29 resources from Andreessen Horowitz (a16z) we point founders to, and the questions each answers.
✍️ Essay
✓ Link checkedFreeAdvanced
Why we picked it
A16z's framing of how modern enterprise software is sold: bottom-up product adoption layered with top-down sales, using companies like Slack, Stripe, and Atlassian as evidence. Essential context for deciding when and how to add real enterprise selling on top of a product people already use.
Why we picked it
A seasoned enterprise sales leader gives a crash course on selling into complex organizations like the Fortune 500, including how deals actually get qualified, structured, and closed. The right resource once you're ready to move beyond scrappy founder sales into a repeatable field motion.
Why we picked it
This is a16z's clearest map of where AI genuinely opens new application-layer companies versus where it just produces thin wrappers around a commodity model. It names the actual openings (counterpositioning on business model, workflow orchestration, embedded domain expertise) and spells out why a slick interface over ChatGPT is not one of them. Read it as a lens for judging your own idea, not a shopping list.
The real openings come from owning a workflow and embedding domain expertise, not from a nicer UI on top of a general model.
New entrants can win by counterpositioning on business model (for example per-conversation pricing instead of per-seat), which incumbents struggle to copy.
A frontend that mostly re-skins commodity functionality stays vulnerable, so ask what your product still delivers if the model layer became free tomorrow.
Why we picked it
This is the essay to read when you want to know whether a trend can carry a real company or is just a feature riding on someone else's model. Casado and Bornstein show why AI economics often look more like a services business than clean SaaS: lower margins, real infrastructure and human-in-the-loop costs, and moats that are shallower than the hype suggests. It is from 2020, but the core question (where does durable value actually accrue) is exactly the one to ask about any new AI wave.
AI companies often run at 50 to 60 percent gross margins, not the 60 to 80 percent of classic SaaS, because compute and human review are real recurring costs.
Model access is commoditizing, so defensibility comes from owned data, a narrow workflow, and real switching cost, not from the model itself.
Before betting on a trend, ask whether it supports a standalone business or is a feature a bigger platform will absorb.
Why we picked it
This piece names the exact trap you are worried about: it uses Zynga, which rode Facebook's platform to a huge business and then got squeezed when Facebook changed the rules, as the cautionary case. Kupor's honest framing is that platforms are a great way to bootstrap early distribution, but you have to use that head start to build a direct relationship with your own customers before the platform's interests diverge from yours. Read it as a starting point for deciding which parts of your business must not stay platform dependent.
A platform is fine as a launchpad for cheap early distribution, but treating it as your permanent, only channel is what gets you wiped out.
Zynga's core mistake was never owning the direct relationship with its players, so when Facebook shifted policy it had no independent way to reach them.
The hedge is to spend your platform honeymoon building things you control: your own audience, data, and a second path to your customer.
Why we picked it
This is a clean, concrete example of reading a policy change as demand rather than a headache. Instead of treating new rules as compliance drudgery, it walks through five specific regulatory shifts in Brazil (instant payments, open finance, unbundled licensing, and more) and shows how each one opened a lane for a new company. It gives you a repeatable way to ask, of any rule change in your own sector, who now needs something they could not get before.
A single regulatory reform can be a 'why now' for a whole category: Brazil's PIX instant-payments mandate reached 139 million users in two years and seeded a wave of startups.
Unbundling one big license into tiered, startup-sized licenses is often where the opening is, because it lets new entrants do one piece well instead of becoming a full bank.
When you see a policy change, map it to a specific group of newly-served or newly-unlocked customers, not to a compliance to-do list.
Why we picked it
This essay explains the actual mechanism: when regulation forces incumbents to open up (for example, rules requiring banks to expose customer data), it unbundles a closed industry into modular pieces that new companies can build on. That is the pattern behind Plaid, and it is why a rule that looks like a burden for incumbents is often a market opening for someone smaller and faster. Read it to train the instinct of asking, when a rule changes, what just became a building block that used to be locked away.
Regulation that forces incumbents to open data or unbundle a license turns a closed industry into modular pieces new startups can assemble.
The classic proof: a data-access provision in US regulation is what made a company like Plaid possible, so a compliance mandate for banks became a whole business for someone else.
The founder's move is to look for what a rule change unlocks or standardizes, because that unlock is usually the new market.
