How to split the pie: Shaping Web3 Advertising market under fair competition 🧵

Earlier, I discussed why there will be no ad giants like Google or Meta in Web3. But this left unanswered a big question—what are we going to see instead?

Let's address the elephant in the room 🐘:

Here is the beginning if you missed it.

TLDR: The primary reason for it—the wide accessibility of rich on-chain data and the lack of control in the hands of a single entity.

https://twitter.com/acecreamu/status/1623748389870657536

What defines the desire of a publisher to work with an ad provider? The price it can pay. What defines this price in Web2? Data. Let’s take it out of the equation. What’s next?

1. The size of the network

A wider availability of users and ad campaigns leads to more efficient optimization through finding the most tailored offers for each individual. Same as a network will always be more effective than a 1:1 partnership, a large network will be more effective than a small network.

But can one really win the network?

The interesting detail of traditional ad tech, is that publishers always have multiple ad providers (ad networks) integrated or even put the impressions out on larger cross-network auctions through SSPs.

Simply speaking, for a publisher, it’s always beneficial to have multiple bidders, not only inside the network but even at the network layer, as apart from the obvious competition in price it creates a backup ad source and helps to maximize the revenue from a single user.

Fun fact: As the user sees more and more ads, they exhaust the network’s ad demand, and the price of impressions goes down, so, it makes a financially sound decision for a publisher to sell this user as a fresh new user to another ad provider even if they will show the same ads!

It’s hard to imagine that publishers will suddenly become monogamous in Web3. But they do have a reason to put more traffic into a provider who pays more. Back to our equation, how can anyone pay more, if publishers and advertisers can be matched as well?

2. The answer is in algorithms

When everyone has the same data and the same networks, wins the one who can process that data more efficiently and understand the user intent better.

This includes multiple layers, from wallets analytics and predictive profiling to more traditional ad campaign optimization, user ranking, and audience optimization. In the end, leading to one result—higher CTR that will create more value for advertisers  and will capture more $

Fun fact: at times when we don’t see many AI/ML experts in the Web3 marketing space, @slise_xyz is in the unique position of being founded by a team led by an ex-Meta AI Researcher who has also built AI startups before 📈

Conclusion #1:

• The specifics of the advertising market will not allow the formation of mono- or duopoly in the environment with free-flowing data.

Conclusion #2:

• We will see a fair distribution of traffic and revenue between multiple strong players with their shares being weighted based on their algorithmic advancement.

Now, it’s time for you to tell me where I'm wrong 😉

But if you liked this piece, you can read this and coming startup-building thoughts in a human-readable format at my substack: https://acecreamu.substack.com/p/how-to-split-the-pie

How to split the pie: Shaping Web3 Advertising market under fair competition

Why there will be no ad giants like Google or Meta in Web3 🧵

What if there were no data walls?

Can we possibly build a fully decentralized ad exchange that connects publishers, advertisers, and end-users directly?

Let's find out:

As I write it, 3 companies, namely, @Google, @Meta, and @Amazon, control the majority (54%) of the digital advertising market.


They are followed by other consumer giants like Twitter, LinkedIn, Bing, TikTok, you name it, and only a tiny part of the market (<5%) is designated to independent ad networks, not associated with any consumer app.

Why is it so? The answer is in data.


Consumer products aggregate first-party data they receive from their users and create walled gardens for advertisers, where the latter can only access this data for advertising purposes through the native advertising solution of the given company, only within their product.


In Web2, this leaves no space for ad networks to compete, as even with all the cookies and shady tracking, they hardly can get users’ geolocation and rough demographics, not even mentioning interests or buying intent, making targeting super inefficient.


Soon, in 2024, the situation will get even worse as Chrome and Safari will stop supporting 3P cookies, basically destroying independent ad platforms and strengthening their oligopoly.


The walled gardens of data are what protect the dominance of the centralized ad giants, and their walls are only growing over time.


In Web3, anyone can access on-chain data, because it is on a chain.


Moreover, this data on a chain presents the lump sum of user activity and interests in the context that may be relevant to web3 advertisers. Making any additional first-party data marginally useful, and leveling the playing field.


Say, now, some Web3 clone of Twitter wants to start selling ad surfaces. They can create their own product, which will be the same as others, maybe less sophisticated.


