A go-to-market strategy for startups starts with one channel

A startup with no budget and no sales team does not need a miniature marketing department. It needs one founder to find a narrow group of buyers, contact them personally, learn why they act, and turn those lessons into a sales motion another person could eventually follow. For a B2B product or another considered purchase, I would choose founder-run outbound by individual email and ignore every other acquisition channel for the first six weeks.

That choice is deliberately restrictive. Content takes time to compound, ads charge you for learning, events scatter attention, and partnerships hide the customer conversation behind someone else. Direct email gives you the shortest path to a specific buyer and a visible reply. It also exposes weak positioning quickly. If thirty carefully chosen people ignore a message, you have learned something useful about the segment, the trigger, or the claim.

This plan does not fit a low-priced consumer app that depends on habit, entertainment, or network effects. It fits a founder who can name a business problem, identify who feels it, and ask that person for a conversation. The work is sales, even if your title says product, engineering, or CEO.

Your strategy is a set of exclusions

A useful go-to-market strategy says who you will pursue, what urgent problem you will lead with, how you will reach the buyer, and what evidence will make you keep going. It also says what you will refuse to do during the test. A list containing outbound, partnerships, community, content, events, and social media is not a strategy. It is a list of unfinished jobs.

With no cash and no sales staff, your scarce resource is founder attention. Split it across six channels and none receives enough repetitions to teach you anything. One post performs well, a friend makes an introduction, and two cold emails get replies. You cannot tell whether the results came from the channel, the relationship, the message, or luck.

Set a six-week constraint: one segment, one painful use case, one buyer role, and founder-run email. Keep product work limited to problems that block a live opportunity or prevent a new customer from reaching the promised result. You can still answer inbound interest, but do not count it as proof that outbound works. Record it separately.

Paul Graham argues in Do Things That Don't Scale that founders usually have to recruit their first users manually. The important part is not the romance of handcrafting an experience. Manual recruitment lets the founder see the distance between what the company says and what a customer understands. I agree with the method, but I would add a hard condition: unscalable work must create a record. If the lesson stays in your head, you are accumulating anecdotes rather than building a motion.

Your six-week test should produce four things even if revenue remains small: a segment definition that predicts interest, language that earns replies, a consistent sequence from first contact to decision, and evidence about whether customers reach the promised outcome. Those are the raw materials of a go-to-market system.

A segment must be narrow enough to find

Choose a segment you can identify account by account without buying a database. Company size and industry alone rarely make a workable first segment. A target such as small healthcare companies leaves too much variation in regulation, workflow, budget, and buyer authority. You will write vague messages because the group has little in common.

A strong segment combines observable facts with a shared operating condition. Consider a hypothetical tool that reduces manual payroll reconciliation. US accounting firms with 10 to 40 employees are still broad. US accounting firms with 10 to 40 employees that run payroll for clients, advertise support for multiple payroll systems, and are hiring payroll specialists form a more useful group. You can find them through company sites and job listings, and the hiring signal suggests that workload matters now.

Define five fields before you collect names:

• The account type you can recognize from public information.
• The operating condition that makes the problem frequent or expensive.
• The person who owns the current process and can test a change.
• The event that makes the status quo harder to tolerate this month.
• The conditions that exclude an account even if it looks prestigious.

The exclusion field prevents wishful prospecting. If your product requires a dedicated operations owner, a five-person company may love the idea but lack anyone who can adopt it. If installation takes two weeks, a buyer seeking a fix by Friday is a poor first customer. Saying no to these accounts protects your reply and win rates from contacts who never had a fair chance of buying.

Do not choose the segment only because you know many people in it. Access helps you get interviews, but a convenient network can hide weak demand. Separate warm contacts from cold prospects in your records. If both groups describe the same trigger and advance through the same process, confidence rises. If only friends take meetings, you have evidence of goodwill, not a market.

The first segment is a testable wedge, not a permanent identity. Keep it until the evidence says the problem lacks urgency, the buyer cannot act, or customers fail after purchase. Do not broaden it because the first ten emails felt uncomfortable. Broadening before you understand the rejection removes the very pattern you need to see.

Customer interviews must uncover a buying event

Talk to prospects before you polish a pitch, but ask about a recent event rather than opinions about your concept. Steve Blank's customer development work treats startups as a search for a repeatable business model and insists that founders gather evidence outside the building. That advice is often diluted into chat with customers. Casual conversation is not enough. You need facts about behavior, consequences, authority, and timing.

Ask the person to reconstruct the last time the problem occurred. What triggered the work? Who noticed it? What did they do first? Which spreadsheet, service, or colleague did they use? How long did the process remain unresolved? Who approved spending, and what else competed for that money? Ask what happened after they chose the current workaround.

These questions distinguish pain from politeness. A prospect may agree that a problem sounds important while living with it for years. A buyer who recently changed a process, hired someone, paid a consultant, or escalated an error has shown effort. Effort is stronger evidence than enthusiasm because it carries a cost.

Do not ask whether someone would buy your product. The answer rewards imagination and kindness. Ask what they already tried, what they paid, why the attempt failed, and what deadline now applies. When you show the product, ask for a concrete next step: access to relevant data, a second meeting with the decision maker, a small paid engagement, or a clear refusal.

Steve Blank also warns that customer development is not a focus group that hands engineering a feature list. I have watched founders obey every request from early calls and end up with a product assembled from incompatible exceptions. Treat a request as evidence of a job the buyer cannot complete. Confirm the same job across the segment before changing the product, then decide whether the missing capability blocks the promised outcome or merely matches a preference.

