The First 100 Customers Course #23: How LinkSquares Turned Nearly 100 Buyer Interviews Into Its First 10 Customers

TL;DR

LinkSquares did not use cold email as a sales channel first. Vishal Sunak says the founders used one licence of an outreach tool and a four-email sequence to spend roughly nine months trying to interview nearly 100 general counsels. They built a convincing Rails prototype with no working backend, listened for recurring post-signature contract pain, and delayed a larger product bet until the language repeated.

The same narrow motion later produced the first 10 customers: one referral deal priced at $1,000 per month, followed by nine buyers the founders describe as complete strangers reached through cold email. The useful mechanism has two proof gates. Repeated buyer evidence earns the right to build. Stranger revenue earns the right to call the motion repeatable. The warning is that replies, famous logos, and friendly purchases can all feel like traction before they are.

What you will build

You will build an interview-to-customer system with seven outputs: a narrow buyer/problem memo, a four-touch learning sequence, a structured interview ledger, a demoable wedge, one paid-proof offer, a stranger-revenue checkpoint, and a channel expansion ledger. Every output has a pass condition and a stop condition so activity cannot hide weak evidence.

This motion fits B2B software where the buyer is identifiable, the problem creates a measurable cost or risk, and a founder can reach the role directly. It is especially useful for a technical founder who can build quickly but does not yet know which workflow deserves a product. It does not fit a mass consumer product, a product with instant self-serve adoption, or a market where purchase behavior cannot be observed through direct conversations.

Case snapshot

Element	What the sources support	Evidence limit
Founding problem	Thousands of executed agreements were difficult to review during an acquisition	Sources disagree on whether the relevant population was about 2,000 or 6,000+
Narrow buyer	General counsel at a company larger than the founders' previous employer	The founders initially knew none personally
Discovery	Nearly 100 conversations over roughly nine months	A later founder article says hundreds over a year
Prototype	Demoable Rails interface built in about 100 days without an operational backend	Retrospective founder estimate
First response burst	Five or six recognizable tech-company GCs replied in week one	None had become customers by the 2021 interview
First paid customer	A referral to a CFO with an active review problem at $12,000 per year	Identity, contract date, outcome, and renewal are not public
Next nine	Net-new strangers reached through cold email	Founder-reported; outreach denominator is unknown
First-ten timing	Around early 2017	No independent customer-order audit

The source language needs care. In the complete SaaS Podcast interview, Sunak describes the first buyer as a referral. The host then confirms that the other nine came from cold email, and Sunak emphasizes that they were strangers. A later founder interview speaks about 10 unaffiliated companies as the desired milestone. The public evidence does not justify rewriting the first referral as unaffiliated, so this course keeps the arithmetic explicit: one referral plus nine strangers.

The model: two proofs, not one

The causal chain is narrow buyer → problem-led learning email → structured interview → repeated language → demoable wedge → paid proof → stranger replication → channel expansion. It contains two different thresholds that early founders often collapse into one.

The problem proof answers whether the workflow repeats across independent accounts strongly enough to deserve product capital. The distribution proof answers whether buyers outside the founder's relationship graph will pay for substantially the same outcome. A founder can have one without the other. Ten interviews may expose a real pain without revealing a viable acquisition route. One friendly customer may pay without proving a stranger will even take the meeting.

LinkSquares began with genuine founder-market evidence. During Backupify's acquisition, the team could not quickly determine what thousands of executed customer contracts allowed. That was a painful event, not a market. Mentors pointed Sunak and Chris Combs toward general counsels, a role neither founder knew. Cold email connected the observed problem to the person who owned it. The interviews then revealed a narrow gap: existing contract-management products focused on work before signature, while legal teams still lacked searchable insight into agreements they had already signed.

Step 1: name one buyer and one expensive moment

Do not begin with an industry or a department. “Legal teams” is not narrow enough. Choose a person who owns a consequence and an event that makes the consequence urgent. LinkSquares chose the general counsel. The expensive moment was an acquisition, audit, regulatory change, or internal review that forced the company to answer what it had already promised across scattered contracts.