Or, they can join the existing network with an efficient and highly-optimized exchange and advanced algorithms, operating on top of the same data but across many more sellers and buyers allowing our friendly social platform to bring the highest-paying advertisers for their users


The network quality and algorithms are the cornerstones of ad giants of the future. And not an association with a particular consumer interface.

>At 8:29 @friedberg shares how the efficiency of Google’s ad exchange helped them build an actual monopoly around their ad network and lock the publishers in.

https://youtu.be/CwQMMzoeH9s?t=509 @theallinpod


If to fantasize, one can even argue that ad exchange is just like any other exchange. You have the bidders, you have the askers, you have unique impressions enriched with the right data, that can be matched with the most relevant and highest-paying advertiser mathematically.

Meaning, it can be standardized and decentralized while all the operations being performed by smart contracts and not a single entity.

Even more, don’t forget that the data access is also permissionless, giving us two big terms of the equation, that can operate in a purely decentralized way. Allow publishers to deposit the inventory, and advertisers to create campaigns, and voilá, your AdChAiN is live.

Maybe there can be pools of unique data or analytics engines or pools of advertiser’s reward tokens, or maybe even end-users can put their 1P data in pools of particular advertisers they trust in a DataDAO manner and earn fees from ads like they earn fees from swaps on Uniswap.

In either case, the speed and cost of the transactions won't allow implementing anything like that today. Considering that each transaction has low value but needs to be fast and cheap, maybe there is a space for new more extreme roll-ups or L3’s specifically for adtech purposes


What we can be sure about, is that the future of advertising looks exciting.

That’s why we build @slise_xyz. The ad platform for web3, serving better ads for a better internet of the future.

Be the first to read this and new pieces on my substack:
https://acecreamu.substack.com/p/decentralized-ads

Why there will be no ad giants like Google or Meta in Web3

Investing in founders and not in companies, how do you like the idea?

The math behind really makes sense.

Investing in the founders is indeed less risky than investing in the companies—they die less often, but win just as much 💀

🧵:

Similar ideas have been in the air for a while, mostly in form of turning an hour of your time into stock (that should appreciate), but this project does it differently—you can sell a share of all your lifetime assets 👀

And I don't know what to think about it

On the one hand from the venture point of view the motivation it's not much different from startups

You get capital upfront which you can use to accelerate growth, scale more aggressively, or capture the market, leading to higher revenue in the long term than if you bootstrapped

95% of 10M is more than 100% of 1M

In this context, giving away some share of yourself (😬) to get early leverage looks reasonable

But why the heck does it feel so wrong!?

Emotionally it feels like selling your soul

How to sleep at night knowing that all you have is not fully yours? Could you ever be self-employed, or there is no more "self", but only a board of directors?

And if by selling 100% you become a slave, this way you're a 5% slave?

But on the other hand, taxes didn't go too far from it. We are ok giving away 20-40% of all we have for our whole life and sleep tight knowing this

And say it will be 21% or 25% instead, who cares, but in return you get capital that you can invest in yourself as you wish...

It's a really nuanced ethical question, would be curious to know what thinkers like @naval think about it, and do reply with your opinion!

The funniest part, prepare for it.. Is that it's called HUMANISM *mic drop*

Source: https://humanism.is/

Investing in people - Humanism or not himanism?

Earlier we got backed by @cory who is also organizing Zfellows.com—a program for aspiring founders to work on their project for a week and meet mentors like CEO of Netflix, CEO of Figma, @naval, and others

Last week I participated in the program, here's how it went🧵:

The program is kept low-key and is not very intense, it aims to inspire, more than anything else, so every day there was just one joint morning call and one fireside chat with a guest

All the participants are young but very impressive. Many students from @MIT and @Stanford doing early-stage projects, someone who already went through @alliancedao, someone who worked in @SpaceX, someone who was invited to join @Apple straight out of high school—12 ppl altogether

The best advice we've got in the beginning was to ask guests tricky hard questions you won't hear on youtube. It really made meetings more valuable and memorable (no guests' feelings were harmed)

Now, the guests:

@edwardlando founder of @misfitsmarket delivery service soon approaching 1Bn in revenue. Edward has tried hundreds of different things and still prefers to work on a few projects in parallel. Main takeaways:

- The secret to managing a lot of tasks and not burning out is using leverage and hiring good leadership teams, not trying to do everything yourself

- Doesn't matter how many things you've failed, people will forget them all as soon as the first one succeeds
@edwardlando

