By Arun Agrahri · Published
Every framework page on Gavel ends with one place where another named expert gives the opposite advice. This page collects all 84 and sorts them by the founder problem they belong to. Each one gives both views in our words, a note on when each view fits, and a link to the page it came from.
When your team is growing · 11 splits · Scaling research
locates the difference in the person: personal humility joined to indomitable will, with ambition pointed at a cause bigger than yourself. He argues it is a place you can grow into, and uses the last decade of Steve Jobs as the evidence.
locates it in the machine. Build an idea meritocracy where thoughtful disagreement is designed into the room and opinions are weighted by track record, not seniority, and the organisation stops depending on one person having the right character.
They are different layers, not opposites. Dalio's machinery still needs someone willing to be told they are wrong in public, which is Collins's humility under another name. The failure mode is waiting to grow into level 5 while making every call alone.
From Level 5 Leadership
says breakthrough is cumulative, not decisive, and asking which push caused it is, in his words, a nonsense question. You keep turning the same wheel in a consistent direction until the momentum becomes visible from outside.
points out that a network product has a threshold before any of that applies. Below the atomic network, some number of people connected at the same time doing the same thing, the wheel is not slow, it is disconnected, and pushing harder changes nothing.
Collins was studying companies that already had a business and asking what compounded it. Chen is answering the question before that one. Find out whether your wheel is heavy or simply not attached to anything before you commit to another year of turning it.
From The Flywheel Effect
locates the problem in the system. Believability-weighted decision making inside an idea meritocracy designs thoughtful disagreement into the room and sets the weight of each opinion by track record, not seniority or confidence.
locates it in the individual. Resulting, motivated reasoning and hindsight bias are the primary source of decision errors, so the work is personal process: pre-set criteria, kill conditions, and separating decision quality from outcome.
Different layers, not rival theories. Duke's moves work on a decision you own; Dalio's work on decisions a group owns. The failure mode is building Dalio's machinery for a call one person could have made in ten minutes.
says you will not get rich renting out your time, and to reach for the permissionless forms, code and media, first. Labor ranks last.
holds that revenue plateaus around seven to ten million the moment the entrepreneur becomes the bottleneck, and that the escape is delegation and systems, which is labor leverage.
Both are right at different scales. Naval is describing how to get off zero without a gatekeeper. Hormozi is describing what breaks once you are past it.
treats silence as the expensive failure. Ruinous empathy, withholding a hard truth to spare someone's feelings, is in her account the most common and most damaging communication mistake, and the fix is to challenge directly while showing you care.
starts from the reader. Put yourself in the audience's position first, and let warmth and generosity earn the right to be heard. Directness that arrives before trust reads as an attack, whatever the speaker intended.
The deciding condition is who the words are for. Scott is describing feedback inside a team, where the cost of not saying it lands on the person you withheld it from. Handley is describing writing aimed at people who owe you nothing and can leave. Internal communication rewards candor; audience-facing work rewards empathy first.
From Radical Candor
treats team health as the precondition for everything above it. Build vulnerability-based trust first and debate becomes the pursuit of the best answer; skip it and the same debate becomes politics, where people manoeuvre to win rather than to be right.
ran Intel's RISC debate to a correct conclusion and accepted the wreckage. Half the people who had worked on the processor left, and by his own account those relationships never healed. He calls the debate itself one of the most important ways of handling the question.
The condition is what is at stake in the argument. Lencioni is describing the standing operating state of a team, where artificial harmony compounds quietly for years. Grove is describing a one-off bet-the-company call under a deadline. Note Grove is not disagreeing that the cost is real: he names it plainly and says these are not fun and easy processes even among people who know each other well.
defines culture from day one through rules specific enough to enforce, imposed top-down. Respect for a founder's time becomes a $10-a-minute fine for lateness. His claim is that specificity is the whole mechanism, because a vague value cannot be obeyed or broken and therefore changes nothing.
Shaan Puri and Alex Hormozi
argue against pre-defining values at all. Watch what the company already does under pressure, then name and codify the behaviours that emerged on their own. On this reading a dictated value competes with the real norm and usually loses.
Both camps agree vague values are useless, so the split is about timing rather than specificity. A company with no history has nothing to observe, which is Horowitz's case for writing rules on day one. A company that already has working norms risks overwriting them, which is the case for naming what is there. The failure mode Horowitz warns about applies to both: a rule enforced on some people and not others teaches a culture about who is exempt.
rules out talk about the past once a behavior is chosen. Colleagues give one or two ideas for the future, the person says thank you, and nobody debates what already happened.
argues for criticism given in person and as close to the event as you can, about the specific work. In her account silence is the costly failure, and a review that waits has already waited too long.
They answer different moments. Scott's criticism tells someone about a problem they cannot see yet. Goldsmith's feedforward starts after the problem is known, and his clients get that diagnosis from confidential feedback first. Use candor to name the behavior, then feedforward to change it.
From Feedforward
argues that no single leader can move every piece in a fast, complex fight. Build the shared picture, push authority down until it feels uncomfortable, and act as a gardener who sets up the conditions. He says a leader who keeps a hand on every chess piece will lose.
Brian Chesky
argues that leaders should be deeply involved in the details. Being in the details is how he understands how his teams perform and drives the company forward, and he does not count it as micromanaging.
The condition is who holds the context. McChrystal's model assumes you have already shared the full picture, so the people below you can decide about as well as you could. Chesky's position fits a fast shift, such as AI, where the founder holds context nobody else has yet. Even McChrystal's gardener stays completely engaged: what he gives up is making each decision, not knowing the details.
argues a company is a professional sports team, not a family. If you would not fight to keep someone and you think you can recruit better, you pay a generous severance and make the upgrade.
Brian Halligan
counters that outside candidates look shiny in interviews and you have not seen their warts, so managers overrate them and underrate homegrown people. When the call is close, he gives the internal person the shot.
The split is how sure you are that the replacement is better. Hastings' test assumes you can tell, which is why he digs through LinkedIn for references more loyal to him than to the candidate. If you cannot check an outsider that well, Halligan's rule stops you from trading a known performer for a good interview.