Complete this worksheet before sourcing names:

• Buyer: the role with budget or operational authority.
• Trigger: the dated event that makes the problem urgent.
• Current workaround: the documents, people, or tools used today.
• Consequence: the time, money, risk, or delay created.
• Why now: the change that makes an old workaround unacceptable.
• Exclusions: 20 account or role conditions that do not fit.

Pass condition: five target accounts contain the same role, trigger, and costly job. Stop condition: each account needs a different buyer or consequence. In that case, return to problem selection before building a list.

Step 2: write a learning sequence before a sales sequence

LinkSquares used one ToutApp licence and a four-email sequence. The public transcript describes the first message rather than all four. Sunak introduced himself, named analogous operating experience at a venture-backed SaaS company, described the difficulty of understanding executed agreements, and asked whether the problem sounded familiar enough for a 20-minute conversation. He was not pretending to have a finished product.

I handled [specific workflow] at [credible context]. When [trigger] happened, we could not [important job] because [current failure]. Does that happen in your team? I am interviewing [role] for 20 minutes before deciding what to build.

Send the sequence to 40 verified people in one role. Use touch two to add one concrete failure mode, touch three to ask for the right owner if the recipient is wrong, and touch four to close the loop without manufacturing urgency. Tag every outcome: delivered, reply, accepted call, wrong person, no pain, repeated pain, or active trigger.

Pass condition: at least two accepted conversations from 40 clean delivered contacts plus one useful wrong-person referral. Stop condition: zero calls after 40. Change the buyer/problem sentence before sending more volume.

Step 3: turn interviews into comparable evidence

Five or six impressive general counsels reportedly replied during the first week. The founders celebrated. None had become customers by the 2021 interview. That is the most useful failure in the story: a reply validates the relevance of a sentence, while a purchase requires urgency, authority, trust, budget, and a usable outcome.

Use the same seven fields for every interview:

1. The last time the problem occurred, with a date or concrete event.
2. The workaround used and the people involved.
3. The measurable consequence or risk.
4. The buyer's own language for the problem.
5. The current tool and existing budget.
6. The next trigger that could create urgency.
7. The other roles needed to approve or implement a purchase.

Ask for the last real event instead of asking whether someone would use your imagined product. After each set of five calls, group the evidence by trigger, workaround, and consequence. Do not average away contradictions. A disagreement about buyer, urgency, or implementation may signal two different markets.

Pass condition: the same workflow and costly consequence appear independently in five accounts. Stop condition: respondents like the concept but cannot name a recent event, or every conversation points to a different job.

Step 4: build only the demoable wedge

Sunak estimates that he built the first Rails interface in about 100 days. It had real pages, buttons, dropdowns, and tables, but it did not have an operational backend. The point was not to deceive buyers. It was to make the workflow concrete enough that a general counsel could react to the proposed job instead of abstract wireframes.

The repeated interviews narrowed the wedge to post-signature search and analytics: a source of truth for executed agreements, including scanned PDFs, and structured answers about dates, renewals, termination, liability, and other terms. Independent VentureBeat reporting in 2020 later corroborated that searchable repository and extraction product, but it does not verify the early development dates.

Your proof can be clickable, concierge, spreadsheet-backed, or partly manual. It should perform one end-to-end job and expose the riskiest assumption. It should not simulate an entire future platform.

Pass condition: a buyer can restate the value, supplies realistic input, and asks to test, use, or pay for the next version. Stop condition: feedback stays at colors, layout, and disconnected feature requests.

Step 5: price a live problem

The first LinkSquares buyer came through a former CEO who introduced the founders to a CFO. The prospect had an active contract-review exercise and immediately recognized the pain. When asked for the price, the founders named $1,000 per month, or $12,000 per year. Public sources do not identify the company or reveal implementation, outcomes, renewal, or discounts.

