a restaurant and tell the head chef how the food should be prepared. I do not do that. I order, eat calmly and do not debate with the head chef about how they created the dish.

How are people supposed to know when you want debate? Some things are business-critical and others are not. Some things you can read up on; others are impossible. Everything you can read up on, you should do before the meeting if you want to discuss the substance. Be factual in your arguments rather than emotional and bringing up sensitive things. Then you meet different founders...

I meet different founders almost daily now. Two or three call, pitch, email or write on forums, so many cases come in. I like investing in young people who are driven by impact, who want to change society or do something better for people. Above all, I want to invest in those who have great potential to do that.

And who give energy rather than taking it. If you feel excited after the meeting, you are a good team that can get in sync, agree and move forward faster. That is needed today. I used the podcast as an example of twenty percent higher speed. You can also listen to the same podcast thirty, forty or fifty percent faster.

After a while, it feels normal. In the same way, a startup needs a faster pace at the beginning. Then you turn it down as the company grows and the stakes get higher. Early on, you need more risk and big, bold bets. Otherwise, someone overtakes you or takes the idea, because ideas are free today.

Sometimes I laugh when people put an NDA in front of me before they have pitched and say I have to sign before they say anything.

Then we can talk about their exciting idea. I get two, three or seventeen ideas a day. I do not care about your idea and I am not going to execute it.

Tell me what it is, and then we can talk about an NDA in step two if there is actually something to protect. Customer lists, I understand that. Purchasing behavior, a factory or an ingredient you want to protect, of course. We sign for things like that, but not because someone just has an idea. I recognize that. People have come to me and said they have an idea but cannot tell me, because then they want to be co-owners and all that in my existing projects. Okay, please keep the idea, because the way I see it...

...execution is the hard part. I have loads of ideas about what I could improve in my different businesses. Actually implementing them is a completely different matter and usually harder than you think.

Incredibly much harder. Looking from the outside at Mockberg or Såpa, it seems as though everything has just gone well since I came in. Yes, it is going well, but you do not see the journey there. In every company, it looks like a straight arrow upward, EBITDA grows year after year and things get better. But in reality, the path looks like this...

...up and down, back, up and down and back again. Okay, now we have reached this point. If we think about your call and what we have learned...

...and what we have learned, what can the listener take away? Nice, a very good question. A lot of people listen to the podcast today, and many of you also get in touch with ideas and other things. I am really grateful, because I enjoy getting involved in lots of proposals. I am happy to participate and respond. But if we are collaborating, you need to give up control in certain areas where you might not be the best in the world. I am not saying I am the best in the world either. But if you think I am good, or pretty good, at e-commerce and marketing, you need to be willing to relinquish control of e-commerce and marketing...

...or come with, just as you said, data, proposals and hypotheses. Then that is fine. It is like I tell everyone in the organization: "Data levels all arguments." If someone says, for example, that Claude is statistically better than ChatGPT for a project and therefore wants to use Claude, there is no discussion.

If there is good scientific support, tests, a good hypothesis or a way to test it, we do not need to discuss it. Go full throttle. I usually say that I would rather hand out a speeding ticket than a parking ticket.

A nice summary. Thank you. Let us turn the page. Something really fun happened this morning before I came to the studio. Speaking of investments, I actually signed a new project.

Really? We teased it a few episodes ago, and now it is done. I sat with the guys at the Qliro event. They traveled up together by train or car.

There are three founders who launched posters about a year ago, and things were slow. I met them, really nice. Right? Young, hungry, and they have not scaled e-commerce before. They have done a little work with dropshipping and various products and then found posters.

The first thing that struck me was: oh, not another poster site. What is unique here? Then, when I started digging into the numbers, I saw: wait, there is something interesting they have done, and that is automation. They do not use someone else's print-on-demand and delivery, but keep the margin themselves. They want to create local jobs in a small town in Sweden.

It is a bit like the Såpa case. I love it when people want to support the local community and create jobs in the country. It is not an outsourced printing factory somewhere in Europe, as is common, that only prints when an online order comes in, where all the company has done is make an AI image of a picture. That was fun. At the same time, AI can do a lot for posters today: creatives, content and copy. When I signed, I thought it could become an AI sandbox. Everything in the organization except the actual production of the material, printing, warehousing, picking and packing, can already be automated today at a very high level that is...

