The Physics of Startups
Most 0-to-1 and product-market fit advice is a pile of things that worked for someone, once. This show is built on causal relationships, not tactics.
Rob Snyder has helped 50+ startups go from $0 to $1M ARR. He's the creator of the PULL framework and author of "The Power of PULL" which explains why customers buy and why some startups scale and others stall.
For VC-backed and bootstrapping founders who are tired of advice that sounds smart but doesn't work.
More about Rob: https://www.robsnyder.org
The Physics of Startups
No more PULL Hypotheses! (introducing the PULL Anecdote)
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
Sorry, folks. The PULL Hypothesis is out. Rob observed that founders and teams were holding too tightly onto the PULL Hypothesis and too tempted by abstractions, which prevented them from understanding and serving customer demand.
BUT - fear not. Today, we discuss the PULL Anecdote, which, speaking personally (as Rob's cohost), I think is a hugely liberating concept and I'm excited to put into action. By tying explicitly to a single, real human being's real PULL, the PULL anecdote overcomes important limitations that were making it hard for founders to get out of "abstractions" to "reality" and really seek to understand what customer's are trying to accomplish but are blocked.
Post: No more PULL hypotheses
PMF Camp! Interested in the Sept PMF Camp? Info here: LINK or reach out directly to me: rc@reframeb2b.com
Links
- Rob's Substack: The Physics of Startups
- AI Rob waitlist: https://robsnyder.org/ai-rob
- Work with Rob 1:1: https://www.robsnyder.org
- Get The Power of PULL: Amazon (https://amzn.to/48KC1yv), Barnes & Noble: (https://bit.ly/4dyMTSv)
- Connect with Rob on LinkedIn: https://www.linkedin.com/in/rsnyder1/
Hey Rob. It's good to see you, man. You are you're you're back from not just vacation. You flew out to work with uh uh one of the early startups uh that you helped get past a million now they're they're growing past that. Um it's it's good to see it to be back in the saddle.
SPEAKER_01Yeah, well, one thing I was thinking about, which is like, you know, I have an 18-month-old daughter now, and one of the things that I have become quite aware of is like as being a dad, one of the one of the kind of like rites of passage is uh now when I travel for this trip, I arrived at the airport an average of four hours before my flight. Just you know, sitting, feeling good. I like I also have this kind of like a desire to start learning more about the World War II battles, uh, certain World War II battles.
SPEAKER_00And so it's like, I don't know, there's there's just something about being a dad that that forces you down this path. I have been hoping it was a generational thing, but uh it's it's probably like an individual evolution thing. No, I'll just start tucking in polos into my jeans. Like that's that's where we're going.
SPEAKER_02Uh well, hey Rob, we've got uh a really you know, you know, you gave us you gave us a gift and you're taking the gift away from us. Uh this week's post is called No More Pull Hypotheses. We're gonna we're gonna talk about what was happening, what you were observing with pull hypotheses, uh how founder like the limitation basically that it that was causing founders to or the problem that they were stumbling over. Uh and then we're gonna get into uh the pull anecdote. Real fast before we jump into that, uh, we're gonna publish this one, I think, September 5th. We've got a PMF camp that's coming up. It starts September 14th. Um, just a few quick sentences like that. The goal of this um uh is to get you know, startups that are a good fit are looking to get Rob's direct engagement on their pool, on their go-to-market, on their outbound, um, and to see an you know another group of startup founders like develop their intuition by watching Rob uh and and the whole group really interact around how other founders are approaching outbound, how they're how they're thinking about the case study factory, um, doing teardowns of live calls. It's gonna be uh it's a two-week, uh, two-week PMF camp, starts on a Monday, ends on a Friday, includes uh four working sessions, a couple AMAs with Rob, and uh two teardowns of full sales calls from other founders in the community. Rob, um, anything that you would add on on that before we jump in?