From The Keeper Test
puts every company he buys on one ERP, one HR system, one CRM and one culture, starting the day the deal is signed. He calls a roll-up that leaves its companies on separate systems a mess that whoever buys it next has to fix.
Michael Girdley
runs his holding company radically decentralized. On Greg Isenberg's How to start a Holding Company episode he pointed to Berkshire Hathaway, Roper and Andrew Wilkinson, who keep purchasing, HR and finance inside each company instead of centralizing them.
Girdley names the deciding variable himself: how alike the assets are. Fourteen fast-food franchises should share bookkeeping and ordering; a spread of unrelated businesses should not. Jacobs only buys inside one industry, so his companies are near-identical by design. If your targets do the same work, integrate from day one. If they do different work, keep the center thin.
From Roll-Up Playbook
When you have too many ideas · 10 splits · Focus research
locates the problem in the individual: resulting, motivated reasoning and hindsight bias drive decision errors, so you fix decisions by fixing personal process.
Donella Meadows and Peter Senge
argue behavior is produced by structure, feedback loops, incentives and delays, so blaming individual judgment misses the cause.
Both levels operate at once. Kill criteria are a personal-process intervention that works best when the structure backs it: a review cadence on the calendar, and incentives that do not punish the rep who walked away from a dead deal.
From Kill Criteria
Annie Duke and Douglas Hubbard
put a number on confidence. Hubbard pushes it further: anything that matters is observable and therefore measurable, and calibrated probability estimation makes even intangibles quantifiable.
Nassim Taleb
draws the opposite line. We systematically overestimate our ability to predict in complex domains, and a precise-looking model there is more dangerous than admitted ignorance because it manufactures false confidence.
The reconciliation both camps accept: calibrate where outcomes are bounded and repeat often enough to be scored; where they are not, stop forecasting and build for robustness.
From Thinking in Bets
builds defensively: understand failure through inversion, eliminate ignorance through the circle of competence, survive surprise through a margin of safety.
takes the other side for company builders. A bad plan is better than no plan, and definite optimism with a concrete vision is what produces transformative outcomes, because iteration without a bold plan will not take you from zero to one.
The line between them is whether the domain is predictable. Invert where failure modes are known and the downside is fatal; plan boldly where you have genuine insight nobody else has. Taleb rejects both premises and preserves optionality instead.
From Inversion
puts a floor under the Highlight at roughly an hour, on the logic that anything smaller cannot anchor a day, and caps the day at one.
argues the opposite direction: shrink the behavior until it takes under 30 seconds and needs almost no motivation, then let it grow.
James Clear sits between them at six ranked tasks a day. All three agree environment beats willpower. If you have bounced off Make Time twice, your first version was probably too big.
From Make Time
says the great companies found the single intersection of what they can be best in the world at, what drives their economic engine and what they are deeply passionate about, and then said no to everything outside it for decades.
Jeff Bezos
wandered outside his circle of competence on purpose, guided by curiosity, not by a hedgehog test, which is how Amazon arrived at AWS and the Echo. Neither would have survived a best-in-the-world screen on the day it was proposed.
The reconciliation is cost. Stay inside the hedgehog for bet-the-company calls, and wander on experiments cheap enough that being wrong is survivable. Collins studied companies that made a leap and held it; the wandering that worked was funded by an engine that was already running.
From The Hedgehog Concept
rests the method on judgment: someone decides an appetite up front, and he defends intuition over data-heavy analysis on the grounds that people are fundamentally feeling creatures.
would weight that same call by who has a track record and can explain the cause and effect behind it.
The split is about evidence density. Dalio's approach needs a history of comparable decisions. Fried is working where that history does not exist and waiting for it costs more than being occasionally wrong.
says failure is the cheap risk, because it is quick and it leaves your time intact. The thing that is just okay is never bad enough to make you quit, so it takes your years and your belief in yourself along with them.
says breakthrough is cumulative, not decisive, and asking which push caused it is in his words a nonsense question. The years that look flat from outside are the years the wheel is loading, and quitting there is the error.
The tell is whether anything is moving. A flywheel produces turns you can count while the outcome stays invisible, so flat years are a measurement problem. Mediocrity produces okay and holds there. Count the turns before you decide which one you are in.
gives every minute of the workday a job in advance and redraws the plan when it breaks. Without that plan, he argues, email, meetings and admin overhead take the day and the real work slides into the evening.
plans one Highlight, a single 60 to 90 minute block written down before the day starts, and accepts that the rest of the day is reactive. Everything else exists to protect that one block.
The deciding factor is how much of your calendar you control. If most of the day is yours and the overhead is burying you, Newport's full plan is the stronger tool, and he concedes it is tiring. If other people own most of your hours, a full plan gets redrawn all day, and one defended Highlight is the version that survives.
From Time Blocking
reviews everything once a week: every inbox emptied, every project walked, the calendar read back and forward, Waiting For and Someday/Maybe checked. The goal is a system current enough that your head can let it go, and it takes an hour or two, sometimes three.
uses the weekly slot to plan execution. He reads his quarterly plan, builds a weekly plan that protects calendar time for the big projects, and time blocks each day from it.
They use the weekly slot for different jobs. Allen's review makes your lists true; Newport's plan decides what the week is for. If things keep slipping or resurfacing in your head, run Allen's review first, because a plan built on stale lists schedules the wrong work. If your lists are clean and the big projects still never get time, Newport's weekly plan is the missing piece.
From Weekly Review
built the premortem to fix the plan at kickoff: the failure list cuts the team's overconfidence, and a closing two-minute round on what each person will do leaves a stronger plan and a more resilient team.
counters that running a premortem rarely changes the plan much, and that its real output is kill criteria: the signals that mean stop, each tied to an action before anyone is invested.
It turns on whether the plan can still move. At a kickoff, before roles and budget are locked, Klein's version pays because flaws get fixed. Once you are committed, sunk cost is the risk, so turn the same list into Duke's stopping rules.