...a hundred times better, no, a hundred is an exaggeration, but better than the alternatives on the market. I have sat down with another entrepreneur myself. I cannot say which company or which product, but I saw all the content and how he created it. Nothing on his site is real anymore. When he did it himself, the pictures had different lighting, different tones and things like that.

Now he has the same lighting regardless of the product. The tone becomes similar, which makes the site more cohesive. Then he has made videos from the photographs.

Photos, descriptions and videos, everything is AI. He showed what it looked like when he tried two years ago and what it looks like today. They are physical products held up by a person, without showing their face. It was extremely difficult to distinguish from reality. Even if you could have suspected AI, it was good enough not to care.

I understand. You are buying the result, not the path to it. Unless it is an investment piece by Andy Warhol, Picasso or someone like that. People think Picasso is very expensive because they know a particular painting. But he painted, what is it, five thousand things just on napkins and small canvases, several hundred a day.

There are cheap Picasso paintings online. Then we can define what cheap means; it is relative for one of the world's most famous, perhaps the most famous, artists. But anyway, so I have entered the poster industry, not directly but indirectly as a strategist. We will set up workshops and guidelines. I found the team interesting: well put together, hardworking, eager to learn and not questioning things too much. That is what we were just talking about, teams that give energy and teams that take energy.

Some people want to learn and run fast. Others sit there with question marks and debate buttons and background colors. I do not have that time today, and did not really have it yesterday either. We do not need to debate everything.

Some things are simply: do it, just do it. Create a foundation that is eighty to ninety percent good. Ship it and then fix it. Let us launch and publish before we keep improving. [unclear in the source: Iteration är i perfektion.] Then we can work on the final percentage points. For eighty to ninety percent of the journey, there are fundamentals and templates we cannot get around. We are not unique in the industry. We are not building space capsules to the moon or SpaceX, which is going public soon. We are not doing anything advanced, just driving traffic to the site, getting people to buy and come back. Let us stick to basics. There is a saying: procrastination can be disguised as perfection. You say: "I am a...

...perfectionist," but really you are putting things off. Yes, even the best often put things off until the last minute because they are administratively boring or something else. That does not have to be directly negative, but we need to move away from perfectionism. In certain niches, it is required.

In high fashion or with an exclusive designer bag, you cannot release something that is eighty to ninety percent finished. The competition is brutal and the target audience wants perfection. But a website selling posters?

Is there already a website? Yes, they have one. At its peak, it turned over one million in some month. [unclear in the source: Den de har satt sig här.]

Then things have gone badly, so it is a good case. What is bad? It has gone from perhaps 1.2 million, say one million in net sales, to six hundred thousand and then four hundred thousand.

I see a rapid downward trend. Up as quickly as down after they found a poster trend. I see no underlying asset value, no returning customers and no proof of concept beyond the site and the print shop.

Really, it is the team you are investing in. Can this team execute at the level you want over the next three years? Three to seven, or three to ten? If the answer is probably yes, it could be a good bet. On my side, the contract discussion was one factor. We sent a contract to three founders. It could have become a week of discussion before we got a proposal back: "These are the things we want to change." Then we would send it back and let it go through a few rounds. Here, we sent the contract and all three had signed the same day, except me. I had not had time, because I had barely seen the contract...

...that my lawyer had sent. The lawyer and they had talked things through. I thought: wait, I also need to read my contract before I sign.

But it shows that you are willing to take a little risk and compromise on things without negotiating every tiny detail. You see the bigger case: it could become very valuable. We do not need to debate who has the final say on marketing. We are bringing Fedja in; you are good at that.

There is also something there I can recommend. I was consulting for an entrepreneur I know who said: "Fredrik, I want you involved in this. We are going to sign a contract to become a reseller of the company's product." So I took the contract they had received, put it into AI and said: "Criticize this." I did not even need to read it first; I wanted to see the summary. Then you can decide whether it is worth moving forward. The terms were so bad that I thought: you will never, ever make a profit. You are just paving the way for the other company, without rights or exclusivity...

...and doing the initial, hardest work. Once you have demonstrated proof of concept, they can freely offer the same thing to any number of resellers.

It was probably a terrible contract. But I still need to add some nuance to the idea, because today you can criticize anything with AI. If I take an ordinary rental agreement, though I do not have a rented apartment, or say the contract from the bank...

