# Your positioning is mostly likely a guess *Use 10-15 won/lost deals to find where your positioning should point, and whether the fix is a new message or a new product.* **Written by JD Prater, 3x VP of Marketing.** This piece is part of *Open Folders*, which draws on marketing science, consulting research and practitioner experience to open the parts of the discipline many marketers never got taught, so they can lead all of it. Newsletter: https://jdprater.substack.com Website: https://jdprater.com Original post: https://jdprater.substack.com/p/your-positioning-is-mostly-likely --- We'd just moved to sales-led at an early-stage B2B startup, and the CEO was really working through what that meant. We were sitting in the office with a whiteboard up, trying to map some things out, when he asked the question he'd been chewing on. Who is our ICP? We had some ideas about who it should be. None of them were formalized, and none of them were rigorous. My first thought was, dang it. I should've had that answer a month ago, or at least a hypothesis we were vetting against. Over the next few weeks I built out who I thought the ICP was. We didn't have much product-market fit yet, so it took a lot of interviewing the founders to get the information out of their heads, and a lot of looking at what our customers were trying to do. The answer showed up in the demos and the deals we won. We'd thought it would be developers and a product-led motion. It turned out to be a PM at a much bigger, more established company, where the problem was bigger than we'd thought. That's when I learned you're figuring out the type of workflow, not just the type of company. --- ### A filled canvas is a record of what the room believed If you've sat through a value prop workshop, you've made one of these: Strategyzer's canvas, with the customer's jobs, pains and gains on one half and your products, pain relievers and gain creators on the other. It went everywhere after *Value Proposition Design* came out in 2014, and it deserved to. Then the sticky notes go up. Somebody photographs the wall, the photo becomes a slide, the slide becomes the source of truth, and eighteen months later the canvas is still hanging there with all the authority of research nobody ran. The boxes are right. The filling method is memory. When a team hands me anything, it's usually a Google Doc: a fill-in-the-blank positioning statement, maybe a messaging house, a persona deck that got built once and never touched again. When there's no doc, it's still in somebody's memory. I've built these off what a workshop remembered, and I've built them from customer and prospect calls with market and category research layered on top. The two documents barely look related. Strategyzer's own definitions of fit all start the same way: you have problem-solution fit when you "have evidence that customers care about certain jobs, pains, and gains." Product-market fit, when you "have evidence" that your pain relievers and gain creators are "actually creating customer value and getting traction in the market." One of their ten testing principles is three words long: **evidence trumps opinion**. They wrote a whole section of the book about testing what goes in the boxes, and that's the part that fell off the truck somewhere between the book and your offsite. So here's the version I run when the calendar says a research project isn't happening. Four passes over the same page, in order, each one filling boxes from something a buyer actually said. In my own work I start wider, with the market and the category, and then I go mining calls, which is my favorite part. The four passes below work on all of it, and on a lot less. **What you need:** 10 to 15 pieces of evidence from the segment you care about, in whatever mix you have: won and lost calls, customer interviews, reviews, case studies, win/loss notes. Any market or category research you've got goes on top. **What you end up with:** a canvas where every box traces to a source, one ranked list of underserved outcomes, and a call on whether your problem this quarter is a message problem or something no message can fix. ### Pass 1: name what they'd use if you didn't exist April Dunford's positioning framework starts from one question: if you disappeared tomorrow, what would your customers use instead? It goes first here because every score later on is graded against it. I've also gotten it wrong. At a question-and-answer ad platform we were sure the fight was Google and Facebook, and advertisers were comparing us to LinkedIn and Reddit. I write the answers across the top of the page, above both halves, before I fill in a single box. Everything underneath has to answer to that list. I fill it from market and category research, plus what buyers say they're using now. For most products the honest list includes something that isn't a product at all: a spreadsheet, a whiteboard, somebody