Lead Scorer

The First 100 Customers Course #38: How groundcover Found Its First 50 Through Founder-Led Sales

A sourced course on groundcover's narrow first use case, founder-led outreach, early pricing and product-assisted follow-up, with a practical first-buyer worksheet.

By Lead Scorer 18 min read The First 100 Customers Course · Lesson #38

TL;DR

groundcover's first buyer bought a specific visibility use case from an eBPF sensor that had no dedicated user interface. Founder Shahar Azulay says he opened at $100,000 and accepted a three-year agreement at $10,000 per year. He later used personal relationships, about 10–15 personalized LinkedIn contacts on a good day, and product activity to build the early pipeline. In his SaaS Club interview, he associates roughly 50 customers with the first $1 million in annual recurring revenue. These are founder-reported milestones, not an audited funnel or proof that LinkedIn produced all 50.

The repeatable lesson is to sell a narrow, provable job before a replacement story. Find a buyer already paying to live with an unsolved edge case. Show the smallest result your product can deliver now. Price that scoped result explicitly. Then keep the founder close to outreach, installation, objections and the next promise. This course turns groundcover's path into a two-week first-buyer exercise for a technical B2B SaaS founder.

Paperboard course cover with the exact groundcover title in navy marker and a red underline
groundcover first sold a specific visibility gap, then used founder-led outreach and product signals to develop its early customer pipeline.

What the evidence proves, and what it does not

The interview gives unusually useful operating detail, but the numbers have different jobs. Azulay says the first customer arrived about three months after starting the company, through a connection to Lemonade's head of DevOps. There was no proprietary UI yet: the team offered its sensor and dashboards in Grafana for a few gaps the incumbent stack could not easily see. The $10,000 annual agreement was one first deal, not a typical price for every account.

He says network, personal outreach and later product-led conversations produced an initial group he describes as 50 to 100 customers. The host also asks whether about 50 customers got the company to $1 million ARR, and Azulay agrees approximately. Neither statement supplies a cohort table, source attribution, conversion denominator or exact date. A separate founder retrospective reports five customers and $32,000 ARR at the end of 2022. Those are different milestones, not two versions of the same count.

Independent checks support the setting, with narrower scope. A Product Hunt listing documents a September 2022 public launch. The current pricing page shows public per-host plans and a bring-your-own-cloud architecture. Neither proves the first contract, historical discount level or the number of buyers from LinkedIn. Keep those as attributed claims.

The mechanism: earn the right to widen the promise

A crowded technical market is a hard place for a new vendor to announce that it will replace the incumbent. Buyers have existing dashboards, alert rules, data flows and renewals. A broad replacement pitch asks them to take on migration risk before the startup has proved it can do one job. groundcover took a smaller opening. The first buyer already used Datadog, but wanted visibility into particular traffic and cloud issues. A sensor that exposed those facts was valuable even without a polished interface. The buyer did not need to believe that the whole observability stack could be replaced.

This creates a four-stage sequence: an observable gap inside an existing workflow; a bounded installation with a result the buyer can check; a paid agreement proportionate to that result; and repeated founder conversations that reveal which gaps recur. The sequence is not a claim that cheap pilots automatically lead to expansion. Azulay says early customers did not necessarily become much larger accounts. Treat the first contracts as learning and reference opportunities, with their own economics, rather than booking imaginary future ACV.

StageGroundcover evidenceYour output
GapMissing visibility in an incumbent environmentOne named, repeated job
ProofSensor plus Grafana, before a full UIA test with a pass condition
PurchaseFounder-reported $10,000 annual first contractA scoped paid offer
LearningFounder handled prospecting, demos and procurementObjection and activation register

Step 1: pick a buyer with an existing bill and a visible gap

Write down the current workaround before writing an outreach message. groundcover did not ask engineering teams whether monitoring sounded interesting. They already used monitoring tools. Its opening was a data gap created by the way those tools collected and priced telemetry. The buyer could recognize the gap inside a production system, and the technical team could test whether the sensor made that gap visible. That is more concrete than asking whether a buyer wants a new platform.

Choose one role with direct responsibility for the problem. Azulay concentrated on heads of DevOps. For another SaaS, the role might be a revenue operations lead reconciling billing data or a support lead losing handoffs between channels. Record the incumbent tool, the expensive workaround, the evidence the buyer would accept, and the person who can authorize a small paid test. If you cannot name those four fields for a prospect, the account is not ready for this motion.