...for the mortgage, remove the bank's name and say "summarize and criticize," AI will list ten terrible things. They are so bad that I would not dare sign.

Good qualification. But when I get the suggestions, I do not trust them blindly. It is a common mistake to just get a suggestion and stop there. I sit and bounce things back and forth. There are...

...pros and cons. Likewise, when I buy a product, I ask AI what its advantages and disadvantages are so I can weigh them up.

One hundred percent. That is also the point I want to make: if you are going to discuss a contract with an AI, it needs to get a balanced picture.

First, you can write: "Criticize the contract." Then, in another chat: "What good things are there about it?" In some way, that gives you a more balanced picture.

Or, like me, let different AIs discuss it with each other. They have statistical evidence that if an AI criticizes something it has created itself...

...it will favor it, even if it is more wrong. They have measured that. If you ask Claude to write a text and then criticize it, it is more...

...positive toward its own text. Not that surprising. Two different chats with different tasks still usually give fairly good nuance, even if it is the same AI. Your way is better, mine is a quick fix. Do not trust AI blindly, because nowadays when...

...we send contract proposals, my lawyer can see straight away whether the response is AI. There are things AI completely misses in your own... You can be bad at prompting. Shit in, shit out. You miss the bigger picture: what am I getting into, what do I have today and what could it be worth...

...with or without the partner you bring in. Whether it is me, you or someone else, the starting point is: what do I have today? Something worth zero, or ten million? Say there is ten million in value because I have built up customer lists and worked on the business, have inventory, built a site and have things worth ten million.

Without you, Fredrik, perhaps the company continues and is worth fifteen million in five years after hard work. Probably, but that does not have to happen. It could crash and go bankrupt, because startups and small companies always have high risk.

Now I bring you in, and you add value by developing the company with AI, with your chatbots that talk to each other. Then perhaps the company suddenly goes from ten to fifty million. The potential exists, even if it does not have to happen.

It could also be thirty or a hundred. Let us take a middle scenario of fifty million together with you. If I am willing to believe that and that it is possible, you have added thirty-five million, from fifteen to fifty.

Then of course I am willing to share half the company with you if you make that happen. Even if you only reach thirty-five, you have created twenty million in added value and are still worth the money. In a bear case, a bad scenario, you have still done a good job. Having an AI star on paper can in itself be worth a stake in a future sale, even if you have not touched the company.

Perhaps not fifty percent, but twenty, twenty-five, ten or five. It is worth something. That is what I try to explain to entrepreneurs. You may have product-market fit or an idea, but it is not worth fifty million when the profit is zero. You have a growing scaleup or startup, have done the product work for one or two years, or an old brand that has been going downhill for ten. It is not worth that much.

Look instead at the potential over the next three to five years if we work as a team and achieve the results, rather than starting too high in valuation. Then I immediately lose interest and can barely be bothered to discuss it with the entrepreneurs because they are so detached from reality.

There is a reality today and a reality we want to take the company to. We create the difference together. It is not something that already exists or is a given because we look at a particular curve. We are here now. Of course, I could be wrong, and the company might take off. Then they can do it without me, and I have not spent any time. Every yes to something is a no to something else. There is only so much mental space in an entrepreneur's brain, and it fills up quickly. Most of mine already goes to Västerbottens Såpa. The little that remains, I can use for leisure, hobbies or something else. There might be room for a project there.

I have said at various events that I am going to get involved in a health project this year, because I am so interested in it and passionate about it.

Are you discussing that now? We are. Hopefully, we will reach an agreement soon and can share something in the fall. One, two or three? No, I am joking, something in health, healthtech, life science or molecular biology.

That is exciting. You mentioned at the beginning that a lot can be measured, but at the same time so much cannot. For example, we can measure blood sugar with a CGM monitor.

But how do you measure hormones? I was at an investor event where one pitch came from a company building sensors to measure hormones as well.

Interesting. I see more sensors coming out. A new one measures gut flora. Gut flora has been strongly linked to brain health and longevity: more and better bacterial strains lead to better well-being, less inflammation and less risk of chronic disease and its consequences.

That was why the case with [unclear in the source: Albaht] was interesting. You test children's stool and speak with a dietitian and a doctor. I bought it immediately...

...and have since tested myself regularly. What does the gut flora look like? How does it change with kimchi, kombucha or more fermented vegetables? I have only seen positive progress from when I did not take care of my gut flora to now.