SPEAKER_01Uh I I just I just say like the the thing that has been consistent. We've now done, I have no idea how many times we've done this now. We've probably done it like 10 times. And so it always kind of evolves each each time. Um this time the flavor we're doing is just like way, way, way more feedback per startup. So basically, all of the sessions, say for say for one are like, you're gonna bring something and get it roasted or torn down or or uh you'll get helpful feedback on it. And then one of the things that so everyone everyone who comes into it is like, yeah, I want feedback on my thing. And so that's that's awesome. Whether it's your outbound message, whether it's your um you know sales deck, whether it's you know, whatever, whatever that thing is. Um one of the things people do not expect to get a ton of value out of, but do find they get a ton of value out of, is like watching other founders' sales calls and watching me tear down other founders' sales calls because it's actually often um you you'll have a lot more epiphanies when you see somebody else. And like you see something really clearly on somebody else's call or in somebody else's thing, and then you're like, wait a second, I need to apply that to myself. So um that that's been that's just been a fun, fun thing.
SPEAKER_02Yeah, I I totally agree with that. Um the the the public teardowns that you do and uh are some of my favorite things, the most impactful things that have led to changes in my business. Um you mentioned this there's a lot of one-on-one or like a lot of direct feedback. We're capping the uh PMF camp at 15 startups. Um and so if you're interested, reach out, ASAP, um, and we'll just go from there. Well, Rob, if there's if there's no uh oh, we'll put the I'll put the link in the in the show notes for anybody that's interested. It's just like a short application. Um nothing else. If there's nothing else on that, why don't we jump into no more poll hypotheses? Um all right, Rob. You've we got everybody running on poll hypotheses, and now you're saying like, all right, total like orthogonal. We're going in the other direction on this one. Um why is this one on your mind? Uh, maybe like who's it, who's it for? Let's just start there.
SPEAKER_01Yeah, so this is this is um especially relevant if you're going zero to one. And and you could be going zero to one on a new product, you could be going zero to one on uh like your second product on your first starter. Like it could it could be kind of anybody who is trying to go zero to one on anything or in the kind of like zero to one stage. Um uh this is gonna be relevant for you. And it's going to be relevant specifically because I'm taking a concept away. And I'm taking a concept away that sounds right, but there's a one word that seems to cause us to occasionally, not always, but steer straight into the iceberg. And that word is hypothesis. Now, like zooming out if you're new here, uh, the pull hypothesis was a is was a thing that is it's basically just a statement of I believe this person has pull or this kind of person has pull. Um, and pull, if you're also new here, the pull framework is a way to state uh somebody who's weird not to buy your product, who would be weird not to buy your product, where they're trying to do something, but they can't do that thing. They need to do that thing right now, can't do that thing. They're grinding against their existing tools, methods, and options, and you show up and unblock them. So they have like a blocked product on their to-do list, you unblock them. That is what the pull framework is. A pull hypothesis is a um, I believe this specific person has pull, and it's um, and what we're killing today is the word hypothesis.
SPEAKER_02Yeah, and here's what you write uh about that. Um hypothesis comes with baggage. It feels like the kind of thing that originates in your brain, which leads a lot of founders to sit around and guess as a way to create their pool hypothesis, this despite the real person rule in my book, which states that a pool hypothesis must be based on a real person with pool. So, Rob, what were you? And and honestly, I'm pretty sure I was guilty to a certain extent of of uh abusing the word hypothesis or uh in the pool hypothesis here over uh on a couple occasions. But Rob, why don't you give some color? Like, what did it look like? Um, you know, for a founder that's using a pool hypothesis, yeah, um, just so that they can self-reflect on it, uh, what is what is going wrong? What does it look like when it's being used wrong?
