Ben Cogan @benjamin_cogan

Very soon, AI will let brands know all publicly available info about their customers. This is going to revolutionize ecommerce. I caved and subscribed to ChatGPT Pro for access to “Deep Research.” This is the tool that scours the internet to write a 5-20 page pHd-level report on a topic of your choice. As many have said, it’s great. What I haven’t seen is someone using the tool for deep research about individual non-famous people. So I tried it out on myself. Here’s the report: chatgpt.com/share/67a39906… Honestly while it’s not exactly what I’d write about myself, it’s very good. It unearthed a couple of interviews with me about my first company, Hubble Contacts, which anyone could find on Google. But it also went back to a 2012 Princeton University article about me which I didn’t even know existed. Overall, I’d easily give it an A. This got me thinking about what you could do with this tool. Because while Pro users are limited to 100 Deep Research searches a month, I’m sure very soon the cost of this will approach 0, effectively allowing for unlimited searches. What could you do with that? There are some obvious use cases. For example, if you’re a sales representative getting on a call with a new lead, wouldn’t it be helpful to get a 500 word summary of who the lead is? Their work history, where they went to school, where they live, even their hobbies? Or what topics they’ve written about LinkedIn (this is getting meta)! Of course, this has the potential to be really weird, so sales reps will probably be trained to not make what they know about someone *too* obvious. But I think this is going to happen. How might this affect ecommerce? Well, maybe the days of one size fits all email campaigns will soon be over. Talk about segmentation–in theory, you could write a personalized email to every single customer in your database. Do you run a sports goods store? Maybe one customer is skiing in their public Facebook profile picture, picked up by Deep Research 3. Boom, they get an email about new skis that just went up for sale. You can take this even farther. With a tool like Retention, you can often identify a site visitor’s email address without them explicitly giving it to you. If Deep Research were fast enough, could you determine who the shopper is from their email and personalize your site itself based on their publicly searchable interests? Why not? Again, the above techniques clearly have the potential to be creepy. And, if pushed too far, they could even violate data privacy laws (read: do not try in Europe). So once these tools become widely available, I’d recommend that brands start slow. For example, maybe a bag brand emails customers offering to customize their bag with their school logo. Risky? Maybe a little. But probably worth trying. Deep Research heralds a brave new world where companies can immediately know almost all publicly-available information about their customers. Buckle up.

@danfleyshman Yes that would be great! Will DM you.

Ever since I began acquiring DTC e-commerce businesses, people have asked me which business categories are “the best.” But I’ve concluded there’s no such thing as a universally good or bad category. I’ve heard the usual advice: look for categories with 60%+ gross margins, have a high average order value, minimize SKU complexity, feature lightweight products for cheaper shipping, etc. In my experience, though, these factors rarely—if ever—predict success. Why? Because any advantage that makes a category appealing draws in more competitors, quickly eroding those advantages. Meanwhile, businesses in categories with difficult challenges often turn those challenges into competitive moats, creating a powerful point of differentiation for those who master them. Some examples: I often hear that apparel is a “bad” DTC category given SKU complexity. And it is true that apparel businesses constantly need to produce new styles, colors, sizes, and products. This is expensive and capital-intensive. But the businesses that get it right have incredible defensibility–think how difficult it is to replicate the thousands of SKUs required to compete with the best DTC apparel businesses like Bombas (socks), Vuori (athleisure), or True Classic (T-shirts). Their complexity is a moat. Furniture is also supposedly a “bad” ecom category. Shipping a couch or bed frame can cost hundreds of dollars. True, but then consider how, for many consumers, having a 100-pound box shipped straight home is easier than lugging it themselves from a store. Wayfair, the ecom furniture company, is still worth nearly $6B. Or consider categories which many merchants want to avoid because of onerous regulations. One category I know better than most is contact lenses, where the seller must validate every single customer’s contact lens prescription. This is burdensome and complicated but, once you establish proper compliance, serves as a meaningful barrier to entry for merchants who can’t figure it out. 1800 Contacts is a $1B+ defensible ecom business because they got it right. Once you start thinking this way, you realize that it’s often the unloved “bad” categories that offer the best opportunities. For example, I remember a conversation in 2017 with the CEO of Dame—a brand specializing in women’s “adult” products—about marketing. Because of what it sold, Dame was banned from advertising on Facebook. No Facebook? Talk about a tough DTC category. But this forced Dame to figure out more creative, organic ways to acquire customers. As customer acquisition costs gradually rose on Facebook, eroding margins for brands in other, “good” categories, Dame thrived because its organic acquisition funnel insulated it from rising marketing costs. The moral of the story is: the ecommerce market is more efficient than you think. There is no such thing as “good” or “bad” categories because the categories that are tougher for you to crack are also tougher for your competitors.

I usually write here about DTC and e-commerce, but this post is about something far more important. It’s the 15th anniversary of the founding of Giving What We Can, an organization made up of people who have pledged to donate 10% of their income to the most effective charities. Recommended charities include the Against Malaria Foundation, which provides bed nets to people at risk of malaria, and GiveDirectly, which gives unconditional cash transfers to some of the poorest people in the world. I heard about GWWC right after its founding and soon joined as Pledge #117 in 2011. Since then, GWWC has grown significantly and is now approaching 10,000 pledgers. As the organization scales, I was fortunate to join the U.S. board of GWWC earlier this year. It’s an amazing organization full of incredibly generous people trying to make the world a better place. I’m looking forward to @givingwhatwecan hitting 10,000 10% pledges soon. If you’re interested in learning more about the 10% Pledge (or a less demanding Trial Pledge), please send me a message.

@dollarcommerce Just builds trust over time (did with us at least)

@dollarcommerce 100%. The agencies who took short term hits because they were honest when things weren't working made so much more from us in the long run because we trusted them.