Speaking of measurement, I want to add why it is so important. I did a measurement and added those things, and mine went to hell instead. Okay. I had not told AI exactly what I had started taking. I asked why I was starting to feel worse and could feel it in my body, worse and worse. We discussed it and tried to figure it out. It asked: "Have you started having sauerkraut and apple cider vinegar lately?" Yes. "Okay, you have low levels of," I think it is called [unclear in the source: Dow], "so you cannot break enough down." Then I accumulate it and get too high a histamine load.

Because I cannot break down enough histamine, adding those things had a negative effect. I had to stop, and then I started feeling better. Then perhaps I need to find a way to increase the substances needed to break it down. I could not just add the supposedly healthy things, because things got worse. Yes, everything you put on or in your body is subjective. What is good for me does not have to be good for you. So: straight facts on the podcast.

Have you listened and thought about adding things like that? Do baseline tests and tests after adding something or changing your diet significantly. It costs a little, but then you have the facts on the table.

Supplements also cost money. That is how I usually justify it: buying a supplement costs money, but the blood test might show that you do not need it or should leave it out. For many people, for example, adding iron or selenium could be good.

Selenium is otherwise difficult to get through food. But I have an excess and need to avoid supplements containing selenium. You can easily get too much if you eat nuts, especially Brazil nuts. Two nuts a day can give you too much.

I [unclear in the source: är paranötter på något vän] but still manage it. There are other foods too, but normally it is difficult to get through a regular diet. Interesting to see.

That is on the side. Let us go back to the case we have just signed an agreement on, the poster shop. It could be interesting to follow. I will see if I can invite one of the founders, or all three.

That would be fun. We will see if we can do another zero-to-one-hundred journey. I am convinced, and so are they. I see it as high risk and high potential reward, as always. I usually see stackables as very safe bets.

What is a stackable? Såpa. If you have proof of concept with a slightly unique product, a high return rate of customers and good reviews, so the people who get the products love them, you get lots of repeat purchases without that much marketing.

But did you not say this case had few returning customers? No, I did not say anything about that. Did I say that? I thought so, but I could be remembering wrong.

I do not think I said that. But of course, a poster shop has a low repeat-customer rate, and it is difficult to make it high. It has also been around for such a short time that they have not had time to build it up. Low-risk cases are often stackables, like soft soap, or coffee if someone comes up with a better one. Now they have done that. I actually bought a coffee that is freeze-dried. They do something with the bean and freeze it where it is ground, or something with the bean. Then it arrives at your home frozen in doypack pouches, and you take a whole one like a puck. Is it tested for mold and heavy metals and things like that too?

Exactly, they have tested for that and microplastics. I was worried about that: what if there are loads of microplastics? It does not come as an espresso capsule with plastic around it; you can use the whole puck. Here you take the puck, add hot water and mix, and it makes fantastically good coffee. More flavorful than anything I have ever tried. I thought: wow, perhaps this works. I heard about the company on, what is it called, The Diary of a CEO, where Steven Bartlett has invested in them and advertises them. I am a bit of a sucker for that kind of advertising he does.

I use Wispr Flow, which he advertises a lot, and many of his other products. It works really damn well. So nice not to have to type and just dictate instead.

So the app means that when you are on the go or sitting in the car, not driving, but do not want to type too much, you can dictate?

It understands what you say in Swedish, English and other languages. It removes pauses, micro-pauses and little "ums." It almost looks as though you wrote it yourself. There is a little way to go, but it is good enough for me. It makes me more productive. I can do twenty to thirty percent more in a day with it, without...

Then one addition is that I have another entrepreneur acquaintance who has connected another piece of software to Wispr Flow. It provides different tones of voice. If he is writing customer emails versus a newsletter, it knows the different tones and changes how it transcribes. He can say one thing to Wispr Flow...

...and it changes the meaning a little. But that is already built in, is it not? Yes, but it is an enhancement. For example, I might rattle off gibberish in an order or structure that does not suit a newsletter or customer email.

Then it takes what Wispr Flow has done and rewrites it into a finished email in the right tone. Okay, fine. I think the built-in flow works fairly well. But he has made a more professional version for newsletters and customer flows. I spoke with John Ekman about six months ago. He had written a book...