SPEAKER_01So the word hypothesis just makes you think of uh I need to think um first. Like I need I need to come up with a hypothesis. And so the method that most people do, and and this is I think due to the word hypothesis, is you'll open a dock or you'll go walk to a whiteboard and you'll just start brainstorming plausible hypotheses. And these plausible hypotheses, like you know, I I was doing a workshop uh this week with a with a another company, and and they basically showed up to the workshop with a huge dock of possible poll hypotheses. And that's that's the normal thing. It's like, okay, cool, who do we think might have pull? Um but which which like this totally sounds like a reasonable thing to do, but the challenge is that something that originates from your brain, it's usually based on like some abstract market logic or something of like, well, agents are taking over the world, so therefore, or this startup just took off, so therefore, right, like it's it's kind of abstract, and you take that and you turn it into a belief about a specific person or like a hypothesis about a specific person, that hypothesis is basically guaranteed to be wrong. Which again, like okay, cool, it's it's okay if it's guaranteed to be wrong, but the problem is the the the word hypothesis, so it that causes you to do this brainstorming exercise and come up with something that is basically guaranteed to be wrong. And then the word hypothesis, for some reason, it makes us cling to it's my hypothesis, yeah, which blinds us to what poll actually is. It prevents us in a lot of different cases, not all the time, but but often enough, it like having that hypothesis is the thing we go out looking to validate the hypothesis. Um and and so I've seen startups do tens, hundreds of sales calls, like making minimal tweaks to their hypothesis that they have no reason to believe other than the fact that it came from a whiteboard and it has the word hypothesis and it sounds like science, right? And so, um, and so that's the like problem with the the word hypothesis is it causes us to go to the whiteboard, create something that is wrong by default, and we can explain explain why, and then cling to that. Yeah.
SPEAKER_02That there's a a couple parts of that. You know, I've I ran with a poll hypothesis for longer than I should have. Um, and they don't, it's kind of like they feel totally believable. It's like you're in this like kind of mind warp thing. And as a result of it feeling like something that you think exists, like even objectively, um, with without having validated it, like I built too many slides, like I redid our website, like you know, there's there's like a lot of this investment that happens, which is an indication of I'm holding way too strongly onto something that a uh, you know, your real person rule, a real human being has not said to me in a meeting is something that they're trying to accomplish and are willing to pay for. Uh, but uh Rob, are there you know, what was maybe we hit real quick the intent. Like the intent of the poll hypothesis was not originally to, you know, go down a theoretical rabbit hole. Like there was there was a purpose of uh related more closely to learning or learning through error, or you know, what what is the thing that we want to not lose from the pull hypothesis that that that was the reason you came up with it?
SPEAKER_01Yeah, so it's so basically we're just killing the word hypothesis. The concept is going to stay the same, but something about the word steers us into the abyss. So we're still going to have a pull framework, but the difference is that we're not going to call it a hypothesis, and we're still we still have the real person rule, and we're and we're turning it from we're just turning that when when we come up with the when we use the word hypothesis, we create this kind of plausible reality in our head. And we have to get rid of that plausible reality that doesn't reflect actual reality. And so all everything else is the same here. It's it's but we're just gonna call, we're gonna use different words because words have like really, really important implications of uh they they they create behaviors uh uh for us. And so we're just moving from pull hypothesis, which causes us to create this plausible world often, to pull anecdote, um, which is just a way to combine pull hypothesis and real person rule into one thing, one much simpler thing, uh which I like. We'll come back to that. Um uh which is to say, okay, cool. I don't even we can't have nice things. I can't use the word hypothesis. I have to find an actual anecdote of a person with pull. It could be me, or it could like, you know, or it could be somebody else, whatever, but it has to be an anecdote. I don't get to go to the whiteboard or to a doc and and kind of like write out a bunch of a list of hypotheses.
SPEAKER_02Yeah. Okay, let's let's hit a couple things on this. Uh, you write about how using an anecdote clarified um like the workshop that you ran this week. You you know, they they they came in with a document, a bunch of plausible pull hypotheses, probably struggling to to choose like where do I where do I focus, where do I go with this. You move to an anecdote, it made it, you write that it made it way more clear. What actually happened? Like um uh and and then let's I I want to start with like how did it feel like what was the result? And then let's get into what is an anecdote actually look like.