From The Premortem
When you're not sure anyone wants it · 10 splits
calls everyone who might buy, confirms the problem without pitching, then asks for money on the same call. He sold Trends and Hampton before either existed, and he reads a buyer who balks at a real offer as a problem that is not big enough.
Y Combinator partners
argued on YC's Startup Ideas You Can Now Build With AI episode that sell-before-you-build is outdated for AI products. A model can now produce output no customer would think to ask for. Their advice is to work with the newest technology and follow your curiosity toward what it can now do.
The split is whether the buyer can picture the product. Parr's businesses answered known problems, a peer group for CEOs and a research newsletter, where a 20-minute call surfaces the need. For a capability that did not exist last year, the person on the phone can only describe the old workaround. If the problem is known, sell first. If the capability is new, build first.
From Presell Calls
runs the tactical, qualitative lineage of Jobs to Be Done: interview people who actually switched, surface the forces around the switch, and study the struggling moment, because the causal story of one real switch teaches more than a tidy metric.
Tony Ulwick
runs the outcome-driven lineage, treating the job as a quantifiable statement with a direction of improvement, a metric, an object and a context, stable and testable at scale. The rigour is the point, because a fuzzy statement cannot be prioritised.
Both call themselves JTBD, which is why founders reading two sources come away confused. Moesta tells you why the people you talked to switched; he does not tell you how many of them exist. Pair the switch interview with sizing work before you bet the roadmap on a job you only heard described in a dozen rooms.
From The Switch Interview
trusts a single survey question put to active users, how would you feel if you could no longer use this, and reads 40 percent answering very disappointed as the fit signal. One number you can run this week and compare against the last time you ran it.
distrusts what people say in a survey and reconstructs what they actually did, interviewing recent switchers to find the push, pull, anxiety and habit that moved them. Stated preference is not behaviour.
Stated preference against revealed behaviour is a real methodological split, not a style preference. Ellis tells you whether you have fit; Moesta tells you why, which is the part you need before you can do anything about a bad number.
says write the success metric down before you build, ask how you would cheat it, and add the countervailing metric that catches the cheat. Any metric you can move you can also game, and an experiment with no OEC will faithfully optimise the wrong thing.
Archie Abrams (Shopify)
runs core product without traditional KPIs, on the view that intuition and taste guide the decisions that matter and a metric target quietly narrows what the team is willing to try.
The split maps onto what you are building. Growth and optimisation work needs Kohavi, because that is where gaming is cheap and the effects are small enough to fool you. A core product bet has no comparable baseline to measure against, which is exactly where a scorecard turns into theatre.
says no result counts as evidence until the split passes the sample ratio check. Lost traffic is biased traffic, so a failed split makes results look super good or super bad. His team at Microsoft blanked the scorecard rather than let anyone read one.
John McEvoy
ran about ten paywall and onboarding experiments on his own app over two or three months. Conversion went from 0.5% to 8%, and revenue from around $8K to over $30K a month. His rule: A/B test paywalls and onboarding flows relentlessly (Starter Story, 9:12).
The deciding variable is whether anything outside the test can confirm the win. McEvoy had that check, because monthly revenue nearly quadrupled and a broken split cannot fake revenue. A team shipping small lifts it can see only in the scorecard has no outside check. At the rates Kohavi reports, about one test in twelve has a broken split, so the SRM check is what stands between that team and a roadmap built on noise.
says size the test before it runs: an MDE of 5% or less at 80% power, which for a store converting at 5% means about 200,000 users. Below that a significant result is likely exaggerated, so build the culture and the platform, and test only for large effects.
tests a prototype with five target customers on the Friday of a design sprint, not with internal stakeholders. In Sprint, five customers reveal the major patterns without statistical overhead, so there is no reason to wait for a large sample.
They answer different questions. Five sessions show whether people understand and want the thing, because confusion repeats as a pattern. They cannot show whether a shipped change moved conversion by 3%, which is the question Kohavi's power formula answers. Use five customers to decide what to build, and the power formula before you claim a change moved a metric.
optimizes for a repeating weekly rhythm with whoever the product can recruit, on the theory that the habit compounds.
argues the opposite trade: pick a fixed set of reference customers, six to eight for B2B, and design with the same people until they will vouch for the product.
One method buys breadth and consistency; the other buys depth and a reference list you can sell with. Neither survives if you only do it once.
grows the tree downward from an outcome and refuses to let a solution into the opportunity space. Several solutions branch under one opportunity so the team can compare them, and every branch has to trace back to something a customer said in a story.
shapes work the other way. A senior person defines the problem and a rough solution together at the right level of abstraction, leadership bets on that shape for a fixed six-week cycle, and the filter on ideas is whether they keep coming back rather than whether an interview surfaced them.
The deciding condition is whether you already are the customer. Basecamp builds for a market Fried has lived in for two decades, so his intuition is compressed evidence and the tree would mostly re-derive what he already knows. A team handed an outcome in a domain it does not live in has no such stock, and shaping from intuition there is guessing with a deadline attached.
sets the bar at a countable set of specific people, six to eight in B2B and twenty-five in consumer, who will personally put their reputation behind the product. Not signups and not a satisfaction score, a named list.
reads fit from a survey across active users, with 40 percent answering very disappointed as the line. A hand-picked group of advocates is, by construction, the most favourable sample you could have drawn.
Idiodi is measuring depth, Ellis breadth, and each is blind to the other's failure. Twenty-five devoted references can still be very nearly the whole market; a healthy survey number can hide that nobody would stake anything on you. Get the references first, then check whether they generalise.
treats the MVP as an instrument for answering a question, which puts rigor at the centre. Minimum is context-dependent and often demands more care than teams assume, because a test run below the market's threshold of viability returns a result about your polish rather than your idea.
tells founders to launch the jankiest version that does the job, on the grounds that time spent perfecting anything before contact with users is time spent on assumptions. Shipping rough and early is the fastest way to find out whether anyone cares.
