A B2B SaaS founder’s guide to the first 100 customers via signal mining
A weekly cadence for finding companies actively complaining about the problem you solve — without a paid acquisition budget. The exact subreddits, X searches, G2 / Capterra review triggers, and job-post signals we mine, with templates.
By Bora Esen
The hardest part of running a new B2B SaaS is finding the first hundred customers without a paid acquisition budget. Pre-PMF you cannot afford performance marketing. You also cannot wait six months for SEO to compound. The fastest path is signal mining — finding companies that are actively complaining about the problem you solve, then reaching out with a cite-led, low-pressure message. This guide is the weekly cadence we use at Leafer and the cadence we recommend to every pre-series-A founder we talk to.
The weekly cadence
Two hours every Monday, two hours every Thursday. Four hours total per week. Anyone can run this; the founder is usually the right person because the founder picks up nuance the SDR will not.
Monday is “source.” Pull the new signals from the week — Reddit posts, X complaints, job posts with relevant tech-stack language, G2 / Capterra negative reviews of competitors. Filter for ICP. Score for intent.
Thursday is “send.” Approve every draft in the queue, send by EOD, capture replies in the CRM. Tuesday and Wednesday in between let prospects breathe and let replies trickle in from the previous week.
1. Reddit: the four-question filter
Reddit is the easiest signal source because the data is structured (subreddit, score, comment count) and the language is unfiltered. Build the universe — five to fifteen subreddits where your buyer hangs out — and run a daily scrape.
For each post, ask the filter four questions: Does the post describe a problem your product solves? Is the poster the kind of person who can buy your product (founder, ops lead, engineering manager)? Is the sub a place where commenting is welcome (most are; a few hate vendors)? Has anyone already replied with the answer? Filter at three yeses and post a public comment instead of a DM.
2. X: saved searches + the cite rule
X is messier than Reddit. The way to make it usable is with five to fifteen saved searches, each carefully worded.
Patterns that work. “anyone using [competitor] for [use case].” “looking for [category] that does [feature].” “our [tool] just broke again.” “why is [legacy tool] so [complaint].”
For every match: cite the tweet, write a one-line response, post publicly. DMing the user without a public cite first is the fastest way to get muted or reported. The cite-first pattern feels like a peer reaching out, which is what you are.
3. G2 and Capterra: the three-star reviews
Software review sites are an under-rated signal source. Filter for three-star reviews of your competitors — these are users who almost like the product but have a specific complaint. The complaint is your wedge.
Reach out via LinkedIn (the reviewer is usually identifiable from their job title in the review), cite the review verbatim, offer a three-minute demo. This is about as warm as cold outreach gets: the reviewer has already written down what is wrong with the product they currently pay for, in public, under their own name.
4. Job posts: the tech-stack signal
Job posts are the most underused signal in B2B. Companies hiring a “Senior Engineer with Postgres, Redis, BullMQ” experience are companies running on Postgres, Redis, and BullMQ. If you sell to that stack, the job post is a buying signal in disguise.
Build a daily LinkedIn Jobs scrape on your relevant keywords. Sort by company size to filter for ICP. Reach the hiring manager or VP Eng via cold email — “saw you are hiring a Postgres-fluent engineer, we thought [tool] might shave a week off the onboarding.”
5. Three message shapes that work pre-PMF
Three patterns, in descending order of how much they ask of the reader. We publish no response figures against them — Leafer does not send, so we have no delivery data, and a number we made up would be worth less than the shapes themselves.
The cite. “Saw your post about [problem]. We built [tool] specifically for that case. Three minutes if you want to see what we did.” Strongest of the three, because the first clause proves you read something before you wrote.
The peer. “Hey — fellow founder. We hit the same wall last year and ended up building the way out. Happy to share what we learned even if you do not use the tool.” Weaker signal, wider applicability: use it when you have ICP fit but no specific post to quote.
The artefact. “We wrote up the approach in a guide [link]. Worth a skim if [problem] is on your roadmap.” The lowest response rate of the three by construction — it asks for nothing — but the people who do come back have read the thing and arrive pre-qualified.
What 100 customers actually looks like
Model it yourself before you start, with your own conversion assumptions, and you will find the whole thing turns on cadence rather than on any single-week result. Four focused hours a week, sustained for a year, is what gets a pre-PMF B2B company to a hundred customers without paid acquisition. A brilliant week in month two changes nothing if month three is empty.
The hard part is keeping the cadence. The cadence beats every clever tactic because compounding outbound is the most reliable way out of zero-to-one for B2B SaaS. Leafer is the platform we built so we did not have to rebuild the cadence every week. The cadence is the point.