Why we picked it
This is the clearest short case that how you reach customers is a real edge, not an afterthought behind the code. Horowitz frames the route to market as a function of your product and your buyer, and shows (Dropbox vs Box) how two teams won different slices of the same market purely on go-to-market, which is exactly the lever two non-technical founders have. It is a starting point for thinking about your beachhead as a distribution question, not only a technology one.
Your channel to customers is derived from your product and your target buyer, so a niche you can reach cheaply is often more defensible than a slightly better feature.
Two companies can carve out separate, durable positions in one market on go-to-market alone (Dropbox's viral self-serve vs Box's direct enterprise sales).
Picking the wrong route to market sinks otherwise good products, which is why founders should treat distribution as a first-order decision, not a later one.
Why we picked it
When a big platform ships the same feature you are known for, this article explains why that feature buried inside a broad suite often loses to a dedicated tool built only for that job. Big platforms get pulled in many directions and end up serving no single user perfectly, which is exactly the opening a focused product exploits with better UX and a business model the incumbent will not fully commit to. It walks through real unbundlings (StubHub out of eBay, Twitch and TikTok out of YouTube) so you can see the pattern rather than take it on faith.
Why we picked it
This is the essay that reframes the whole question. Instead of asking whether your DAU/MAU is high enough, it tells you to look at the histogram of how many days per month people actually use the product, and it says plainly that not every product should have a daily-use curve. For a low-frequency product, that is the honest starting point: match the metric to the real shape of usage instead of forcing a daily lens that will always look like failure.
DAU/MAU is a single blunt number that hides the variance in how often your users return; the power user curve shows the full distribution of active days per month.
A flat or right-skewed curve is not automatically bad. Some product categories (professional, investment, and other infrequent-use tools) are healthy at low daily engagement.
Pick a frequency lens that fits how people naturally get value, then track whether a real core segment keeps coming back at that cadence.
Why we picked it
A16z boils the whole decision down to one honest rule of thumb: usage-based pricing tends to fit products whose main user is other software, while subscriptions tend to fit products with human users. It is short, opinionated, and gives you a lens to reason from instead of a list of pros and cons. Read it as a starting point for framing your own call, not a verdict.
Why we picked it
Chan's piece is the honest version of the ads question: advertising is a real model, but it degrades the product once you are stacking thousands of impressions on people, and it only pays off at genuine scale. She walks through how Chinese platforms layered in subscriptions, tips, and purchases instead of leaning on ads alone. Read it as a starting point for deciding whether ads should be your first model or a later, secondary one.
Why we picked it
This is the reference definition founders actually cite when they argue about take rate, and it gives you the honest benchmark band in one line: take rate usually runs from a low single-digit percent up to the mid-30s, depending on how fragmented your market is and how much work the platform does. Instead of chasing a magic number, it points you at the real levers (fragmentation, substitutes, operational value-add), which is the right way to reason about where your rate should sit. Read it as a starting point, not a verdict, then pressure test your own number against it.
Take rate (also called rake) is simply the percentage of gross merchandise value the marketplace keeps, so define your GMV clearly before you argue about the percent.
Benchmarks span a low single-digit percent to the mid-30s, so there is no single right number, your fair rate depends on how much matching and operational work you do.
Managed marketplaces that do more of the work (logistics, trust, guarantees) can justify a higher take rate because they add more value to both sides.
Why we picked it
This tackles the exact fear you named: a buyer who wants your product but cannot predict the bill, and sometimes builds it in-house just to control cost. Instead of one trick, it lays out a menu across pricing, sales, product, and support (committed spend, in-product usage dashboards, forecasts, true-ups) so predictability becomes a design choice, not an afterthought. Pair it with hard spend caps and threshold alerts, which are the blunt-but-effective safety net it complements.
Why we picked it
This is the canonical a16z reference that defines the marketplace vocabulary you actually need before you build a model: GMV vs revenue, take rate, match rate, and market depth (which is how liquidity really shows up). It keeps you honest about the fact that a marketplace has two sides to feed, so your unit economics have to reflect both supply and demand, not just orders. Treat it as the shared language, then plug your own numbers into it.
Why we picked it
The margin leak most bootstrapped founders never see coming is their own infrastructure bill, and this is the essay that put a number on it: across 50 public software companies, committed cloud spend averaged about half of cost of revenue. It reframes cloud from a convenience into a line item that quietly decides whether you are running a healthy business or a break-even one. Read it as a starting point for asking where your COGS actually goes, not as a case to flee the cloud on day one.