The condition is the market's tolerance for rough. In consumer software with a forgiving audience the jankiest build is a valid experiment, and Tan's advice wins on speed. In regulated, enterprise or high-trust markets, the viable half is what makes the result readable at all, and shipping under that bar produces a clean negative about the wrong variable. Ries is not arguing for polish; he is arguing that you cannot learn from a test the customer refuses to take seriously.
When people don't get what you do · 9 splits · Positioning research
starts from what the customer would do if you did not exist, then positions inside a market category buyers already understand so they arrive with the right expectations. The category is a tool you pick deliberately, and usually one that already exists.
argues the durable win often comes from counter-positioning: a stance, and sometimes a whole new category, that the incumbent structurally cannot follow without damaging its existing business. Fitting an established frame can hand the advantage back to the leader.
The alternatives exercise tells you what buyers are weighing today. It does not tell you whether winning that comparison is defensible tomorrow. Run Dunford to get chosen, then ask Helmer whether the reason you were chosen survives the incumbent copying it.
puts the market frame of reference at step nine of ten, because the category you claim should fall out of the alternatives and the capabilities only you have. Start from the category and you inherit its expectations before you have checked whether they help you.
hands you a fixed fill-in-the-blank statement with the category near the front, so a whole team can align fast on one crisp sentence. The structure is the feature: it forces the decisions and keeps everyone saying the same thing.
Moore optimises for alignment across a large team; Dunford optimises for getting the answer right. If ten people have to repeat your positioning next week, the template earns its keep. If you are still not sure what you are, filling in the blanks will just make the wrong answer sound official.
argues a perfectly logical position is also a forgettable one, because buyers run on memory and heuristics rather than spreadsheets. The distinctive, slightly weird asset an optimiser would strip out is often what actually gets you chosen.
builds positioning as a rational chain: competitive alternatives, the capabilities only you have, and the value a specific segment cannot get elsewhere. Get the logic right for the best-fit customer and the pick becomes obvious. The work is precision, not flourish.
Weird without a clear frame is just noise, and precise without distinctiveness is a position nobody recalls when the buying moment arrives. Sharpen the logic first, then protect the one strange thing your optimiser keeps trying to delete.
says focus everything on one beachhead and lead it before moving, positioning inside a category pragmatists already recognize, since a new one leaves mainstream buyers without a frame of reference.
Kim and Mauborgne
push the other way, arguing the largest opportunities sit with noncustomers outside the segment you are studying, and that uncontested space is the whole prize.
The practical reconciliation: start narrow and inside an existing category, and revisit noncustomers and category creation once you actually lead your beachhead.
claims the right name buys asymmetric advantage before launch, and that a descriptive name gives that advantage away.
starts one level up, with what customers would use if you did not exist, and would tell you that no name rescues positioning you have not done.
Both are right, in sequence. Placek's close ally is Rory Sutherland: familiar enough to be understood, strange enough to be interesting.
From The Naming Process
chooses letters for the signal their sound sends before anyone reads the meaning, and treats a descriptive name as an advantage given away.
Colin McIntosh
judges a consumer brand name on whether it is spellable, shareable and memorable, connotes the product, is good for SEO and has the .com available.
Both lists reward a name that is easy to say and remember. They split on meaning. A brand that sells through search and word of mouth gains from a name that says what it is; a company entering a crowded category where every rival sounds descriptive gains from a name that sounds like its promise.
treats the small group as the destination. It grows only when its own members bring people who share the worldview, and there is no next segment by default.
treats the small segment as a staging ground: dominate one beachhead to earn the references that let you cross into the next.
April Dunford lands between them, narrowing by which buyers care disproportionately about one differentiated value. Sean Ellis turned the same question into a measurement at forty percent very disappointed.
collects the words first. Surveys, interviews and review mining produce a hundred or more tagged insights, and the count of what people actually said decides the hierarchy, the content and the language of the page.
works from a model of the market instead. Where the prospects sit on the awareness ladder, and how many competing claims they have already heard, tells you what the headline has to do before you learn anything about an individual buyer.
The split is whether you have customers yet. Maguire's method needs a base to mine, so it is the stronger one once traffic and reviews exist. Schwartz's ladder is what you have before that, and it stays useful after: it tells you which of Maguire's top-ranked messages the reader is ready to hear.
treats the mid-market as a naming problem rather than a segment, because asking five people to define it returns five different answers built on revenue, market cap or headcount. She collapses it into two games, small business powered by marketing and enterprise carried by sales, and tells founders to go at tier-one logos early with an ACV headed for $75K to $150K.
Y Combinator
defaults early-stage companies to mid-market precisely because it is the softer target, arguing that enterprise sales cycles run too long for a young company to learn anything from. Faster iteration loops beat a bigger logo when you are still finding out what you are selling.
The deciding factor is whether the problem exists outside enterprise at all. If it does, YC's learning-speed argument wins and mid-market is the cheaper classroom. If the problem is structurally an enterprise problem, there is no smaller version to practise on, and Abel's point stands: you are playing the enterprise game whether you named it or not. Note YC concedes the same edge, that a motivated champion inside an enterprise can move as fast as a mid-market deal.
When you have to close deals · 7 splits · Sales research
treats the pitch as a designed sequence: a short setup on the market and your point of view, then most of the call on differentiated value against the alternatives the buyer is actually weighing. Structure is what stops a sound positioning doc from producing a flat call.
would say the twelve minutes you spend talking are twelve minutes you are not learning. His opener is a label or a mirror that gets the other side explaining their own position, on the view that the person doing the talking is the person giving away information.
They fail in opposite directions. Dunford without Voss is a polished monologue at a buyer whose real objection never surfaced. Voss without Dunford is a warm conversation that never gives the buyer a reason to pick you. Run the setup, then shut up and label what you hear before you spend the second half.
says most teams build the sales process on probability, so many leads in and so many converted, when the question that matters is how many people are ready to buy. He finds where each buyer is on the six-phase timeline and helps them take the next step.