...and I asked how he finds time for everything. He is also a serial entrepreneur, founded Conversionista among other things, and has the amazing event Conversion Jam, which I have been to several times.

He is really nice and a very good speaker. I said: you have written a book too, how do you find the time? I want to write a book but have never even sat down with the first page. He replied that he had a long car journey as a passenger, with a few hours to kill. Instead of listening to music, he dictated his thoughts...

...into a tool that took them and structured them into a foundation. He could use that with Claude or ChatGPT to write it up into chapters. Then he read and edited them into what he had intended from the beginning. He had saved sixty to eighty percent of the time it normally takes to write a book, because perhaps it was not his first. I thought: wow, so damn smart. Soon you will see me as an author. Bam.

For those who criticize that kind of content creation: I do not remember the name, but a very famous female author also found it difficult to sit and write. She had a cassette dictaphone, lay on the couch with it on her stomach or chest and dictated short stories.

Love it. That is different from asking AI to create slop right from the start. Here it was her ideas that she lay there dictating and that then became real short stories, with an editor who polished them and helped out. I get it. So cool. You do not really need more than that. But okay, that aside, let us move on. It would be fun to see the poster company turn over a hundred million in its first financial year. I do not think that will happen, but say the second or third. For someone in a similar situation to the poster company: what are the first steps you now sit down and take with the company, so others can...

...do something similar? A really good question. First, I review the whole flow. How does the customer find us? What do they buy? Do they come back, and if not, why? Ask questions all the time throughout the purchase flow.

What questions do you ask along the journey? How do people find us, and is it good enough? Probably not; it can be optimized. Once people have found us: what is the quality of the traffic? You can see that in the conversion rate.

High conversion means good traffic? Yes, but that is subjective depending on the site and the product. If you sell, for example, Twistboxes for ten thousand kronor...

...a good conversion rate is 0.5 percent. A product costing ten thousand requires time to think and several visits before a purchase. For an accessory, lifestyle product or poster, a good level might be three to five percent.

A stackable, apart from toothpaste, such as pharmacy products, haircare and skincare, needs high conversion. Order values are often low and products cheap, so the level needs to be higher to justify the customer acquisition cost.

How do you measure how they found you so you can increase traffic? I shopped at Västerbottens Såpa yesterday...

...and you have the form on the thank-you page: "How did you find us?" with different suggestions. It is important and provides both qualitative and quantitative evidence.

Qualitative in that they can also write a comment and think about it: how did they actually find us? Sometimes it looks as though the customer came from a Google ad. But that is not the reason; perhaps you talked about Wispr Flow on the podcast and the person then googled it. Then Google Ads appears as the source, because someone searches for what we talked about. It is difficult for the listener to click straight through from the podcast to Wispr Flow in a way that can be tracked.

You have heard about it on a podcast or from a friend over dinner. Mom says perhaps you should use Västerbottens Såpa. "Heard from a friend" is a difficult scenario to track. A post-purchase survey like that can make it clear that the friend influenced the purchase, at least the most recent purchase. But that does not have to be the only thing. A friend or a podcast like ours can get you to discover something. Then you do research, see ads and get help from them. Without the ads, perhaps you would not have bought. We can talk about something today, someone does research and then gets retargeted for...

...weeks before buying. Without retargeting during those weeks, perhaps they would not have become customers. If they do not become customers, they definitely will not become repeat customers. In the long term, you have thousands of kronor to lose if the advertising flow is not there. It is very important to think through the whole customer journey: what does it look like? The first step is information, and a post-purchase survey is a really important part.

That is one thing. Another is what we call abandoned cart, before the customer goes to checkout. I do not like disrupting the actual checkout. The question could be: "Why do you want to leave the site? What made you not buy from us today?" Ninety-five to ninety-eight percent of visitors do not buy. Post-purchase questions only reach the two, three, five or ten percent who actually buy. That is a small proportion, and not all of them answer either. Most actually do not answer, but some provide information. Asking before people have left is important too. It could be on...

How many answer? It sounds like: if you do not want to buy, fill out the form. People still answer; generally, they want to be nice. Not a large percentage, but with the volume of sessions, even a few percent becomes a lot. A small startup can easily get a thousand responses over a certain number of days. I know that...

...I was on a site recently. The menu said "New bestseller." I clicked and ended up on a 404 page. So some site is broken and does not work. Yes, it happens. Then the reason might be: "The site does not work, you dumbass." [laughter] Or: "Shipping time is too long." There are two approaches when setting up a flow like that. One is an open question.