on the team doing it by hand. Every rating you make later is a comparison against this list. A buyer's satisfaction is always relative to whatever they're doing right now. Grade satisfaction without naming the alternative and you've measured your product against your own roadmap. **Where this goes wrong:** you write down three vendors from your win/loss dropdown and move on. When the list comes back all competitors, I treat the pass as unfinished, because most of the deals I've lost went to the status quo rather than another vendor. On a call, I just ask what they're doing now, or what they were doing before. That tells me their status quo. It also tells me why now, which is the critical event, and that one's super important. **Output of pass 1:** the alternative set, written where you can see it while you do everything else. ### Pass 2: write the jobs as outcomes you could argue about This pass does the most work and asks for the most discipline. Ulwick's format for a desired outcome is a direction, a metric, an object of control, and the context, written so it stays true whether or not your product exists. Two of his own, both about listening to music: minimize the time it takes to get the songs in the desired order for listening. Minimize the likelihood that the music sounds distorted at high volume. There's no product in either one, no feature, and no adjective smuggling in the value. Each one names something the person is trying to move, in a direction, that you could put a number on. Here's a made-up example, close enough to a real week that you can map your own product onto it. Say you sell scheduling software to field-service teams, and your buyer is the dispatcher who builds tomorrow's routes. Out of their calls, reviews and interviews, you might write: - Minimize the time it takes to rebuild the day's route when a technician calls in sick. - Minimize the likelihood that a customer hears a different arrival window than the one the technician is working to. - Minimize the time it takes to prove to a client that the crew arrived when the invoice says. Now compare those to what usually lands in the jobs box. "Better route planning" is a feature with a compliment attached. "Save time" has no object, so nobody can tell you whether it happened. "Make data-driven decisions" is your mission statement, copied into their box. Ulwick's own rules for these statements: measurable, controllable, actionable, devoid of solutions, and stable over time. Calls and customer interviews are the best source for this pass. Work through them and write five to ten of these. Use the buyer's nouns, not your category's. If a line only makes sense to somebody who already knows your product, it belongs in the value map on the other side of the page. Outcome statements are how you write a workflow down, which is what that whiteboard was missing. I hold myself to one rule in this pass: if I can't say which source a sentence came from, it doesn't go on the page. **Where this goes wrong:** you write the outcome your product is best at. The test is whether the sentence would still be true if your company folded tomorrow. Buyers will hand you feature requests all day, and they sound like outcomes. When someone names a button or an integration, ask what it would let them do, and keep asking until the answer has no product in it. **Output of pass 2:** five to ten outcome statements in the jobs box, each traceable to a source. ### Pass 3: grade each outcome twice Two scores per outcome. High, medium, low is enough, and a 1 to 10 scale is false precision at this sample size. **Importance:** how much does this buyer care about this outcome? **Satisfaction:** how well does the alternative from pass 1 already deliver it? I do this with the evidence open beside the page, and I still catch myself rounding satisfaction down on the outcomes my product happens to be good at. Four combinations, and only one of them is money. - **High importance, low satisfaction.** That's the wedge. That's where positioning points. - **High importance, high satisfaction.** These are table stakes. You have to prove them and you can't win on them. - **Low importance, high satisfaction.** You're overspending on something the market stopped caring about. - **Low importance, low satisfaction.