Jeb Blount
treats volume as the job. Prospect every single day, keep 30 times your quota in pipeline opportunities, and read an empty pipeline as a failure to prospect consistently (Fanatical Prospecting).
They answer different questions. Blount decides how many buyers you reach; Moesta decides what you say to each one once you do. A team with a thin pipeline starts with Blount. A team whose demos keep stalling starts with Moesta, because more leads pushed through the wrong phase only stall faster.
From The Buying Timeline
deliberately withholds resolution, alternating what is with what could be so the audience feels the gap before it gets an answer.
Barbara Minto
says lead with the answer, then support it, which is the house style at most consulting firms and in most executive readouts.
Both are correct for different rooms. A board asking for a number wants the pyramid; a team you are trying to move wants the waveform.
built sales at HubSpot as a measurable system: hire against traits the data actually predicts, coachability over aggression, train one repeatable process, and manage the funnel by numbers instead of instinct.
Carles Reina
scaled ElevenLabs past 350 million in ARR and argues the systematised version is exactly what broke outbound. The tools treat every prospect as a transaction and buyers can smell it, which is why reply rates cratered. Personal beats scalable.
Roberge is talking about the inside of a sales org, Reina about the message that leaves it. You can run a rigorous hiring and coaching formula and still insist every first touch is written by a human, and the operators with the best numbers usually do both.
treats every negotiation as holding at least three Black Swans, unknown unknowns that would change the deal if surfaced. The party with the most information usually wins, so calibrated questions and labels exist to extract what the other side does not realise they are giving away.
Daniel Pink
argues information parity has already transformed selling. Once a buyer can look up everything a seller knows, hoarding and extraction stop paying, and honest servant-selling plus problem-finding replace them.
The deciding factor is how much of the relevant information is public. Pink is describing consumer and transactional markets, where the internet closed the gap and an extraction posture now reads as manipulation. Voss is describing complex, multi-party negotiations where the decisive facts are constraints, deadlines and internal politics that were never written down anywhere. Robert Cialdini sits between them: genuine expertise creates an information advantage you earned rather than one you hid.
From Calibrated Questions
opens by labelling the other side's position and letting them talk, on the view that the person doing the talking is giving away information. Tactical empathy is not warmth, it is a way of getting the real objection said out loud.
Matt Dixon
analysed 2.5 million sales calls and found most deals are not lost to a competitor at all: 40 to 60 percent of qualified pipeline is lost to no decision. His answer is to teach the buyer something they did not know, name the risk they cannot see, and take it off the table.
Empathy wins on trust and challenge wins on indecision. If your deals die quietly instead of to a rival, understanding the buyer better is not the missing piece, and a smaller first commitment or a guarantee usually is.
From Tactical Empathy
treats influence as rules people already carry. Give first and the other side feels they owe you, at a negotiating table as much as in a restaurant. The work is picking the first move.
argues that negotiation runs on emotion. Show you understand the other side's feelings and name them out loud. He calls this tactical empathy, and says it beats any logical argument or compliance trigger.
They meet at listening. Cialdini listens first so the other side owes him a hearing. Voss calls listening the cheapest concession you can make. If you are designing the offer, the onboarding or the follow-up, use Cialdini's first move. If you are across the table from someone upset or dug in, a gift can look like a tactic, so start with Voss's labels.
When you set or raise prices · 5 splits
argues surveys understate willingness to pay because buyers hedge, so the live sales call is the only honest test. Say the price, wait for the gasp, and let refusals rather than a spreadsheet find your ceiling.
counters that willingness to pay is measurable before you ever quote a number. Survey best-fit buyers on specific features, find the 20 percent that drives 80 percent of it, and price the packaging around that. Guessing high is not a strategy.
The split is really about your sales motion. The gasp test needs a human on the other end of a call who can react; a self-serve checkout gives you no gasp to read, only a bounce you cannot interpret.
From Gasp Pricing
would measure first. Learn willingness to pay from best-fit customers, let the value metric and the 20/80 split set the number, and treat any single quote on a call as one data point rather than the decision.
says the survey lies and the live call does not. Quote a high anchor in a real sales conversation and read the gasp, because what a buyer says on a form and what they pay when the card is out are different numbers.
Both are right in different motions. Run the measured survey if you sell at scale through a checkout; run the gasp test if you sell person to person and can watch a face change.
prices against provable value and treats a low anchor as a trap. Alex Hormozi pushes the same direction harder with the gasp test.
The land-and-expand camp
argues for deliberate underpricing in AI tooling to win the category first, then expanding wallet share later.
Ramanujam does not reject that outright. His condition is that it only survives if there is a real land-and-expand plan, not just a hope that price can be fixed later.
From The AI Pricing 2x2
ran the Van Westendorp meter on real prospects and priced Superhuman off it, treating the survey as a genuine signal from buyers rather than a guess. Asked well, what people say maps closely enough to what they will pay to set a starting price.
warns that what a customer says in a survey and what they actually pay are different things. Anchor on the value metric and willingness to pay for specific features, and treat any single survey as one input, never the decision.
Vohra ran the survey on people already using the product, which is a narrower and more honest population than a cold panel. The further your respondents are from having used the thing, the more Ramanujam is right about them.
treats pricing as an operating rhythm you run on a live business, one change a quarter, value metric first.
disagrees on sequence: run willingness-to-pay conversations before the product is built, on the grounds that a price discovered after launch is a price you are stuck defending.
Alex Hormozi disagrees with both on method, finding the ceiling by naming a number high enough that the buyer flinches. Campbell's version is the one that survives having no sales team, because the metric does the negotiating.
From The Value Metric
When AI changes what's worth building · 5 splits · Product strategy research
holds that intelligent design beats Darwinism, and that iterating at random does not get you from zero to one. The answer comes first.
is the counterweight: the fastest way to learn what is true is to ship something ugly and watch what happens.
