The Cannonball / Lead Bullet Portfolio
Allocate growth energy between a few fundamental rebuilds and many small experiments — by product stage.
- Difficulty
- Moderate
- Time to result
- ~months to results
- Steps
- 5
- Confidence
- 90%
Growth work splits into cannonballs (fundamental, months-long changes to how the product works) and lead bullets (small optimizations and copy tweaks). Adriel Frederick argues teams drift toward lead bullets because they're easier to design, build, and defend to a boss — producing incremental thinking that never adds to anything big. The fix is an explicit energy budget between the two, with the ratio set by product stage.
Origin
Adriel Frederick's synthesis from Facebook's growth team. The 'no silver bullets, just lead bullets' framing originates with Ben Horowitz (Andreessen Horowitz); Frederick extends it by adding cannonballs and a stage-dependent portfolio allocation.
Core principles
- 01There are no silver bullets — many lead bullets, and a few massive cannonballs.
- 02Cannonballs are defined by investment size, not just impact: the fundamental changes required to make a thing work at all.
- 03Small experiments carry a hidden cost — the bigger work you didn't do with that time.
- 04A hard allocation constraint forces you to pick only the genuinely good small experiments.
- 05The cost of experimentation is time; early on, that cost is prohibitive.
How to run it
- 1
Identify the core actions of your product
Determine the handful of actions that actually drive growth (at Facebook: find the product, get into the product, find your friends, build the habit of coming back). Everything else is noise.
Pro tip Frame each as a friction-removal target rather than a feature.
- 2
Separate cannonballs from lead bullets
A cannonball is a fundamental, expensive change required to make a core action work — sign-up with phone numbers, reliable global SMS delivery, good friend recommendations. A lead bullet is button copy, layout, small flow tweaks. Classify by size of investment.
Pro tip If it takes a few months and rebuilds how the product works, it's a cannonball — say so out loud so nobody mistakes it for a hack.
Watch out Lead bullets are table stakes, not strategy. Doing them well is just good product communication.
- 3
Set the allocation by product stage
Very early: close to 100% cannonballs — you already know the big things from talking to people, so go build them and don't run experiments. Mature: the balance shifts toward refinements, because there may only be one cannonball left and experimentation is cheap at scale.
Pro tip A mid-stage rule of thumb Frederick offers (explicitly as illustrative numbers, not doctrine): roughly 80% of energy on cannonballs, 20% on lead bullets.
Watch out Do not port a mature company's experiment-heavy culture into an early-stage product — the time cost of waiting for data will kill you.
- 4
Enforce the constraint to filter the small work
Cap the lead-bullet budget. The constraint itself forces the team to select only the small experiments that are actually worth running, instead of shipping ten easy ones because they're easy.
- 5
Grind on the cannonballs, and accept the ugly middle
Cannonballs don't look glamorous and don't produce weekly wins. Protect the team doing them, let them work, and judge on the fundamental change, not on interim metrics.
Pro tip Frederick's example: Tom Allison and one engineer disappeared into a corner for months to rebuild a recommendation system properly — then it crushed it.
In the wild
Facebook's 'people you may know' recommendations spiraled new users down a single friend group — one or two friends produced mutual-friend suggestions from that same cluster, never reaching the user's other social circles. Frederick spotted this by watching someone use the product. The fix was not a hack: it was a fundamental change that seeded a new user across multiple friend groups at once, which then let recommendations compound outward.
→ Described as an absolute game changer, especially for new users, and reusable as a seeding tool for the whole recommendation system.
Over three months a team could run ten small experiments or push on two big ones. Frederick estimates roughly 60% of experiments succeed and 40% should be turned off — but even the successful small ones carry the opportunity cost of the big work not done. The pull toward small experiments is laziness: easier to invent, design, build, and report to a boss ('I moved the number 0.02%').
→ Incremental thinking that produces a stream of small wins that never add up to anything meaningful.
Common mistakes
Growth-hacking your way to a product
Assuming small clever changes alone can produce a durable business is disrespectful to users — they catch on. Hacks buy initial traction; without fundamental value people disappear within months.
Experimenting on everything at the early stage
Early on you already know the big things from talking to people. Splitting traffic 50/50 and waiting for data costs time you don't have and degrades the product for half your users.
Letting lead-bullet laziness set the roadmap
Small experiments are easier to design, build, and defend, so they crowd out the cannonballs by default unless you impose an explicit allocation.
Is it for you?
Best for
Growth leads and product leaders deciding how to spend a quarter's team capacity, especially those under pressure to show weekly metric movement.
Not ideal for
Teams with no ability to run experiments at all, or organizations where a single fundamental change is already mandated and the allocation question is moot.
From the transcript
“there's no silver bullets just many lead bullets”
“maybe it's 80 of your energies on those big cannonballs 20 on the light bullets”
“very early on when you're building a product you kind of know what the big things are you've talked to enough people you have enough…”
“there's a laziness that can creep in where you're just finding a lot of little things because they're easier to come up with”
“one make it easy to find the product number two make it easy to get into the product three stupid easy to find your friends”
From the episode
Humanizing product development
Adriel Frederick (Reddit, Lyft, Facebook)