You have a question and an open answer that someone analyzes. That is common and better; I will explain why. The other is predefined answer options and an open field at the bottom. Both provide qualitative and quantitative input.

With open answers, people used to think it was a hassle to analyze a thousand responses with misspellings and different ways of expressing things. Difficult to structure the reasons when a thousand people leave a thousand different responses.

Difficult to quantify, but with AI it is easy today. You might get slightly fewer responses, but those who actually take the time can have a greater impact on what you should change on the site.

The other thing is that just because something is easy to do does not mean the answer you get is true. With answer options, you might just click something: "I just want to get off the site. Get this shit out of the way." What is the reason you are not buying? I want to get rid of the question and press the first available option without even reading it. Send, then I leave. This abandoned-cart pop-up is a pop-up that...

...when the mouse moves toward the close button. It only works on desktop. On mobile, you have already clicked away, and most people visit on mobile. But there are enough desktop visits for it to be good enough for most.

We have traffic and conversion, but traffic quality affects conversion. How do you know where to look? I have seen the same page perform completely differently depending on the traffic coming in and how educated visitors are beforehand...

...which has an enormous impact. We could record and see what visitors did on the page, meaning the screen, but not their personal details.

Conversion can be ten times better when you educate them first. A longer ad tends to provide higher-quality traffic. Quick-click ads with little text, short content or an image with little information can get clicks, but do not deliver higher conversion when the customer lands.

With video, people might click away, but those who reach the site are more inclined to buy. Sometimes you need static images for fast, cheap clicks, and sometimes educational video for high conversion.

Traffic quality on the site can differ tenfold. I was at Facebook's offices many years ago. They gave a talk about how many people at the time optimized for how long people stayed on the site.

Then they said you are optimizing for retirees. Likewise, if you have an ad and think: "People did not watch it to the end." But that might mean they had already made their purchase decision after fifteen seconds, went in and bought. How long they watch the ad is not... It can get complicated. What data do you look at in that case?

I look at that too: bounce rate, how many people leave the site after arriving, and hold rate, how many watch a large part of the ad. Important parameters, but not the most important. It is the combined picture.

How many stop at the ad, watch it to the end or watch a certain number of seconds? How much of it is consumed? That is important. Then how many click through, and how many come to the site and buy on the basis of that particular ad.

You can isolate a lot here. Take a poster shop: we make a new poster and put it in ads. Different creatives, but they all drive to that particular poster. Then we see in total that we have spent, for example, twenty thousand kronor on five ads for that poster. It has sold forty-seven thousand, and then we can calculate the correlation directly. It is one new poster among five thousand. So in all likelihood, most of the traffic has found it recently precisely because we set up these ads.

So we have spent twenty thousand and sold forty-seven thousand, deduct VAT and get a ROAS of around two-something. Then we know the advertising effectiveness is 2-point-something. With existing products, it gets harder. Twistboxes can have a longer and more complicated customer journey that requires more work from a growth manager or strategist.

At least so far, I have not managed to prompt AI to provide the proactivity and strategic level that fully helps me. I wish, but we are not far off.

If we take posters, we then have traffic and conversion. What do you look at next, if you are going to scale a company where...

...their revenue has fallen? Then I think: how do we build a long-term brand and something sustainable over time? Posters people actually want, that stay on the wall for a long time. Can we do something? A lifetime guarantee is one thing. Can we create a level of innovation so people do not keep replacing their posters, but take them with them? I like that part. Is there something else you can charge a premium for?

Do you not like them replacing them and buying new ones all the time? No, because it is bad for the environment. I think a lot about the impact on the planet. If you want posters, buy them from us. If you want something on the wall, buy it from us. I want to compete for the space on the wall. But once you have bought something, I do not want you constantly buying new things from us. Someone, I do not remember if it was on the team or another team, said: "You should have a subscription so people get new posters every month." I thought: what a fucking bad idea, then people need to throw away quite a lot of...

...posters, with transport and waste. I do not like the idea. Even if it is commercially possible, I do not want us to take that path.

Some products are not suited to that, but it is interesting that you think about the environment even with posters. I think about the environment and people in every company. We only have one planet, and I want to look my daughter and son in the eye when they grow up and ask: "What have you done in life, Dad?"