Both hold at different altitudes. The contrarian answer picks the direction; the janky launch tests whether you can execute on it. Annie Duke's kill criteria keep the answer a bet rather than an identity to defend.
describes a model where teams are given problems, not features, and anyone in the building can run discovery and land on something that works for the customer and the business. Installing the titles and the ceremonies without that is why most adoptions change nothing.
is describing a company that has none of that yet: ship the jankiest version, filter a hundred curious people down to the five or six with a hair-on-fire problem, and let their reaction decide what is real. Discovery at that stage is the founder and five users, not a capability you distribute.
Cagan is writing for companies that already have a product worth operating; Tan is writing for the ones that do not yet. The failure mode is installing the model before anyone in the room has found fit, which produces the ceremonies Cagan warns about.
assumes you will have rivals and tells you how to beat them. Diagnose the crux, concentrate force at the decisive point, and apply a named source of power such as leverage or focus. Competition is the terrain, not a mistake.
calls competition a loser's game. Profits get bid away in a crowded market and companies converge on each other, so the real move is to build something new enough that nobody is standing there to fight.
Which one applies depends on whether an unclaimed position exists. Thiel is right when you can find one, and Rumelt is right the rest of the time, which is most of the time. Both agree on the failure they are arguing against: spreading effort thinly across a market you never diagnosed.
ran the test and left. Intel pioneered semiconductor memory, and when Japanese manufacturers became a 10x force in it, Grove exited the business the company had invented. A real inflection point does not respect what you are known for.
found the opposite pattern in the companies that made the leap. They picked the one intersection of what they could be best in the world at, what drove the economics and what they cared about, then said no to everything outside it for decades.
The condition is whether the industry itself changed shape or only your position in it. Collins studied disciplined consistency inside stable rules; Grove is describing the moment the rules stop holding. Note that Grove agrees with Collins most of the time: he ruled RISC out at 2x to 3x and kept the existing architecture, and half the RISC team left over it.
diagnoses strategy by walking the organization and asking every team what they are building and why. Convergent answers that trace up to a business metric mean the strategy was deployed; scattered answers mean it never left the slide. She runs the same probe on the vision, and a statement nobody can elaborate on is a tagline.
would call a company that recites one story aligned but not strategic. His four hallmarks of bad strategy include fluff and goals mistaken for a plan, and his heuristic is to strip the buzzwords and see whether anything substantive remains. On that reading a well-cascaded slogan is the more dangerous case, because the alignment makes it look solved.
The two tests measure different things and a company needs both. Perri's is a deployment check: did the strategy reach the people doing the work. Rumelt's is a content check: is there a diagnosis and a guiding policy underneath. A firm can pass Perri and fail Rumelt, which is a whole organization executing the wrong plan in unison. Failing Perri while passing Rumelt is the cheaper problem, because the thinking exists and only the communication is missing.
When nobody finds what you built · 4 splits · Distribution research
says pick one of the four and spend the first four hours of every day on it. Focus and repetition on a single channel beat a thin presence everywhere, especially when you are small and short on time.
argues the durable win is owned distribution: an audience, a community and channels you control, built mainly through content, so you stop renting attention from platforms. That is a slower, compounding play than grinding cold outreach.
Hormozi buys you customers this quarter and stops the day you stop. Isenberg buys you an asset that keeps paying and takes a year to show up. Pick against your runway, not your temperament.
treats the ad as a mechanism with a known failure point. Most ads die at the hook, so name the pain in three words, deliver the meat, and make one clear ask. Every element is there to be tested and tightened.
argues a perfectly optimised ad is also a forgettable one. Buyers run on memory and heuristics, not spreadsheets, and the distinctive, slightly odd element an optimiser would strip out is often the only reason anyone remembers you at the moment of decision.
Both are right at different budgets. When you are testing dozens of variants against a cost target, Hormozi wins because the mechanism is measurable. When you are one of forty near-identical ads in a feed, the measurable version is exactly what makes you invisible.
assigns one motion per phase and says the playbooks do not blend. In the tornado the job is to standardise, take coverage and grab share while the market is buying, because the reason people buy has flipped from your vision to wanting what their peers have.
argues share taken in a land grab is not the same thing as a business worth owning. Unless the share converts into a Power, scale economies, network effects, switching costs, the margins arrive the moment the tornado ends and competitors are still standing.
Moore is describing how demand moves; Helmer is describing what survives after it stops. Run the tornado playbook, but name which Power the share is buying you before you spend the quarter buying it.
sorts channels by ownership before performance. Paid search and paid social run on an algorithm somebody else controls, so spending there funds a competitor's distribution and leaves you exposed to a policy change you had no say in. She makes building the earned channel, powered by virality and word of mouth, the growth team's first priority.
puts paid ads alongside warm outreach, cold outreach and content as four equally valid channels, and tells founders to pick one and spend the first four hours of every day on it. On his account the failure is dabbling across all four, not choosing a channel someone else owns.
The two are answering different questions, and the deciding factor is your time horizon against your runway. Hormozi is optimising for a founder who needs customers this quarter, where depth in any one channel beats theoretical durability. Verna is optimising for the compounding asset, which pays off over years and starves if you never fund it. Note Verna does not tell anyone to switch paid off; her failure mode is cutting rented spend before the earned channel produces anything.
When you need an audience from zero · 4 splits
keeps at least three or four gives per person before any ask, and aims for 70% or more of the impressions a person sees to be gives. Once ads run, the ads carry the ask and his short-form content carries zero calls to action.
Donald Miller
holds that customers do not take action unless they are challenged to take action, so every marketing touchpoint needs a clear, direct call to action. He pairs the buy button with a transitional offer, a free resource, but the direct ask never leaves the page (Building a StoryBrand).
They are counting different surfaces. Miller writes for a website, email or landing page someone chose to open, where a missing ask wastes the visit. Hormozi counts the feed, where every ask spends trust and too many send people elsewhere. Put the clear ask on the page you own and keep the feed giving.
treats controversy as the growth engine. In a feed that rewards provocation over polish, an authentic and deliberately divisive take buys the attention that a well-behaved launch never gets, and the distribution is what then guides the product.
would say durable growth comes from sharp positioning, not provocation. Attention you win by being divisive is attention you have to keep re-winning, and it tells a buyer nothing about why they should pick you over the alternative they are already using.