** Leave it alone. Reviews and any competitive or pricing research are the best sources for satisfaction, since a review is a buyer writing about what they're using now. At the ad platform, that research was a pricing analysis against LinkedIn and Reddit: what you get and what you don't. It put us in the middle of the funnel, better performing than Reddit, which was going after brand, and cheaper than LinkedIn, which was too expensive at the time. A good tool turns into a bad meeting right about here. You're building a ranked hypothesis from evidence you already have, which is a lot less than Ulwick's full version: a single job commonly turns up 50 to 150 outcome statements, ranked with statistically valid research. Label yours that way and it's the most useful document in the building. Leave the label off and you've handed your CEO a guess again, this time wearing a lab coat. **Where this goes wrong:** the room grades from memory again, and the scores drift toward whatever the product is good at. Grade with the source open, and write it beside the score. I'd rather carry five outcomes I can point to than thirty I can't. **Output of pass 3:** a ranked list of underserved outcomes, every score labeled as a hypothesis. ### Pass 4: put the four forces under pains There's one more problem with the customer side of the page, and better inputs won't fix it. Bob Moesta and Chris Spiek describe four forces on anybody deciding whether to switch: the push of a situation that broke, the pull of the new way, the anxiety of what happens if it doesn't work, and the habit of whatever everybody already knows how to do. Push and pull are what your marketing is built to create. Anxiety and habit are what kill the deal after your marketing worked perfectly, and the canvas has nowhere to put them except pains. So in my version, pains gets a strip underneath with all four forces on it. Moesta and Spiek built their model for switch interviews, and squeezing it into a strip loses something. I use the strip anyway, because anxiety written on the page is harder to ignore than a box labeled pains. Anxiety rarely makes it into a deck, because there's no campaign for "they're scared of you." It's easy to fill push and pull from your own messaging and leave anxiety and habit blank, because no buyer said them out loud. Those two show up as stalls and silence rather than objections, so go looking in the deals that went quiet. At a speech AI startup, anxiety and habit showed up as demo requests. Our product-led motion was built for developers, but when we moved up-market into established workloads like call centers, a PM had to get involved, and a PM didn't know what to do with an API key, so they booked a demo. On the call, the anxiety was security, scalability and go-live time. The habit was build versus buy. They also needed a business case that turned a better experience for their customers and internal teams into money saved or made. A signup funnel can only ever show any of that as a conversion rate. **Output of pass 4:** the four forces filled under pains, with the anxiety and habit lines quoted from actual deals. ### Read the result: two kinds of low score, pointing opposite ways You now have a ranked list of underserved outcomes. Before you touch a word of copy, split that list one more time, because a low satisfaction score has two completely different causes. **A capability gap.** Your product doesn't deliver that outcome well. The score is accurate. No amount of positioning fixes it, and anything you write gets found out in the first demo. That belongs in a roadmap conversation, or it changes who you should be selling to. **A perception gap.** Your product delivers it, and buyers don't know, don't believe you, or can't find the proof. Same low score, opposite meaning. You have a wedge sitting unused, and you can fix it this quarter, often with proof they can check rather than new copy. The test is one question per outcome: if a buyer used us for a month, would they rate this differently? Don't answer it yourself, since you're the person most likely to round in your own favor. Answer it from the won deals and your case studies. If the buyers who picked you name that outcome as a reason, it's perception. If even your customers rate it low, it's capability, and no campaign will survive contact with the trial. I had both kinds at once at that ad platform. Our win rate was falling because we were selling mobile app-install ads, and advertisers wanted thousands of installs at a price we couldn't get to. We could deliver a couple hundred. That's a capability gap, and the sales team already knew it. They didn't like chasing those deals anyway, because the close rate was low. This was pre-AI and pre-recorded calls, so we pulled the customer list out of Salesforce and enriched it ourselves: the couple hundred accounts that had stayed a while, spent real money and given us feedback. The pattern got clear pretty quick. They were high-growth SaaS companies in the Bay Area, a couple of years from an IPO, and their founders already wrote answers on the platform. Their marketing teams kept telling us they wanted to promote the founder's answer, and a budget holder who already knew the platform was the one who said okay, go for it. The perception gap showed up once we were talking to those companies. Everyone assumed our audience was Reddit's audience. Market research settled it. A comScore study showed half of it wasn't on Reddit at all, and that it skewed college-educated and high-income, which made it