SPEAKER_01Yeah, so um, yes, and and this is like, okay, so if you if you look at a list of hypotheses, you can uh what happens is you can create a ton of hypotheses, all of which sound like they should work. They all sound like they make a bunch of sense. You can kind of map out where they could lead as a business. You can you can do all of this work on them. And then what you're doing um is you have a bunch of hypotheses in a doc or on a whiteboard, and you then have to like figure out how to prioritize across these different hypotheses and which one's most attractive, which one's most plausible. You have to do all of this work, and you're doing it in like 40 different dimensions um because you're like looking into the future, you're like doing all this different stuff. Do we know like and well, what you know, but what about Claude Code? What about the you know next model release? And you're doing all of this kind of insane work in your head. And so what happens is you come up with a doc, you spend hours just like hitting your head against the wall, trying to come up with some, and then you come to some conclusion at the end of that that's like, cool, we're gonna prioritize this one or these seven uh different directions. Um again, all of which are unlikely to have any grounding in the real world, to have even a single person with real pull, but you can't see that because they're all on equal footing in a dock. They all seem, they all have words, they all they all seem coherent. And so what happens is you just say, okay, cool, we we're not allowed to do that. We can't do that. What evidence do we have of actual pull from a real person? And when you do that, almost everything in that doc gets cleared out, and you say, Oh, interesting. Yeah, so uh helping kind of I've done this on two sides. Uh the one is for a zero-to-one company, one is for a kind of like later stage company looking to launch new products. Um later stage company looking to launch new products is like, what have we seen in our existing sales funnel? What have we seen from customers? Are they asking, are they trying to buy additional stuff from us? Have we had conversations where people have tried to buy stuff from us that we didn't have? Right. And it's like, it's that's where the anecdote comes from. And that's where that's and so what we're looking for are anecdotes of real people who have had some form of pull or like seem to have had some form of pull, but we didn't double-click into it and we can go back and figure that out. And that's what we're looking for in those occasions. And what that does is that takes that the the entire, basically the infinite possibilities and constrains it just to like what what is real based on what have we seen to be real?
SPEAKER_02You know, it's it's funny. It's like how I feel like I spent 15 years being like, oh, that's just an anecdote, and like discounting it to zero, and then kind of like doing the opposite, which was like going to I don't know, what does data say, etc.? And increasingly with your approach, it's like the anecdote is the thing that really matters. But uh let how let's before we go to the zero to one, let's say, um, let's say that maybe maybe they're not, you know, uh uh Series B, Series C, like maybe they're not huge, but it's a startup that has some customers. What should they be looking for in an uh in a customer anecdote to develop their kind of pull anecdote here?
SPEAKER_01Yeah, so I'm saying, like, if you if you what often happens when you have existing customers and it's it still feels like you're kind of pushing, is that um you will not focus on the anecdotes of your existing customers or the people who have been in your sales process or the people who have churned out or whatever, right? And instead you go to the whiteboard to theorize, to create a hypothesis. And so so if I were gonna keep the word hypothesis, I would say the hypothesis has to be generated from an anecdote of a real person. And so that's that's always, but but the word hypothesis prevents uh that from happening. So that's why we're getting rid of it. But what they have to do is they have to look into the anecdotes of sales conversations they've had or whatever, or whatever. Um, and if you're PLG, it's helping a company that's PLG right now, it's like, um, well, we need to go and talk to the people who have either signed up and used it and churned or or what. Like we just but we need to get to the anecdote of real people who have chosen us or not chosen us for some reason and try to try to figure it out from those anecdotes. And then you come up with like a okay, you come up with a pull anecdote, which is here's a real person who had this on their to-do list. They could, for this reason, they tried this other option, didn't work for this reason, therefore they chose us and they bought it relatively quickly. It's like, okay, cool. I have an anecdote that is a real thing. I need to see if that repeats and if that's the right shape that repeats. And so then you can either go and look at your other existing customers and see with that with if that repeats with them, or you go out and use that to run a sales sprint, right? And say, okay, cool. There, I'm just gonna see if this anecdote repeats rather than this hypothesis that I just crafted in my mind. Yeah.
SPEAKER_02It feels like um the pull hypothesis previous version was closer to like, I'm a um, you know, go-to-market director that's trying to implement and, you know, there, or I'm a go-to market director in a SaaS company that's trying to implement XYZ or that's running into, you know, these limitations. Is this one like should people be using like the name of a human being? Should it be written in the past tense? Like, how should how should they think about crafting this um in a way that's effective?