@kevinhartz angel investor in PayPal who've sold a company for almost a billion and co-founded @eventbrite

- If the difference of IQs in a couple is greater than one standard deviation, the prob of divorce is ~100%

- Choosing a partner is no different from choosing a co-founder

@rabois member of PayPal mafia, GP in Founders Fund, and co-founder of unicorn

- Neither he nor Peter Thiel believes in AI or technology in general, they believe in solutions to problems and value creation

- He doesn't see it now in #web3

- The main trait of Peter Thiel is an ability to extrapolate one data point to a whole picture (+ recognize the best talent early)

- All founders they invested in are outliers by one (not always obvious) axis. Ordinary people do not change the world
@rabois

@adamguild founder of owner.com who dropped out of school to build his company and now makes millions in revenue at the age of 23

- Believes that startup building is a competition that you can win only by giving your all. Only hires ppl who understand this as well

- Re burn-out: the ability to control own mental state is the main skill for a founder, and he has derived a methodology to stay positive and motivated

- Why not be a founder: higher prob to build wealth by being an early employee; the process is harder than wildest expectations

My main personal takeaway is how normal and humane those "legends" are. They read emails, argue, and joke just as me and you, and the best attitude for them is a normal friendly attitude. It will affect my empathy and how I communicate in the future

Great experience. Thx @Cory!

Z fellows experience

I've been working on #AI in the best #research labs like @MetaAI and have 200+ citations incl @MIT

I've been also bringing AI to the thousands of users in @get_suggestr #product

The gap can't be larger!
Here are 8 important lessons, every ML engineer need to know to become CTO:

1/ No one cares how big your NN is, how innovative the approach is, or what's the cross-val split. Basically, anything that will fail your paper review means zero in real life.

The only question business asks is "will it work or not?"
(if yes: what does it take to implement?)

2/ The easiest way to make smth work is to use smth you understand very well and can deploy fast.

Hence, the best solutions irl are often simple algos, hard-coded heuristics, and off-the-shelf solutions like @huggingface

The fewer points of failure it has – the better.

Pro tip: very complex algorithms that are used as a black box to produce primitives (e.g. embeddings from #transformers) work very well in a combination with simple logic on top of them.

3/ The easiest models to debug are the ones that are simple and that you understand very well.

If the good model fails in 2% of unpredictable cases and crashes – it can't be used. Less accurate but bullet-proof heuristics will win.

4/ There is no such thing as a controlled environment!

Everything should work for any user at any time of the day. Forget about the parameters that have to be adjusted or limited knowledge domains. The world is unlimited, and you should be prepared for that.

Pro tip: we did everything to go away from absolute values to sorting, rankings, and percentiles because these are things that let you sleep well at night.

/5 You should know the problem user solves better than a solution!

Only by going deep into the domain and trying to solve it from the first principles, you'll be able to produce the best solution, account for all the edge cases, and see the whole picture.

Throw away all the variables and parameters you don't understand. Here less is more.

For ex, we would recreate CF algorithm and find the best variations for our use case. It's like understanding the code VS  what it does. You should be able to explain your algo without formulas.

/6 Everything has its price, so you can't research endlessly for the sake of research.

The last 10% of accuracy takes 90% of the time, so try to concentrate on the first 10% of efforts and focus on the impact/effort ratio and not AUC. Use the elbow method.

7/ ML doesn't exist in a vacuum. To train the model you need data, to store the data you need storage, to run the inference you need resources, infra, back end logic.

You can't add @PyTorch model when the first one is on @TensorFlo because dependencies will brutally kill you.

You can't use any data, you can't utilize all the resources of the server, and you can't afford crashes because it will take down all the front end (ideally you don't build it like that).

You should think about your neighbors because a hole in a wall is a problem for all.

8/ ML doesn't exist in a moment of time. And it runs more often than just during train and test runs. If it won't run at night then you will wake up and run to your laptop.

Production algorithms need maintenance, and the less maintenance your model needs – the better.

In a conclusion: a business doesn't value the Novelty, Coolness, or even the Beauty of your algorithm, what it values are the Simplicity, Transparency, and Effectiveness of your solution times the Cost of deployment and maintenance

The same btw can be said about the engineer 😉

ML in Research VS in a Startup

One could think: here you got a blockchain, all data is public, you can see every txn on @etherscan–everyone should know everything

And yet, if to ask any creator what's your token retention, what's conversion in the funnel, or what's your target audience, they can't answer! 🧵:

2/ "Ehh, the community is active", "Meh, volume is growing", "Uh, the floor price is X", and indeed, the floor price is not a big secret, you can find lots of market data that is useful for traders.