Then I can say: I have built these companies that stand for something in life, and they have all had a positive impact. That is important. It may be less important to others, but for me it has been a core and fundamental principle in everything I do. A major reason for investing in Twistboxes was that the product is very energy-efficient. More so than any other roof-box brand I could find on the market...

...when I did my research. The fact that it folds up and takes up little space is also important when it is shipped. Whether it travels by ship from the factory in Asia to Sweden or the last stretch by air to a parcel collection point or by truck: the bulkier it is...

...the greater its environmental impact. That was important with Twistboxes. The same with Såpa: reused raw material, whereas everything else I saw on the market was virgin oil, even if it was flaxseed or linseed oil. Reused rapeseed oil, I thought, how interesting.

Can you make a good product from reused rapeseed oil? Now we are getting somewhere. They were also the only Swedish company I found doing it that way. I wanted to invest in that kind of production, not just as a personal impact case, but because it is economically sustainable.

It is also good for building a community around and provides greater returns in the long run. If you think about the best interests of the whole company, as you should in a limited company, the environmental aspect is often important. Many people can get behind the company, which I think is fun. It becomes a win on all sides...

...from doing the right thing. I also want to recommend the documentary The Devil We Know, about PFAS. Look on YouTube, where both the trailer and the whole film are available, for those of you listening.

That has to go in the show notes; I have not seen it. There you see how a company that does not care about it can negatively affect the whole world. I will say no more; just watch it. Good cliffhanger: a company that does not give a shit about the world.

Yes, they really have caused destruction. My God, there is so much of it. All right, good enough, good cliffhanger. I will watch it. Before we finish, we have a new segment. I sent you notes on WhatsApp. Did you see them?

I am opening them now. The question is which trend is most overrated: AI SEO, community commerce, live shopping, headless commerce, loyalty programs or AI-generated ads.

We have six topics that are widely discussed in different e-commerce companies. Many are in the process of introducing one of them, have already done so, or want to. You can choose one of the topics you think is most overrated and say why. Then I can choose one I think is overhyped and explain why. We are now primarily addressing companies with revenue of zero to a hundred million, or a little above, say zero to two hundred and fifty. Those are the core listeners and D2C brands, not multibrand stores, where it can be a little different.

Okay, what springs to mind? I look at all the bad loyalty programs I have been... or I am not even in them, but they are there: "Join our customer club." I am trying to understand how to do it well, because many people say it is good. I am trying to see how. There is surely some company I have missed...

...that has actually implemented it so it creates loyalty. Where does the loyalty come from? Good thought. What do you think of it for D2C? I fell for it too. In 2018, American or Canadian Swell Rewards exploded on Shopify. Everyone wanted it. It was easy to integrate, "easy" in quotation marks, and promised the moon. I fell for it too.

I tried it on at least two brands. Then Yotpo bought the company. They were big, and I thought it was an interesting combination: they are good at reviews, get loyalty and a bigger ecosystem. I loved, promoted and sold Yotpo into my projects. Then I saw their loyalty part go down the drain.

Over the years, I kept introducing it and saw it up close in 2021 and 2022 at D2C brands where I was also a co-owner. We paid a lot, both in startup costs to Yotpo and in ongoing costs and contract commitments. Then comes the most important thing, which I think people overlook: the mental effort. You only have so much capacity and can only get so many things done in a year. A loyalty program takes time. On that, I completely agree with you that it is overhyped, especially for the core listeners. It is not...

...in proportion to what you get out. Db Journey has done it well, I think, with early access to launches and unique VIP products. It is the closest I have seen to something in loyalty that...

No, I do not like it. I have bought the bag from Db Journey and do not need seventeen more. I am a member and get early access to a new bag, but I do not care.

They do not sell out anyway, so I can be third in line. That is okay. With certain other products, you really want to get in before everyone else.

For me, for example, it is local eggs that always sell out. I am a member of a grocery store on Södermalm in Stockholm, with Sébastien Boudet.

To support the local community and the farmers supplying the food. We know that there are only a certain number of eggs each week and that they are reserved for members first. That is a great benefit to me, as someone who eats eggs almost daily.

Is that a loyalty program or a membership? A membership, but like a loyalty program. You are invited to dinners and other things and get discounts on products. It is also an expensive program.