SPEAKER_01Yeah, I think I think it's I think like that's not super I I would make it like the exact person. Uh I'd make it the exact person, I would not abstract it very much at all. And I would try to just make it as close to the actual anecdote as humanly possible.
SPEAKER_02And let's say that a founder has a couple anecdotes that they're they're choosing between. What what makes a more compelling anecdote to go with?
SPEAKER_01Yeah, so um uh what you want to do is you uh the useful thing about the poll framework is you will like actually write out what poll is there, and you can look and you can compare different anecdotes with each other. And you might actually say, Oh, wait, these are actually the same anecdote written just three different ways. That's often what happens. And it's like, oh, here's the repeatable anecdote, or here's the repeatable case study, essentially. Um but what will often happen is you'll write out three anecdotes, and two of them you'll be like, huh, those people could have bought something else, and they just bought us because blank, right? Like they actually didn't have pull, which explains why the sales cycle was kind of slow, which explains why blah, blah, blah, blah, blah, blah, blah. Um, and so pull flame gives you a way to assess, is it real pull? Do I actually believe that this anecdote repeats? Is this good enough?
SPEAKER_02Yeah. Well, um, and then is is there anything else that changes that like I know it's just the word, uh, it's really like a reinforcement of the real human, uh, real human rule. Um, is there is there anything else that changes in the way that you use the anecdote versus uh the way that previously folks were using hypothesis?
SPEAKER_01No, I just I think uh what I hope an anecdote does, and what I have experienced, the idea of that anecdote does, is it doesn't give you a thing to call your own. Um and so you're gonna care less about the anecdote being true. Or or like about the anecdote. Because it's not, it's not your it's somebody else's anecdote. It's not your hypothesis. Somebody else's anecdote, your job is literally just to go and see if that repeats. And if it doesn't, cool, okay. Down to figure out something else.
SPEAKER_02Let's talk about that. Um, and so it's you know, we've we've talked about the Hail Mary question, or um, you know, but I think what you're basically you're saying here is like there's a point in a sales conversation where it's like not working or that they are not pulling. It becomes obvious if you were, if you were, if you took your mind out of your body and observed the call that was happening and you were like, okay, they are obviously not pulling here. Um that's the moment where it's like, I either am tied to this hypothesis or I'm not tied. To it. What should they do? Okay, let's say that the anecdot, you know, using an anecdote makes that easier to stomach. How should they be extracting themselves from a situation where it's like, okay, had an had an anecdote, had a hypothesis, and it was uh it has been, it is not playing out. Like, what should their response in that case be that the hypothesis was preventing?
SPEAKER_01Yeah. So this is this is also the nice thing about it, because when you have something called a hypothesis, your intuition is going to be to go back to a hypothesis and like work on the hypothesis. There's no such thing as a hypothesis anymore. You have to go find an anecdote, which which just pushes you to, you can ask the Hail Merit question. Is there anything you'd pay me? You'd pay someone, you is there anything you came to this conference to look for that you haven't been able to find? Right. Is it like those kinds of things? Um, and and so it's um uh yeah, it like removes that step of going back to the whiteboard every time and like working, doing another version of the hypothesis, doing another version of this awful exercise where you kind of like stare into the infinite of this. And instead, it's just okay, cool. Like I think this actually changes a this I we already had the simplest way to go zero to one. Um, and it's just been made even simpler, which is um you don't even craft a hypothesis in this. You can have a search area or something, you can have a person you're looking for, or or something like you have something, but your job is not to have a hypothesis, it's to go find an anecdote of somebody who wants to pay to solve something, to get something off their plate, to that who's trying to do something blah, and will pay for it. And you don't like that's the job. There is no no other there's no other activity.
