Marketing analytics and proper attribution are getting 🥀 Another trend I’ve been noticing lately is that traditional marketing analytics is becoming less and less reliable. First, user behavior itself is much more complex now. A funnel where someone sees a post on X, Googles the product, compares options with AI, reads Reddit reviews, and then shows up as Direct or Organic is completely normal today. And on top of that, we now have AI traffic and all the instability that comes with it. Here’s a fresh example: In October, ChatGPT stopped passing utm_source and referrer data properly for a few days. In GA4, it looked like ChatGPT traffic had suddenly dropped by 22–57%, while in reality a significant share of those visits had simply moved into Direct. Plus, more than half of web traffic is now bots, which means standard traffic numbers are becoming less and less representative of actual human behavior. So in 2026, it’s already risky to look at one dashboard and conclude that “this channel is growing” or “this channel is dead.” You need to look at multiple signals at once: direct traffic, brand search, MMM, actual LLM answers, and conversions. And honestly, the good old “How did you hear about us?” question is starting to look less ugly every day :)

Wanted to give you guys a quick rundown of what CopyCheck can do: • My holdings: check current value, unrealized gains/losses, and break-even prices. Try different market moves to see how they’d affect your bags. • Review wallet trades: review up to five @solana wallets over a date range of up to 90 days. Filter losses and inspect individual buys, sells, and transaction links. • Compare two wallets: see how entry and exit prices affected returns on matching eligible trades. You don’t need to connect your wallet (I think these days this is the real moat). You can also view token market snapshots, open chart links, and export your reports (FYI, all trade results are shown before fees) The video shows the wallet-review flow, but the article covers the calculations and limits in more detail. Let me know what you think!

It's a nice article on how they scaled the distribution, but it mentions nothing about: → Pricing → Free-to-paid conversion → Churn → Creator costs → Ad spend → Profit margins How precisely did they turn all that traffic into $150K/month in revenue?

There’s something I’d never really considered: Now that Meta, Google, ChatGPT, etc. are starting to roll out AI Ads, what are the chances advertisers end up paying to reach users who would’ve bought the service anyway – without an ad? I started thinking about this after reading this study on the “ghost attribution” trap. It looked at how user targeting works on Retail Media Networks (RMNs) like Amazon Ads, Walmart Connect, and Instacart Ads – and found that the models were biased. Automated RMN bidding models naturally target users who are already highly likely to buy a product (e.g., people who’ve bought the brand 5 times this month). Then the AI dashboard shows a massive Return on Ad Spend (ROAS) and advertisers keep spending. But randomized controlled trials showed that up to 70% of these sales were non-incremental. The consumer would’ve bought the item anyway. The AI was essentially taking credit for organic behavior, masking a massive waste of ad budget. Scary.

Grant capital is not free money. I have very little sympathy for teams that treat it like it is. For context, I'm talking about the @limitlessdefi grant of 75,000 ARB tokens to incentivize liquidity on @arbitrum. I read from Arbitrum’s Watchdog Committee that the project swapped the entire grant to USDC and bridged it to Base instead. The project later appears to have stopped operating. The team was then given until September 10 to explain what happened or return the funds. They did neither. Now Arbitrum is voting to permanently ban the founders from future DAO programs. If an ecosystem gives you capital for a specific mandate and you move the full amount somewhere else, it becomes a governance and accountability failure. So this is definitely the correct response.