Then there is NK-nyckeln, with different colors depending on how much you have shopped. Does that count as a loyalty program? Yes, but not a good one. There are good examples. I think Åhléns has a good customer club. SAS is perhaps the best, EuroBonus. I think the program is valued higher than the whole airline.

The EuroBonus program has an insanely high value. People collect points on all sorts of things to exchange them for flights. It is the kind of product where there are only a certain number of seats on a plane, often in a higher class. Then they are gone, and you have to find another plane, and so on. It is a perfect product to build a loyalty program around. But far removed from what we are talking about here: D2C selling zero to two hundred and fifty million. There, I agree with you that loyalty programs are overhyped...

...for smaller companies. For Åhléns, NK and SAS, it can work well, when people return to the store every week or month. Which of the options do you think is overhyped?

Difficult. You are on the right thing. But right now, AI-ifying the whole world is also overhyped. Everything has to be done with AI. Most of what is released in AI advertising is rubbish and nowhere near adequate, perhaps three out of ten. When you scroll TikTok...

...or Instagram Reels, there are loads of e-commerce gurus recommending automating the entire flow. One AI for social media, one that creates ads, and so on. But what comes out is so bad. Shit in, shit out. Most people who start with the recommended flow get something worse than mediocre. I would happily say mediocre ads; that would have been okay here.

But they are worse. I am not saying it cannot be done; I am one hundred percent convinced and have seen that AI can create good ads automatically while you sleep. But it takes more than the TikTok gurus show. Everyone seems to have the same script, publish the same apps and use the same UGC flow: someone gets a product in their hand and has to recite a script that feels worse than mediocre. Right now, I think it is extremely overhyped.

Not because the value is absent or it is impossible, but because it is harder and costs more time, tokens and money than what is shown.

Then my second question is: does it have to be better than mediocre to perform? Yes, mediocre is by definition about average. To perform, you need to beat the average; otherwise, you are competing with those who have much bigger budgets than you. Of course, you can still get results at a low level, but it costs a lot of money. It is not something we usually recommend on the podcast.

The example I am going to give is actually a bad one. In the US, there is a company with a guy and his brother, I think. They are turning over more than a billion dollars this year and sell some weight-loss drug. They have created six or eight hundred fake pages with fake AI doctors...

...through which they produce ads. It has worked for them, even though ethically and morally it is very bad.

It has performed. I have seen similar things for impotence and other things on the US market, including Swedish entrepreneurs with dropshipping and warehouses in Florida. It is easier to achieve that kind of short-term scaling in the US than in Europe. Laws and rules are a little vaguer and easier.

I know they have been reported in the US by the equivalent of our medical products agency and might lose money. I have not seen their ads, but can almost guess that they are better than mediocre and that the eight hundred pages are relatively good. Are they not?

No, not for the American market. You have to consider that half the people in the US do not even know where Paris is or that it is a city in Europe, so...

I love the prejudices. [laughter] Yes, many do not even know how many states the US has or who the president is.

Do they have Trump? Yes. These are not prejudices but facts. When such a large proportion knows nothing about the outside world... [unclear in the source: för de ser ju sa ganska bra ut.]

My strong take is that it works there. In Sweden? That remains to be seen. I have no data and cannot...

...comment. I have seen it up close and tested a little myself. A talented Swedish woman moved to the US and made similar sites in impotence, with traffic through affiliates. She got going and could make a good enough living to build her own real brand. Then things went the way they did. I do not know how things have gone since we last spoke; it was a long time ago. But it is possible in the US.

I want to add that a Swedish brand with a Swedish product almost, I think exclusively, uses AI ads now, and they perform well. Yes, I am not saying they do not exist.

That was not what I said. We were talking about what is overhyped right now: the approach and the execution. For the average person, it is overhyped, absolutely, in terms of how it is done and sold in.

That was the last segment for the week. If you think what we are saying makes at least some sense, send the podcast to a colleague, friend or someone who wants to watch or listen. Full episodes are now on YouTube, and soon there will be more and better reels, more often, on YouTube, TikTok and Instagram. Subscribe, and if you like the episodes, review and rate them. That gets us back on the morning couch...

...preferably five stars, if you think it is good, of course. Then we will come back week after week. It is almost summer. We will record a few more and then see when the break will be. After that, we also want suggestions for interesting guests.

Great fun. Thank you, Fredrik. See you next Thursday. Bye.