Create a paid-proof memo with:

• one urgent workflow and one decision owner;
• the implementation input the buyer must provide;
• an outcome date and a price;
• the operator time required from the customer;
• the measurement method;
• a written list of what the founder will not build.

Pass condition: signed money plus operational access to perform the job. Stop condition: praise, an introduction, a letter of intent without a trigger, or a free trial with no committed operator.

Step 6: run the stranger test

A referral deal proves that the problem can be worth money. It does not prove that the company owns a route to market. LinkSquares therefore kept going until nine net-new people with no prior obligation bought through the cold-email motion. Sunak describes reaching the first-ten milestone around early 2017.

Keep three early-revenue lists separate:

• Relationship proof: friends, former colleagues, investors, and referrals.
• Stranger proof: buyers with no relationship who buy the same job.
• Bespoke proof: customers who require a private product branch or new ICP.

Relationship advantages are useful. Hiding them is not. Use a warm introduction to learn how a buyer purchases, prices, and implements. Then preserve the test by finding whether strangers respond to the same problem and buy a substantially similar product.

Pass condition: net-new buyers purchase the same job through the same buyer and product wedge. Stop condition: every win needs a favour, a different ICP, or custom software that cannot serve the next account.

Step 7: expand channels only after the loop repeats

Cold outbound reportedly remained effective for LinkSquares in 2021. The founder also named customer referrals, review platforms, SEO, and trade shows as later channels. His founder-authored Inc. account says social content was a weak fit for general counsels, while a later interview warns that paid advertising can consume scarce early cash before a company has enough bankroll.

Channel	Early job	When to test	Stop signal
Cold email	Learn and reach a narrow role	Before product commitment	Forty clean contacts produce no calls
Referrals	Reach a live problem with trust	For first paid proof	Introductions create praise but no access or money
Events	Meet a concentrated buyer group	After the buyer/problem pair is clear	Badge scans do not become qualified conversations
Reviews and SEO	Capture buyers already researching the category	After customers can describe outcomes	Traffic lacks the target role or active trigger
Paid acquisition	Scale a known conversion path	After payback can be measured	Qualified opportunity cost exceeds the sales model

Pass condition: a second channel produces the same buyer, problem, and qualified opportunity. Stop condition: reach grows while qualified conversations and paid proofs do not.

What failed, and what not to copy

First, enthusiastic replies did not become customers. Treat discovery engagement and buying intent as separate states. Second, the founders spent about nine months gathering evidence because they feared wasting limited savings. That reduced product risk but created a large time cost. An AI builder should copy the thresholds, not the duration. Build the smallest proof after five independent accounts repeat the same expensive workflow, then return to calls.

Third, the company had unusually strong founder-market context. Sunak and Combs had lived the contract problem inside a SaaS acquisition. Do not manufacture analogous authority. If you lack direct experience, earn specificity through interviews and transparent prototypes. Fourth, later scale does not prove cold email caused everything. Independent TechCrunch reporting corroborates the origin, product, funding, and later customer base, but the first-ten attribution remains a founder account.

Your 7-day implementation plan

1. Day 1: write one buyer/trigger/problem pair and 20 exclusions. Output: a one-page ICP memo. Do not source leads until it names a role and expensive event.
2. Day 2: source 40 accounts and one operator per account. Output: a clean discovery list with evidence for role and trigger.
3. Day 3: draft four learning touches. Output: an approved sequence with no product claim the evidence cannot support.
4. Days 4 and 5: conduct calls and complete the same evidence fields immediately. Output: five comparable interview records, or a written reason the hypothesis failed.
5. Day 6: group recurring triggers, workarounds, consequences, and language. Output: a one-page pattern map and one wedge decision.
6. Day 7: build or storyboard the smallest end-to-end proof. Output: a test script with buyer input, pass condition, and stop condition.

If the week produces no calls, do not build. If calls produce no repeated recent pain, change the problem. If pain repeats but nobody commits input to a test, narrow the outcome. If a friendly buyer pays, celebrate it and begin the separate stranger test.