Cloud can silently become 50 percent or more of your cost of revenue, so a great top-line business can still run on thin margins.
The cheapest, most flexible choice early on (on-demand cloud) becomes the most expensive one at scale, and the switch happens gradually enough that nobody notices.
Treat infrastructure spend as a first-class metric alongside revenue, because it compounds the same way your growth does.
Why we picked it
Dalgaard, who founded SuccessFactors and sold it to SAP for 3.7 billion dollars, makes the case you need to actually pull the trigger: he has never fired anyone too early. His sharpest point is exactly your five-person problem, junior people spot the mis-hire before you do, and every week you tolerate it they lose faith in your judgment and the best ones start eyeing the door. Read it when you are still hoping it works itself out; it will not.
Why we picked it
Horowitz names the exact thing that happens to your head in a crisis: the 3am spiral, food losing its taste, self-doubt curdling into self-hatred. He wrote this after nearly losing Loudcloud, and the second half is a founder's field manual for the spiral itself: do not carry it all alone, focus on the road and not the wall, play long enough for luck to find you. It is the most honest first-person account of the internal weather of a real crisis, and it ends where you need it to: the struggle is where greatness comes from.
Why we picked it
Horowitz did three layoffs at Loudcloud/Opsware and wrote the definitive playbook: managers lay off their own people (never HR, never a mass email), you say out loud that the company failed to hit its plan and not the person, and you walk in already holding the benefits and settlement details so there is nothing to wing. It is the exact framing you need for the conversation you are dreading.
Why we picked it
A useful correction for founders chasing a market-wide 40 percent before a single cohort loves the product. The authors argue you first need a small group of power users who genuinely pull the product out of your hands, then expand outward. It sharpens the difference between broad lukewarm interest and the intense fit that actually compounds.
Why we picked it
Design partners are how B2B founders validate with depth, and this gives you a practical way to pick the right ones. The urgency, capability, and representativeness test helps you avoid partners who love you but never buy, or who move too slowly to teach you anything. It also makes the point that good design partners should convert into your first paying customers.
Why we picked it
If you're seriously weighing open-sourcing your MVP, this lays out what that path actually demands. It walks through project-community fit, product-market fit, and value-market fit, and why open source is a long commitment to community and governance, not a free megaphone. Use it to decide whether open source fits your product and your appetite for maintaining it.
Why we picked it
A grounded look at what AI and no-code builders genuinely let non-technical people ship, and where they still hit a wall. It is honest that complex logic and data work often stall out and hand back to engineers. Read it to keep your expectations realistic about how far one prompt actually takes you.
Why we picked it
This is the second act in action: once you own a narrow vertical, adding payments and fintech can more than double your addressable market from the same customers. It shows how founders escape the growth ceiling of a small market without abandoning their niche. Useful once you have won the room and need a next chapter.
Why we picked it
Haber argues that tools are now widely available, so the scarce ingredient is knowing what is actually useful inside a real industry. That is a direct validation of the operator's edge: you carry the context that technical founders have to go hire. Read it to see how your lived knowledge of where the bodies are buried becomes the product's differentiation.
Why we picked it
This piece argues that whole categories of local businesses, think laundromats, clinics, and repair shops, were long dismissed as markets too small, and that assumption was wrong. It is a useful counter to anyone who tells you your town's small businesses cannot support a company. Read it to see how narrow, unglamorous local markets became large businesses.
Why we picked it
This gives you the vocabulary for the moment a trend makes a job newly urgent. An inflection point is when customers who tolerated their old solution suddenly become open to switching, and this piece shows how to spot and time that opening. It sharpens the difference between a trend that is merely interesting and one that just made a real customer ready to move.
Why we picked it
When you want to go deeper than headlines, this is a curated map of the resources that actually explain modern AI, sorted from gentle introductions to market analysis. Instead of trusting one loud take, you can read the primary sources and form your own view. Treat it as a menu: pick the introductions and the market pieces, skip the heavy research papers unless you are curious.
Why we picked it
Each year a16z partners publish the shifts they expect, and reading it shows you what a finished trend thesis looks like when smart people commit to one in writing. It is less a source of answers than a model to argue with: where do you agree, where is the signal thin. Use it to pressure test your own list against people who do this full time.