But is it as useful for creators?

3/ What's the token price for a creator? It's an alternative to their stock price. So you tell me the founders know damn well how their stocks behave but don't know much about the product itself or the users? 

4/ Where is the event-based tracking? Where are look-alike audiences? Who are the target personas? What's your funnel bro? It's like if web2 product managers were studying @Similarweb instead of @googleanalytics or @mixpanel

5/ Why #web3 marketing of the future is at the level of the 2000s? 

6 / My speculations: 

- The market is immature, the intelligence which comes with scale is not needed at the bull times when people put money in your pockets for promises

- Web3 is allergic to targeting and any tracking. You can't reach the right segment even if you'd find it

7/ The reasons can be different, but one thing is clear–as projects get more mature and serious, people start thinking more about measuring and controlling what they do (sounds ridiculous to argue about importance of it in web2 🙄)

The only question–how soon will we get there?

State of web3 analytics

If you're looking for a #startup idea or trying to grow your #idea and it doesn't grow, here is one most important thing you should remember:

For the majority of successful companies, exponential growth started after 1-3 years of working on the idea.

Why it is important 🧵:

If you aim for overnight success your mental detector will be distorted and will detect hypes and trends instead of long-term problems.

Even worse, it will challenge your self-esteem, if you're not down from traffic right after launch. 

This is not how great companies are built

Expectations management is crucial. If you and a team understand that it may take a year to find your place on the market – you will act very differently from when you expect instant victory or failure.

Same with #investors. A good investor will not expect you to know all the answers or show the hockey stick at pre-seed. 

People who've been there and who aim for a long-term partnership will look at completely different signals. 

// Inspired by @sinasanm from @learninglooporg 

Main idea search mistake

I almost lost my hope in accelerators, before I found these guys, who believe in teams not only with % they take but also with $ they put. 

So, my top-6 of pre-seed accelerators that accept startups at the idea stage and give 6M+ valuation 🧵:

1/ @a16z START: https://a16z.com/programs-a16z-start/
1M investment, focus on talent and very early stages, 11 teams have been taken so far. Apps are on a rolling basis. 

2/ @sequoia Arc: https://arc.sequoiacap.com/
1M investment, 15 companies in a batch, mainly oriented to Europe. Apps are currently closed.

3/ @Accel Atoms: https://atoms.accel.com/
250k investment at uncapped note! Which means a much higher valuation of the next priced round. Focus on India and SEA. Cohort based.

4/ @pearvc Accelerator: https://www.pear.vc/accelerator
$500-750K at 10M cap. 15 teams in a batch. Quite a reputable fund. Cohort based.

5/ @AforeVC Alpha: https://www.afore.vc/afore-alpha
1M at 10M post-money SAFE. No company is too early. Response within a week. 

6/ @hyper_: https://hyper.com/#apply
300k at 6M post (could be better). Just 4 weeks (could be better). But they claim a very individual approach and elite network of operators.

7/ Honorable mention: @ycombinator (the link you know better than me)
Still the best in the industry, and with their new deal selected companies can effectively get 500k at 6M+ cap, while realistically it will be more like 5M (125k at 7% + 375k at val of next round). 

8/ Honorable mention: friends and family

Jk, they are as broke as you are. But they may give you 20 dollars at 1B val if you ask nicely. 

Top-6 pre-seed accelerators

To the question why do I write and share everything with the public. 

I started without any intent, the tg channel was just my learning diary. Over time I started seeing objective value in it:

1/ It helps to structure my thoughts and offload info from RAM

2/ It allows me to share interesting and unique information that otherwise would be wasted. The feeling of wasting something valuable makes me really uncomfortable. 

3/ It helps to build a reputation. Again, doing something great without anyone appreciating it is wasted potential.

4/ It allows me to be discovered and creates new contacts. The most interesting people I know I met through the channel.

5/ It improves my writing. Writing in English also improves my English. 

6/ Sometimes I receive valuable feedback and insights from the comments. Most of the time the comments are so dumb tho, I just can't 🙄