Lead Scorer implementation

This motion maps cleanly to Lead Scorer because it depends on narrow sourcing, evidence-first qualification, expensive contact data used only for keepers, reviewed outreach, and a reply loop. The product cannot choose the right problem or replace founder judgment. It can make each gate visible enough that volume does not disguise weak targeting.

Phase 1: separate discovery from sales

Start with the `icp-offer-context` skill. Store one buyer, one trigger, the costly consequence, allowed proof, and explicit disqualifiers. Then run `daily-vertical-prospecting` for a 40-person discovery list. Do not put these people into a sales campaign. The output is a learning cohort with source evidence, not pipeline.

Define an ICP for [product]. Buyer: [role]. Trigger: [event]. Pain: [consequence]. Source 40 discovery candidates in [market]. Exclude [conditions]. Keep this list separate from future sales prospects. Do not find paid contact data or activate outreach.

Apply `icp-scoring-rubric` with a minimum score of 7/10. Require evidence for role, company fit, and trigger. Only then invoke `contact-discovery`. That is the credit gate: show the qualified count and expected cost, then require human confirmation before consuming paid data.

Phase 2: draft and approve the learning sequence

Use `cold-email-first-touch` to draft the 20-minute research ask from verified signals. Add three follow-ups only after the first message is honest. Run `outreach-qa-audit` over the full batch. Reject invented personalization, unverified outcomes, a sales CTA, or more than one ask. A human reviews every message, sender configuration, and deliverability result before activation.

Pass condition: every draft identifies one real reason this person owns the workflow and asks only for a learning call. Stop condition: the message depends on generic praise, an inferred trigger, or a product benefit that has not been proven.

Phase 3: promote evidence, not contacts

Store each interview as a dated signal with the same fields used in the worksheet. When five independent accounts repeat the job, create a new sales list. Do not automatically promote the entire discovery cohort. Re-score for active trigger, authority, willingness to test, and fit with the demoable wedge.

From these interview records, identify accounts with the same trigger, workaround, and costly consequence. Explain every score. Move only 8/10 or higher accounts into a new paid-proof list. Draft nothing until I approve the evidence and price hypothesis.

After approval, use `follow-up-sequence` to create the customer motion and keep every message in draft. Route responses through `reply-triage`. Convert repeated objections into source-backed briefs with `daily-topic-briefs`; useful public answers can later create a new signal audience, but that audience remains separate from the original cold cohort. No agent activates a campaign, spends beyond the approved contact-data gate, or sends without review.

Saveable checklist

• One buyer owns one costly moment.
• Forty clean contacts test the positioning sentence.
• Interviews capture recent events, not future opinions.
• Five independent accounts repeat the same workflow.
• The proof completes one job from input to outcome.
• The first paid deal includes money and operational access.
• Referral, stranger, and bespoke revenue remain separate.
• Custom branches do not count as repeatability.
• A second channel waits until the first loop repeats.
• Paid-data and send actions keep human approval gates.

Sources and evidence limits

The full SaaS Podcast episode with Vishal Sunak is the primary source for the early sequence, four-email workflow, nine-month timeline, $12,000 first price, first-ten composition, and later channels. Sunak's Inc. article supports the general-counsel focus and the constraints on social and paid acquisition. The company's Series A announcement and independent VentureBeat report corroborate the later product and announced round. TechCrunch independently corroborates the acquisition origin and later company scale.

The first customer, next nine customers, first-week responses, interview count, and acquisition attribution are not independently audited. Public evidence provides no outbound volume, deliverability, response rate, sales-cycle distribution, early CAC, gross margin, activation, retention, or customer-level outcome. Source wording also differs: the detailed interview says nearly 100 conversations over about nine months, while a later founder article says hundreds over a year. This course uses the narrower, more specific account and discloses the disagreement.

The decision to copy is not “interview 100 people.” It is to define the evidence that earns the next expense. Ask one buyer about the last costly event. Record the same fields every time. Build only after the pattern repeats, then keep selling until strangers prove the route is yours.