Both can be true in sequence. Controversy buys the first wave, positioning decides whether anyone stays once the noise dies down. The failure mode is a company that can reliably get talked about and has never worked out what it is.
says any skill legible enough for someone else to copy will not hold its price, so the work is to become the only person who does what you do. Substitutability, not talent, is what sets your rate.
would call that pricing power without a barrier. A position that lives in one person does not transfer, does not scale past their calendar, and cannot be sold, which is the test a Power has to pass.
Both hold at different scopes. A personal monopoly is a genuinely good answer to what should I do with my career and a genuinely bad answer to what makes this company worth something without me. The move is to convert the one-of-one into something the business owns before it becomes the ceiling.
From Personal Monopoly
splits writing into two passes and drops the quality bar on the first to zero. Get the thoughts out without judging them, then shape what came out for one specific reader. The blank page is what happens when you try to do both at once.
attacks the same anxiety from the other end. Set a timer for 33 minutes and 33 seconds and forbid yourself everything except the copy, with no obligation to write a word. Boredom does the recruiting, because writing becomes the most interesting thing left in the room.
Handley removes the standard, Schwartz removes the alternatives. If your problem is a harsh internal editor, Handley. If your problem is that you never sit down at all, a permission structure that does not require output will get you there when a pep talk will not.
From The Ugly First Draft
When you have no customers yet · 3 splits
picks 100 specific names and works them for thirty minutes a day over a month, following up ten times each. The email has to be specific enough that it could not have been sent to anyone else, which is exactly what makes it unscalable.
Will Cannon
took UpLead to its first million in ARR on the back of one cold-email template. His point is not that personalisation is wrong, it is that a message which genuinely converts is an asset you should run at volume, not send once.
Carles Reina, who scaled ElevenLabs past 350 million in ARR, sides hard with Parr and says automated outbound is what broke the channel in the first place. The reconciliation is sequencing: hand-write until something works, then decide whether the thing that worked survives being sent a thousand times.
From Cold Email Method
says startups do not take off by themselves, so you recruit your first users one by one, by hand, and do the unscalable work that delights them, the way Airbnb went door to door. The manual work is what teaches you what to build.
says distribution first, product second. When AI makes building nearly free the code is commoditised and the moat is an audience, so grow one to about a thousand people, ask them what they need, and build into a warm market.
Isenberg wins in commoditised categories where your audience is your buyer. Graham wins when the product needs a human touch to be any good and ten real users teach you more than a thousand followers.
says a network product has a threshold: some number of people connected at the same time doing the same thing, or the product is worth nothing to any of them. Ten thousand users spread one per company is a product nobody can use while the dashboard reports growth.
says recruit users one by one, by hand, and do the unscalable work that delights them. The manual work is the point, because it teaches you what to build and it is how every startup that looked like it took off actually started.
For a single-player product Graham is simply right. For a network product, delighting one user in isolation buys you nothing, which is why Tinder started inside colleges and Slack started inside one small team. Work out whether your product needs one happy user or a dense pocket of them before you choose a launch plan.
From The Atomic Network
When anyone can copy you · 3 splits
The single-founder, multi-business camp
argues for a portfolio of bets with operators hired to run each, treating distribution as something you scale across many products.
The all-in camp
argues that one audience, one product, one obsession compounds faster than diversification.
Both agree the moat is distribution. They disagree on how many moats one person can dig at once.
From Distribution as Moat
says the durable win comes from a business model the incumbent cannot copy without damaging its existing business. The barrier is not the idea, it is the leader having something to lose by matching you.
positions you inside a market category buyers already understand, so they arrive with the right expectations and grasp your value in seconds. A position nobody has a frame for is a position you have to fund the education of.
Counter-positioning is a claim about the incumbent; positioning is a claim about the buyer. You can have a model the leader will not match and still lose because nobody can tell what you are. Helmer decides whether the win lasts, Dunford decides whether it starts.
says software is not a moat. Features get copied, and distribution is what actually holds.
argues in 7 Powers that software businesses absolutely can hold durable power, through counter-positioning, switching costs and scale economies.
Both are right at different layers: a single feature is not a moat, but a system of features wrapped in switching costs can be. Read them together before you bet the company on either.
When customers keep leaving · 3 splits
says even the right customer with the right job churns if they never reached the result. The cause is delivery, weak onboarding and undelivered value, not the feature set, so the fix lives in the first thirty days.
says churn that is abnormal for your segment almost always means you sold to the wrong ideal customer profile. The product can be fine and the buyer wrong, which makes it a positioning and go-to-market problem, not an onboarding one.
Bob Moesta makes it three ways: the job the customer hired you for changed, so the switch reversed. Onboarding, ICP and job mismatch lead to completely different quarters of work, so diagnose which one is yours before you commit.
says churn that is abnormal for your segment almost always means you sold to the wrong ideal customer profile. The product can be fine and the buyer wrong, so the fix is positioning and go-to-market, not a quarter of engineering.
says the customer hired you for a job, then their context changed and the job moved on, so the switch that brought them in reversed. The fix is to interview the churned and find the struggling moment, not to re-segment.
Alex Hormozi makes it three ways: even the right customer with the right job leaves if they never reached the result, which puts the fix in the first thirty days. The three diagnoses lead to completely different work, so run all three before you commit.
says behaviour happens only when motivation, ability and a prompt land in the same instant, and that motivation is the least reliable of the three. So you shrink the behaviour and place the prompt, instead of trying to make people want it more.
says nobody switches because a prompt arrived. They switch because a struggling moment got big enough, and the forces of progress, push, pull, anxiety and habit, decide whether the switch happens at all.
Fogg is designing the last inch and Moesta is explaining the mile before it. Make the action tiny and prompted, and you still get nothing if the person was never struggling. Find a real struggle and Fogg tells you why they still did not act.