@dollarcommerce Definitely Julia Roberts but only if she started a DTC biz. Otherwise, $10mm bootstrapped biz.

@dollarcommerce Also just running brand campaigns they claim will get you sales years down the road.

@ankurnagpal 💰 thanks!

Legend

@lifeofbi You're becoming the new me!

@lifeofbi 😬

A DTC ecommerce brand’s data is valuable, but only to the brand–not an external acquirer. I remember when Unilever acquired Dollar Shave Club for $1B in 2016. While DSC was massively unprofitable, a large part of the acquisition rationale was not only DSC’s revenue and team, it was the inherent value of DSC’s “data.” Techcrunch, reporting the acquisition, said “Unilever will also benefit from Dollar Shave Club’s valuable customer data and existing customer base.” The idea here is, before 2016, Unilever didn’t know much about their customers because these customers bought nearly all Unilever products through retailers, not directly from Unilever. Because most people buy anonymously in stores, even the retailers don’t know much about their customers. But DSC knew everything about the DSC customer from exactly what they had purchased historically to their home address. This data, supposedly, would benefit Unilever. Unfortunately, it doesn’t seem to have helped much. Unilever sold DSC last year for an undisclosed (read: very low) amount. What happened? Chalk me up as someone who actually does believe that DTC data has real value, even or especially for an omnichannel business. For example, it’s much easier to test new creative on Facebook than in a retail partner’s end caps. Or A/B test pricing. Or even pre-launch new products and gauge demand. All of these learnings can be applied across every channel a brand operates in, not just DTC. Or consider Warby Parker which, famously, uses DTC sales as a lodestar to determine where to next open a retail location. It makes sense: probably the best place to open a store is where you already have a lot of online customers. So DTC data can have tremendous value. But notice in the above examples, the data is valuable to the brand because they can use it to make better decisions for *their* business. This isn’t what Unilever wanted with DSC’s data. The idea was that somehow, DSC’s data would inform how Unilever should make decisions about its *other* brands. This ended up not being true. They may have similar customer bases but DSC data doesn’t tell you much of anything about other brands owned by Unilever like Axe Deodorant, let alone Hellman’s Mayonnaise. Why? Because they have totally different customer bases. The best way to learn about Axe customers is to talk to Axe customers, not assume DSC data will inform Axe decisions. Part of the runup in DTC valuations in the mid 2010s was the hope that acquirers like Unilever would acquire DTC brands for assets beyond just the financials of the business. One of those potential assets was the brand’s data. But the DSC acquisition is a cautionary tale. While DTC data does have real value, its primary–potentially only–utility is to the brand itself, not an external acquirer.

Something I’ve been thinking about a lot recently is what is the best way for ecom business owners to incentivize the CEOs / general managers (GMs) of their businesses. At Agora we’re acquiring DTC ecom businesses. It would be impossible for my co-founders and I to run all of the business ourselves, so this question is directly relevant. But it’s not something I feel like we’ve totally nailed. This would be easier if we acquired minority stakes. If the operators still own a majority of the business, they’re pretty incentivized to manage it efficiently. But we usually buy 100%. One obvious answer is to pay GMs, in addition to a base salary, a percentage of profit (net income) bonus. This is what we’re optimizing for. But this has several problems: Problem 1: Ecom businesses can often maximize short term profit by starving the business of investment. For example, turning ads way down or doing no new product launches will likely improve profit in the immediate term. A GM planning to stay with us for 3+ years would feel the pain of these choices in the form of lower profit in future years. But it’s (understandably) hard for a GM to know if they’ll be with us that long–maybe they should take a short term bonus. On the other hand, maybe a business has reached a point where there actually are few profitable expansion opportunities, and the best thing to do is milk the business for short-term profit. How can we know unless we’re running the business? Problem 2: The businesses already have meaningful net income when we acquire them, so we’d be giving GM’s relatively large payments even if the business is flat. So maybe we should pay GM’s only a percentage of net income *growth* each year? The issue with this is ecom businesses are incredibly volatile and sometimes it’s actually a win to keep profit flat (e.g., year after iOS14). Should the GM get no bonus in that world? Also, what about when it’s 6 months into the year and net income is trending below last year and so it’s very unlikely GM’s will make any bonus. The GM’s then have no incentive to make sure they’re only 10% down vs. 50% down. That seems like a huge problem. Problem 3: Sophisticated GM’s can boost net income in a way that’s detrimental to a business. For example, a GM could make the decision to purchase a ton of inventory and ensure the business is never out of stock on any product ever. Even assuming all this inventory eventually gets sold, using accrual accounting, the net income of the business doesn’t reflect the cost of the inventory until it's sold. But this could be months or years later and until then the inventory is a massive cash drag on the business. But GMs have a strong incentive to do this. The above are just the problems with net income-based incentives. There are different but comparable problems with growth-based, cash flow-based, or combo-based incentives. This is genuinely a puzzle. What do you think the best incentive approach is?

@allenwalton Think it's especially hard because you don't need technical skills to start an ecom business (unlike a software business) so you're basically competing against 100x as many people.

@bzises Extension

@lifeofbi What about net income margins? Taxes and interest count!

@rogueleaderr Definitely part of it but Facebook has largely recovered.

I wrote last year about how poorly DTC companies were performing in the public markets. Unfortunately now, with some key exceptions, it’s worse. Here’s a representative sample of public DTC brands, the same as last year. - Allbirds (Footwear). Down 98% from peak.

@StevenPDennis Definitely agree a lot of it was VC overfunding. Nearly all of the businesses listed in the post were VC-backed.

@msjpenn I think overhead definitely needs to be cut in many cases but also think the marketing spend for many of these brands is unsustainable.