Worksheet: “For [buyer role] at [company type], [current system] fails to show or complete [specific job]. Today they [workaround], which costs [time, money or risk]. In [time window], our product can demonstrate [observable result] without asking them to replace [incumbent].” Leave the cost field blank when you have no buyer evidence; never fill it with a benchmark from another company. The pass condition is three qualified buyers who independently describe the same job and agree that the proposed result would matter. If the job changes in every conversation, revise the wedge before increasing volume.

Step 2: design proof small enough to survive the first installation

The first groundcover sale did not depend on a complete interface. It did depend on an installation that could expose useful telemetry. A technical founder may hear “sell before the product is finished” and skip the second half: the buyer still needs a working result. Make the proof narrow enough that your team can support it. Define the environment, data permission, setup owner, first successful output and reversal plan. If the buyer needs a security review, treat it as part of the offer, not an annoying surprise after the verbal yes.

Ask the buyer to specify the acceptance test in their words. For a monitoring product it might be “show this class of network event in our test environment.” For an AI workflow it might be “process this approved set of documents and identify every uncertain answer for human review.” Record a deadline and a named evaluator. If you cannot show the result safely in the agreed scope, stop the sale and fix the product. A relationship can open a door; it cannot make an unproved deployment reliable.

Step 3: price the first result without inventing a future account

Azulay's first price conversation is memorable because the starting ask and signed annual amount were far apart. He says the CFO offered three years at $10,000 per year and the founders accepted. He later published a per-host list price and discounted it heavily while learning to sell. Today's public plans still price by monitored host, but current numbers should not be treated as the 2022 terms.

“Close at any price” is his opinion about early learning. A founder should make the cost of that learning explicit. Before offering a discount, calculate onboarding hours, recurring support, infrastructure cost, payment terms and the maximum number of similar pilots the team can handle. Write what the buyer receives, what they do not receive, and when the price will be reviewed. The signed amount is a payment signal; an open-ended custom project with no boundary is not the same validation.

A practical decision rule: accept an early deal below the desired list price only when it buys a specific learning outcome, covers the planned delivery cost or has an approved learning budget, and gives permission to measure the result. Stop when the buyer asks for a different product, indefinite service work or promises your team cannot keep. Log the price and the reason for it; do not quietly turn the discount into a standard quote for every new account.

Step 4: use founder access as a research instrument

Azulay says he sent personalized LinkedIn approaches to roughly 10–15 people on a good day, often revisiting people met at conferences or through his network. That is an activity range, not a conversion rate. He could also change the product conversation during a sales call, decide which technical request mattered, and bring the engineering team into a proof of concept. An early SDR would not have had that context or authority. A remote SDR hire without a local team and clear process failed, by his account.

Build a small account register. For each prospect record why this person owns the gap, how the gap was verified, the source of the introduction, the message sent, the next step, and whether the buyer saw a working proof. Send a message that names one observed problem and asks for one specific conversation. Avoid a generic “we replace your stack” claim. On the call, separate discovery from a commercial ask: “Is this gap real, who owns it, and what proof would justify a paid test?” If the answer is only friendly feedback, record feedback, not pipeline.

The first weekly pass condition is ten relevant accounts, five conversations and at least two buyers who define the same test. These are suggested sprint gates for your team, not groundcover's historical ratios. If the accounts are relevant but nobody wants the test, change the wedge or the proof. If people want it but cannot buy, inspect procurement and scope before sending more messages.

Step 5: follow product activity with a useful offer

groundcover layered a product-led motion onto founder prospecting. Azulay describes people reaching the product through public exposure such as Product Hunt, then the team seeing a login or installation and offering a larger demo. The public launch listing supports that the product was exposed in 2022; it does not tell us how many buyers the launch produced. The useful operation is the handoff from a visible attempt to a human conversation, not a claim that every signup needs a sales call.

Define one consent-respecting activation event that indicates a buyer tried the job you sell. A signup alone is weak. A configured workspace, first data import or completed test is stronger. Decide what help is genuinely useful at that point and send a short, reviewed message offering it. If the buyer never reaches the event, improve onboarding before scaling acquisition. If they do reach it but decline a conversation, ask whether the product already solves the job or whether the offered help is mistimed.

Run the motion with Lead Scorer, with the founder in charge

Lead Scorer is useful here as a research and drafting surface for a narrow list. It cannot install an eBPF sensor, certify value, decide a discount or close a customer. Connect its MCP and skills library in Codex or Claude, then begin with the icp-offer-context and icp-scoring-rubric skills. Define one buyer role, one current workaround, disqualifiers and the evidence a prospect must show before contact discovery. Keep two lists: ten warm or event-connected accounts, and ten cold accounts with a public sign of the same problem. Do not merge them, because you need to learn whether the relationship changed the outcome.