When you're stuck in your own head · 3 splits · Mindset research
holds that the destination is better and the path there is a continuous chain of setbacks, so you budget for the wreckage in between. Room for error is what keeps you in the game long enough to collect the long-run return.
Phil Knight
ran Nike the other way, reinvesting every cent into inventory and growth and living in permanent tension with his bankers, on the view that a company growing slower than its demand hands the market to someone else.
The deciding variable is reversibility. Floor it when the downside is recoverable and hoard the buffer when a single bad month ends the company. Housel is describing a personal balance sheet; Knight was describing a land grab with a closing window.
makes the constraint external. Thirty-three minutes and thirty-three seconds in the chair, nothing else permitted, and no requirement to produce anything. The prohibition on everything else is what makes the work start.
makes the constraint internal. Give yourself explicit permission to write badly, get the thinking out in a first pass with no quality bar, then edit for one specific reader in a second pass.
They are treating two different causes of the same symptom. Schwartz solves not sitting down; Handley solves sitting down and freezing. Running Schwartz with Handley inside the block, a timer plus permission to produce garbage, is the version that survives both failure modes.
From The 33.33 Method
says repetition is how habits form. Run the stack in the same order every morning and it gets smoother. Writing or training still never turns as automatic as brushing your teeth, so you keep showing up.
wrote the anchor recipe Clear borrows, and says emotion does the wiring. Celebrate on the spot, right after every new behavior, because you change best by feeling good, not by feeling bad.
Clear likes celebrating small wins too, to feel progress while results are slow. They split on what makes a habit stick. If a stack keeps dying in the first week, shrink the new behavior and add Fogg's celebration. If it survives but still feels like work months in, Clear's answer is to keep running it.
From Habit Stacking
When you keep delaying launch · 2 splits
argues the MVP bar is not enough on its own. Software keeps shipping after launch, so he cuts features rather than quality and insists one part of a first release be good enough that a user repeats it to someone else, because that part is what carries the vision the release cannot yet deliver.
says put out the jankiest version that still delivers value and let real reaction decide, on the grounds that fear of launching kills more startups than competition does. Polish before contact with users is time spent on assumptions.
The condition is whether buyers already have something that works. Tan probe answers whether the problem exists, and in a young category a rough build gets a clean read. Figma launched against Sketch and Illustrator, where the open question was not whether designers had the problem but whether they would switch, and a rough build there measures your polish instead of their demand. Field bar is also not permission to keep building: he names the trap himself, that chasing the awesome thing means sweating every detail and never shipping.
says ship the jankiest version now and let real reaction decide, because fear of launching kills more startups than competition does. The first launch is a probe, so move fast on anything you can reverse.
says measure before you commit. Forty percent of active users answering very disappointed is the product-market-fit line, and scaling on a vibe instead of that number is how founders pour fuel on a fire that is not lit.
The two answer different questions. Tan is telling you how to get data; Ellis is telling you when to trust it. Launch janky, then refuse to scale until the number says the thing is actually working.
From Launch-Jankiest Rule
When your AI-built app has to work · 1 split
argues application software trends toward free as agents climb the levels, so the bottleneck moves from production to ideation and generalists who know what to build win.
Michael Truell
counters that vibe coding is not sufficient for professional development. Programming and math stay valuable, and taste plus logic design only pay on top of real technical understanding.
The split is about what breaks at level four. If your failure mode is shipping the wrong thing, Masad is right that judgment is scarce. If it is shipping a thing that quietly does not work, you still need someone who can read the code.
When people don't believe you · 1 split
argues that traditional PR is dead and the founder should take the story straight to the people who matter, through channels the company owns. Every handoff from PR person to reporter to editor dilutes the message and filters it through someone else's worldview.
Michael Seibel
tells early founders to do their own press instead: build relationships with reporters the way you would with business partners, and offer each story as an exclusive to one tech reporter at a time, with one story every month or two.
Both cut out the PR agency; they split on whether the reporter is the channel. If you can name the people you need and reach them yourself, go direct. If you need the outside credibility of an article, which readers and AI assistants both lean on, Seibel's exclusives earn it, and Meservey agrees the press helps when it is a choice, not a dependency.
From Going Direct
When you're deciding whether to raise · 1 split · Fundraising research
narrows the standard team, problem, market and model checklist with founder-market fit, and puts a founder's willingness to outlast everybody else above the idea. He asks how big a market can become, not how big it is today.
uses customer calls only to confirm numbers he already likes: a SaaS company at $1M ARR growing about 10% a month with low churn, and for enterprise software a net revenue retention above 110%, or he does not invest.
Stage decides. Lin is describing seed and early rounds, where no numbers exist yet and the founder is most of the bet. Lemkin is describing a company at $1M ARR, where the metrics exist and a strong story cannot cover weak ones. Raise on fit before revenue; past $1M, expect your metrics to be the pitch.
From Founder-Market Fit
When you don't know what to sell · 1 split
holds there are only two viable pricing positions, lowest-price leader or high-value leader, with the middle being death. The money model leans on a high-ticket front end that funds delivery.
Kim and Mauborgne
argue the opposite for blue-ocean plays, pricing for the mass of target buyers rather than the premium few, because volume is what makes a new category defensible.
Both hold. Hormozi's version works in service businesses where margin funds delivery; mass pricing works when you are scaling a product and volume is the moat.
From The Money Model
When AI is changing your job · 1 split · Career research
treats your own curiosity as the guide. If you read about a field at night without being asked, and its history fascinates you, go deep in that field. The depth is what sets you apart.
warns against working forward from something you have labeled a passion. He has people picture the daily life they want a few years and ten years out, then work backwards to the career capital that would pay for it.
Gurley answers which field to go deep in. Newport answers what the depth should buy you. If you already know your field, use Gurley's two halves to get good at it. If you are choosing between fields, sketch Newport's lifestyle picture first, then run Gurley's boredom test on what is left.
From History and the Edge
Cite as: Arun Agrahri, “Where experts disagree,” Gavel, October 2026, https://usegavel.com/research/where-experts-disagree