Just came across a very interesting simulation study: What happens if you make AI agents simulate the behavior of typical shoppers – and then test marketing campaigns on them before launching anything in the real world? Researchers built a tiny virtual town and populated it with AI agents. Just 11 agents and 10 locations. Each agent had: • a budget • an energy level • food and shopping needs • memory • social connections • its own persona Then DeepSeek-V3 generated their plans, conversations, and purchase decisions. So the agents went to work, ate, talked to each other, remembered things, invited friends somewhere – and independently decided where to spend money. The actual experiment was very simple: One fried chicken shop offered a 20% discount in the middle of the week. The researchers then watched what happened. • The shop’s revenue increased by 51% from Day 2 to Day 3 – even after (!) the discount. • Its market share grew from 30% on Day 1 to 41% on Day 3. • But the TOTAL market didn’t really grow. So the discount didn’t create much new demand – it mostly pulled customers away from a competitor. • The local diner lost market share and saw revenue fall by 7%. • The coffee shop stayed relatively stable. Even more interestingly, the strongest effect didn’t happen immediately. The authors interpret that as information spreading through the agents’ social network: they talked, made plans together, invited friends – and interest in the place increased through their interactions. The simulation also produced something that looked like loyalty: some agents returned to the fried chicken shop during the promotion, while the diner kept broader appeal even without discounts. So in theory, systems like this could let marketers test: • discounts • promotions • price changes • new product launches • competitive scenarios on synthetic consumers before going into the real market. But there’s a very important BUT. This is still a proof of concept, not evidence that AI can reliably predict real human behavior. The simulation only had 11 agents. Their decisions depend heavily on prompt design, the chosen LLM, the rules of the environment, and even how the virtual town is structured. Imo, this is definitely something to watch. Because if these simulations eventually get calibrated properly on real-world data, part of CustDev, pricing experiments, and campaign testing may happen on AI agents first – and only then on humans.

BTW, @Revolut confirmed it disclosed customer information after receiving fraudulent requests from an email address using a legitimate government agency domain. The exposed information reportedly included identity documents and may also have included verification selfies and transaction histories. Revolut says customer funds were unaffected, but that doesn’t make the exposure harmless. Those details could help someone impersonate a customer or make a scam message much more convincing. I think we should be asking for a proper explanation of how those requests passed their checks, what they’ve changed, and what support affected customers will receive instead of blanket statements.

A few things stood out to me in this piece on compute markets, where AI companies buy access to someone else’s GPUs to train models or run their services ↓ → GPU deals still happen in group chats A lot of that computing power is still bought through brokers, private deals, and group chats. Comparing prices across providers is a mess. TBH, that’s sorta crazy considering how far this tech has come. → Compute could trade like electricity The article argues GPU rentals could become a commodity market, with pricing benchmarks and contracts to manage future costs. Electricity markets started going through a similar shift in the 1990s as they opened up to competition. → The same chip doesn’t mean the same deal Location, reliability, and rental terms all affect the price. Cheap GPU access that can get interrupted is way different from guaranteed availability. An AI chatbot needs to stay available for users, while a company using AI to label images overnight might accept pauses to pay less. → NVIDIA is basically the central bank for compute Its release schedule influences how quickly older GPUs lose value, and it also benefits from keeping existing GPUs busy, since demand for that computing power gives data centre operators a reason to buy more chips. → The opportunity goes beyond renting GPUs Companies are also working on pricing tools, trading platforms, lending, and insurance. The author’s bet is that companies combining reliable GPU access with those financial services could have an edge. Silicon Data says it tracks 95% of neocloud GPU providers and 100% of major hyperscalers. That rental data is already being used to build daily pricing benchmarks for compute. So I think that gives us a glimpse of how GPU rentals could grow into a wider financial market.

This is just another way to give traditional investors exposure to crypto without leaving their brokerage account. @BlackRock's IBIT already proved the model. People buy shares to get Bitcoin exposure. The fund holds $61.6B in net assets. @Fidelity's FETH does the same for ETH. Works through regular brokerage accounts and eligible retirement accounts. No wallet setup required. The appeal is obvious: add crypto to the account you already use for stocks, skip the wallet management, seed phrases, and staking setup. Curious to see how much TradFi money actually flows into TRX from this.

Bruh, I’ve never heard of hackers giving money back lol. Someone found a bug in @Liquid_BTC that let them create unbacked L-BTC, then withdraw almost 4,000 real BTC from the reserve. They returned 3,400 BTC after @Blockstream patched it, but still kept around 598 BTC, worth roughly $47M. Maybe there was a deal behind the scenes, but taking $320M first and keeping $47M is a pretty insane way to negotiate.