A copyable first prompt is: “For a B2B product that solves [one job] for [role], find ten companies with a dated, public signal of [problem]. Record the signal URL, relevant person, incumbent or workaround, and a reason to exclude. Score them before any paid enrichment. Keep warm connections in a separate list. Do not invent contact details.” Review the output yourself. A pass is a short list where every keeper has a real company, a role tied to the job and a source you can inspect. Stop if the agent uses generic firmographics as proof of pain.

Next use signal-research-dossier on keepers and contact-discovery only after your score and credit gate. The dossier should hold at least two dated signals when available, source URLs, an uncertainty note and one plausible opening. Ask the MCP to enrich only approved keepers, then use cold-email-first-touch or linkedin-connection-requests to draft one message per person. A second prompt: “Draft a concise first touch for these approved contacts. Name the verified gap and ask whether a scoped test of [observable result] is useful. Do not claim we replace their incumbent. Mark missing evidence for human review.” Run outreach-qa-audit before any campaign draft proceeds.

The MCP can create a draft campaign and keep every message for human approval. No agent should activate a campaign, spend beyond a credit threshold or send a message without the operator's review. Log replies in five useful buckets: no gap, gap but no urgency, wants a test, procurement obstacle, and unclear. Use reply-triage to prepare a response, then have the founder decide what to send. Capture recurring objections in Content Studio. Publish useful answers to the exact problem, and use signal-audiences to qualify people who engage with that content before adding them to outreach. The loop makes each conversation improve the next small list.

A third prompt closes the sprint: “Compare the warm and cold lists. Show source-verified accounts, conversations, agreed tests, installations, paid contracts, support hours and objections by list. Keep unknown values blank. Recommend continue, revise or stop.” A pass is a decision grounded in buyer behavior, not activity volume. If no buyer will define a test, do more interviews. If tests work but contracts stall, fix offer and buying process. If support consumes the team's capacity, narrow the scope before sending the next batch.

A fourteen-day execution plan

  1. Days 1–2: define one buyer role, one gap, one test and a written pilot cost ceiling.
  2. Days 3–4: build two separate ten-account lists; verify every source and reject weak matches.
  3. Days 5–8: contact the approved people personally; record conversations separately from polite replies.
  4. Days 9–11: run at most two bounded tests with a named evaluator and a written pass condition.
  5. Days 12–14: make a paid offer where proof passed, then compare buyer behavior, support cost and objections.

A small number of strong conversations is enough to decide whether to repeat the wedge. Do not project a first-50 curve from two pilots. The stop condition is a repeated mismatch: buyers cannot confirm the gap, the product cannot prove the result safely, or the only way to sell is to become a different service business. Revise one of those before increasing reach.

Checklist and sources

  • One named buyer and current workaround, with source evidence.
  • A scoped technical test, evaluator, deadline and reversal plan.
  • A paid offer with a known delivery cost and discount reason.
  • Warm and cold lists kept separate, with every response categorized honestly.
  • Activation evidence followed by genuinely useful human help.
  • A continue, revise or stop decision after two weeks.

Primary account: The SaaS Podcast full interview and transcript. Cross-checks: Azulay's dated retrospective, Product Hunt's 2022 listing, and groundcover's current pricing. The first-contract value, outreach volume and first-50 attribution come from Azulay's recollection. No public contract, channel conversion report or cohort table was found.

Frequently asked questions

Did LinkedIn alone bring groundcover its first 50 customers?

No. Founder Shahar Azulay attributes the early pipeline to personal relationships, personalized LinkedIn and email outreach, and later product activity. The interview gives no channel-level conversion count.

Was groundcover's first contract $10,000 total?

Azulay describes a three-year commitment at $10,000 per year after opening at $100,000. The contract is not independently available, so this is a founder-reported figure.

Should a SaaS founder accept any price for the first customer?

That is Azulay's stated preference, not a universal rule. Define a paid, scoped use case and a learning budget, including support cost and a stop condition, before offering an early discount.

Can Lead Scorer get my first 50 customers automatically?

No. It can help build and qualify a narrow account list, enrich selected contacts, draft reviewed outreach, and capture response signals. The founder still owns the product proof, offer, approvals and sales conversation.

Keep reading