SPEAKER_02And and what one, I love that. I I just think it's I think it's way more liberating. Um and it removes that temptation to to tweak. It's like, oh, okay, well, maybe maybe I got this part like framed a little bit wrong. Um, and it's it's increasingly like my belief is tell me I'm wrong here, Rob. It's like if you find pull, like the it should be strong enough to overcome those like kind of small things, but if you until you are willing to just like you know, control a delete basically what a hypothesis is and replace it with an anecdote, you're gonna tweak things that maybe maybe it's around the corner here when really it's it's likely not there. Yeah, yep, 100%.
SPEAKER_01Yeah, they're like oh well, we could probably like if you go back to um like the founding of Microsoft, the if you if you look at the history of Microsoft, um they saw in it in Popular Electronics magazine that there was a personal computer. Um the Altair, like it was and it was like this cool, cool thing. It was brand new. And what did they do? They have a they have a hypothesis and go and like run some experiments and do some research. Or no, they called the person, they called the company out in Albuquerque and said, Hey, would you pay us to do XYZ for you? And he said, the, the, the, uh, I forget what his name was, but he answered and said, uh, yeah, I've gotten seven calls like this. If you come and show me that it works, yes, I'm gonna pay you to do that. And that's when they started writing code. That's when the but if he had said, no, I wouldn't pay for languages, but I'd pay for this other thing, they probably would have done that other thing, right? And so, um, and so the kind of if you look into like the history of entrepreneurship and read these stories, it's not like the the concept of this hypothesis, I think, comes from our desire to like scientize a thing where scientific language doesn't seem to help.
SPEAKER_02Yeah, I I really I like that example. Let's hit a couple things uh before we wrap on this one. One uh that I'm I'm 98% sure if we don't talk about this, somebody is gonna leave it in the comments. Um, one is does this apply to consumer? You mentioned PLG. Does this apply to deep tech? Like how does it um does it abstract across all different are there limitations um that you know we should be thinking about with an anecdote? Uh no. Um uh so there goes the excuse of uh special snowflake. Let's talk about zero to one. Um uh what if you're pre-anecdote? Like what should the founder that is like just getting started, or what do they do with this?
SPEAKER_01Yeah, it's it's like pick some sort of a search area, which like we could call a hypothesis, but it's not a hypothesis of like you want X, it's just a hypothesis of like, here's who I want to talk to, here's who I want to serve, we'll see. Um, and then you basically go and find an anecdote, which is to say you go in person with them, you go talk to them, you get them as something, and try to figure out is there anything you would pay to get off your plate right now? And you don't come with a I believe you want to pay for X right now. You just you come hypothesis free, looking for an anecdote. Is there anything you're trying to buy that you haven't found? Is there anything you're trying to switch out? Is there anything? And then you can kind of pull up, fill out the pull framework from there and decide is that something we actually want to tackle. But your job is it's this should be super liberating. You don't have to sit around and come up with a ton of ideas and kind of predict the entire future. You just have to go find an anecdote and then see if it repeats, see if you want to serve it, da da da. But but the job is just find one anecdote. And if you're not doing the things that would enable you to find an anecdote, then it would be weird if you found both.
SPEAKER_02Yeah. So it feels like I knowing our audience, which has a lot of computer engineers, a lot of folks that are focused on coding and developing AI applications, it's if you're in that zero spot, it's you know, choosing who is the customer that you want to serve, uh, doing whatever it takes using that hierarchy of interestingness, which we talked about, I don't know, two weeks ago or something like that, to try to get those conversations and uh coming in literally with the objective of trying to generate the anecdote. And uh there's probably like you know, some specifics you mentioned, the Hail Mary question. Um, you know, what what would you pay me to do?
SPEAKER_01Are the frame of the conversation is still the same, but it instead of coming in saying, I believe like I want to see if you have pull for X. It's I just want to see if you have what if you have pull for anything and what if so what? Yeah.
SPEAKER_02Yeah. Um awesome. Uh Rob, anything else that we wanna we wanna hit here before we wrap up today? Yeah.
unknownNo.
SPEAKER_02Okay, cool. Well, uh like we mentioned at the start, the PMF camp coming up September 14th. We'll we'll leave the uh my contact information and the link and uh for information in the show notes. Um and hope to see some of y'all there. Otherwise, Rob, we'll catch you in a week.