an audience worth going after. We took it to a few of those companies with our CRO doing the pitching, tested it with a few good reps on each coast, and built a target account list. The native ad unit became our fastest-growing ad unit and revenue driver. Then those customers asked for lead-gen ads and links back to their own sites, and we built those too. Whichever one you've got, the next conversation starts from evidence instead of a vote. Then do it again next quarter. Most of the docs I'm shown are out of date by the time I read them, because they got written once and filed. I treat mine as a living document and refresh it at least once a quarter, from what customers, the market and the competitors are telling us. That used to take weeks, if not months. Ten years ago you sat in on the calls, because the calls weren't recorded. Now we have the recordings, and most teams still aren't using them as much as they should. I have Claude with API access to Gong, and I love mining calls this way. We categorize every call by the job the buyer is trying to get done, their needs, their objections and their unmet needs. Then we grade the unmet needs the way Ulwick does, by how satisfied buyers are with what they use now, and map our competitors against the same needs. That's where the white space shows up, and it tells us how our messaging is landing through our value props. It's the same four passes, run across every call at once, and the mining takes about half a day. The hard part is knowing the right questions to ask, QA-ing what comes back and validating it before anyone acts on it. We also map the needs over time, so you can watch one bubble up and start to scale as more buyers bring it in. That's interesting from a marketing, product and market-trend angle: maybe a competitor launched something, and now people are asking you for it. Map the jobs, needs, pains and gains that way and you start to see where the demand spaces are, which deserves a post of its own. One caution before you take this into a meeting. This kind of research can hand you the ranked list of unmet needs, and it still can't get your organization aligned behind acting on it. That part stays yours. A graded canvas gets the room arguing about the right thing, and that's as far as any instrument can carry you. ### The Monday Version Pull your last ten lost deals from your best-fit segment, with whatever you have on each: a call, an interview, a review, the rep's notes. For each one, write the outcome that buyer was trying to reach as one sentence in the format (direction, metric, object, context) with no product names. Your first tries will read like features, so hold each one to the pass 2 test: would it still be true if your company folded tomorrow? Then ask whether your product already delivers that outcome well, and whether the buyer believed it. If you find an outcome your product already delivers and the buyer still didn't believe it, stop there. You've found a perception gap, and your positioning and messaging are pointing at the wrong thing this week. Label it a hypothesis before it goes anywhere near a deck. If you have nothing on the lost deals at all, start from the other end, the way we did at the ad platform. Pull the customer list, sort it by who's stayed and spent real money, and look for what those accounts have in common. Like Sherlock Holmes, look for patterns, connect the dots. It's a weaker source than a buyer's own words, so label it that way and grade it anyway. I built a free tool for this: [the Graded Canvas](https://jdprater.com/graded-canvas). Run the passes in your browser, or grab the Claude skill on the same page and paste in your notes, reviews and transcripts. It won't write a line it can't trace to a source. Opening Folders, JD --- ## Sources - Alexander Osterwalder, Yves Pigneur, Greg Bernarda and Alan Smith, *Value Proposition Design* (Wiley, 2014), and the Strategyzer VPD companion: the canvas, the three types of fit, and the ten testing principles. - Anthony W. Ulwick, *Jobs to Be Done: Theory to Practice* (Idea Bite Press, 2016): desired outcome statements, the construction rules, and the scale of a full ODI study. - Bob Moesta and Chris Spiek, the four forces of progress, jobstobedone.org. - April Dunford, *Obviously Awesome* (2019): competitive alternatives as the starting point. **How the pieces fit.** The parts already know each other: Osterwalder's foreword to Ulwick's book says outcome-driven innovation research led his team to design the canvas, a lineage I traced in [A Brief History of Jobs to Be Done](https://jdprater.substack.com/p/a-brief-history-of-jobs-to-be-done). Dunford credits Bob Moesta's jobs-to-be-done work as a big influence on how she does positioning, and joining her question to Ulwick's grading is my assembly. The four-force strip is my reduction of Moesta and Spiek's model, which they built for interviews rather than canvases.