用 Jev 给 B2B 冷邮件市场打分排序

Termsheetinator @termsheetinator原推发布于 2026.09.22

在 X 上打开原推

my business partner is not going to like this: we have a cold-email playbook that uses a specific site with ~6,779 B2B categories + over 1M case studies sitting on it lol. here it is kids: casestudies(.)com/categories this is basically an infinite demand gen playbook if u know how to read it. heres exactly how: 1. use the category page as the market map dont sit there scrolling around looking for the coolest software. youre looking for categories where: - 1 new customer could easily be worth $20K+ - companies can reasonably buy inside 3-4 weeks - the buyer is easy to identify + reach - vendors already have multiple public case studies - the result is repeatable across lots of similar companies - the demand gen work can be fulfilled without praying for miracles important: these are hypotheses at first. u do not know a company's ACV or sales cycle just because AI guessed from a category name. 2. turn the categories into a scoring queue have Codex or Claude structure the markets u want to inspect, then use Jev to score each one across the same closed set of questions: - likely ACV band - likely sales cycle band - size of the reachable buyer universe - how obvious the economic outcome is - how much public proof exists - how easy the delivery would be - how confident the overall fit is Jev gives u the order to research and index them. it doesnt turn a guess into a fact. manually validate the top markets before u build anything. 3. open a category, then open 1 company inside it go directly to the company's own website + case studies. use the directory for discovery. dont clone their database or mirror a million PDFs into some commercial warehouse. keep the source URL, follow the site's terms + warehouse YOUR research and scoring. find a company with a few strong case studies and pull: - customer name + domain - what type of business they are - size, location + operating model - the problem they had - the outcome they got - why they were a fit - who likely owned the buying decision - any trigger that made the timing make sense 4. pick the most repeatable case study not always the biggest logo. pick the customer where u can explain exactly why another X amount of companies would have the same problem + want the same outcome. that company becomes the seed. 5. run the seed domain through DiscoLike give it: - the case study customer's domain - a plain english description of the ICP - company size - geography - business model - industry - any tech, growth or operating signal that actually mattered in the case study pull a few dozen lookalikes, then review the list. similar does not automatically mean qualified. every company still needs a reason to be there. 6. send this email to the company that published the case study subject: {{firstName}} {{casestudy_companyname}}? “{{firstName}} - we took the {{caseStudyCompany}} case study from your site and mapped a few thousand companies that look extremely similar to them. same offers, sizes and signals that made {{caseStudyCompany}} a fit for {{caseStudyOutcome}}. want me to send over the first 20?” thats it. youre using proof THEY already trust to show them where the next version of that customer could come from. the part most people will fuck up comes after they reply yes. 7. do the work immediately run the final DiscoLike search through the app, API or MCP, then manually review 10-20 companies before sending anything. reply with the sample directly inside the email: - company - domain - 1 line on why it looks like the case study customer dont attach a giant mystery CSV full of shit u havent checked. give them the useful part right there, then add 2 real meeting times + your calendar. 8. understand what u optimized for that first email is optimized for a positive reply. it is NOT optimized for a booked call. so once they raise their hand, u need to be way more aggressive about converting the interest. if they dont book, call them the next day. “hey {{firstName}}, calling about the {{number}} companies I sent over from the {{caseStudyCompany}} case study. did u get a chance to look through them?” now the call isnt cold anymore. ur following up on completed work they asked u to send. 9. expand the sample into the actual engagement on the call, show them what this looks like across every useful case study they have: - use each successful customer as a seed - map the lookalike market around each one - combine + dedupe the full TAM - find the right decision makers - build the outbound campaigns - agree on the exact buy box - generate meetings - let the client approve which ones are genuinely worth pursuing now ur not selling a random lead list. ur selling a defined number of qualified, pursuable opportunities built around customers they already know they can help. and this is where the $20K+ engagement can make sense IF their customer economics, market size, sales cycle + ability to fulfil it support the price. 10. save the play so it compounds for every market, store: - category - company being targeted - case study URL - seed customer + domain - case study outcome - lookalike filters - estimated ACV band - estimated sales cycle band - Jev score + confidence - sample sent - positive reply - meeting booked - opportunity accepted - rejection reason after a while u dont have a folder of random lead magnets. u have an indexed bank of markets, proof, offers, seed customers, lookalike searches, copy + actual response data. the company already told the internet who they can help + what they did for them. ur job is to find more companies that look like that customer, hand them the first 20 + ask if they want u to build the machine that keeps doing it. hell of a lot easier to buy than “we do lead gen.”