Capture the buyer's exact nouns and verbs after each call. Your landing page may say workflow visibility while buyers say they cannot close payroll on Friday. Use the customer's concrete wording in outreach. Keep your internal category label for planning, not for forcing the buyer to translate.

Your message should name the costly change

A good outbound message proves that you chose the recipient for a reason and connects a visible change to a costly problem. It does not explain the whole product. The job of the first email is to earn a reply from the right person, not to finish the sale.

April Dunford starts positioning with competitive alternatives: what customers would do if your product did not exist. That is more useful than opening with a feature comparison against companies you consider competitors. Your toughest alternative may be a spreadsheet, an assistant, a weekly meeting, or accepting the loss. Name the old behavior and the consequence your product changes.

Use this message card as the source for every email:

SEGMENT: [observable account type plus operating condition]
TRIGGER: [event that makes the problem timely]
BUYER: [person who owns the current process]
ALTERNATIVE: [what that person does today]
COST: [time, money, delay, risk, or missed work created by it]
CLAIM: [specific outcome your product can credibly produce]
PROOF: [direct evidence you can honestly provide]
ASK: [small next commitment]
EXCLUSION: [who should not receive this message]

For the payroll example, the opening might note that the firm is hiring another payroll specialist while supporting several payroll systems. The problem sentence can ask whether cross-system reconciliation is creating review work before client payroll closes. The claim should describe the changed work, such as flagging mismatches before a specialist checks each client file. The ask might be a 20-minute call to compare the current process with a sample run.

Never invent proof. If you have no customer result, say what the product does and offer to work through the prospect's case. A founder who claims broad adoption before earning it damages the one advantage a small company has: direct, credible contact. Your first proof can be a working product, domain knowledge, a useful teardown of the current process, or a paid pilot with explicit boundaries.

Personalization should explain relevance, not display research. Mentioning a prospect's college, podcast appearance, and local weather proves that you can browse. Mentioning the hiring signal and its likely effect on the payroll close proves that you understand the work.

Founder-run email is the first channel

Founder-run email works here because one person can research, send, receive, and learn without paying for reach. It fails when a founder treats it as cheap advertising and sends generic volume. Thirty precise accounts teach more than a scraped list whose members share only an industry code.

Build the first account list by hand. Use company websites, public team pages, job postings, conference programs, and professional profiles to verify the segment fields. Record the evidence that qualified each account. If you cannot write a one-line reason for including it, leave it out. Research is part of segmentation, not administrative preparation.

Send from your normal company address as yourself. Keep the message short enough to read on a phone: a relevant observation, the likely problem, one credible claim, and one small ask. Do not attach a deck. Do not disguise a sales email as a request for advice. A direct note respects the buyer and gives you cleaner response data.

Follow up because busy people miss good messages, but do not manufacture urgency. A useful sequence has a first note, a follow-up that adds one piece of evidence, and a final note that closes the loop. Stop when the person declines or asks you to stop. Change one meaningful element per batch, such as the segment trigger or problem sentence, so you know what affected replies.

US commercial email rules still apply when the recipient is a business. The Federal Trade Commission's CAN-SPAM guide requires accurate sender information and subject lines, a valid postal address, a clear way to opt out, and fulfillment of opt-out requests within 10 business days. Other jurisdictions can require more, so check the rules for where recipients live before contacting them. Legal compliance sets the floor; relevance and restraint determine whether a person sees your note as worth answering.

Email is the channel for this test. Do not add automated social messages when replies slow down, then report the combined result as outbound. Do not launch a newsletter to warm the list. If a prospect replies and wants a call, take the call. If a customer refers a peer, accept the introduction, but tag it as referral rather than crediting email. Clean attribution matters more than flattering totals.

A weekly cadence turns effort into a motion

Run the same operating cycle every week so activity, learning, and customer follow-through have fixed places on the calendar. Founder sales collapses when prospecting happens only after the product work feels finished. Product work never feels finished. Reserve the blocks before the week starts.

On Monday, select and research the week's accounts. On Tuesday through Thursday, send a small batch each morning, handle replies, and conduct calls. On Friday, update every opportunity, review recordings or notes, and decide on one change for the next batch. Keep customer onboarding appointments separate from prospecting so new users do not lose attention when the pipeline becomes busy.

A plain spreadsheet is enough. Give every account one row with source, segment evidence, trigger, buyer, contact date, current stage, next action, next action date, loss reason, expected price, and activation status. Use fixed stage names: researched, contacted, replied, meeting held, qualified, proposal, won, lost, and activated. A stage records an observable event, not a feeling. Interested is not a stage.

Define qualification before the first call. A qualified opportunity has the problem you solve, a current reason to act, a person who can drive the decision, a plausible path to budget, and an agreed next step with a date. If one of those facts is missing, the opportunity stays unqualified. This discipline can make the pipeline look smaller, which is exactly why founders resist it. A smaller truthful pipeline is easier to repair than a large fictional one.

Write a next action while the call is fresh. Waiting to follow up is not an action. Send the sample, meet the operations lead on Thursday, and review security requirements by May 8 are actions. If the buyer will not agree to any dated next step, ask whether the project should close for now. Polite limbo steals more founder time than an explicit no.

Pete Kazanjy's Founding Sales describes founder-led sales as the work of discovering and refining the initial motion before handing it to professional sellers. I would treat documentation as part of discovery from day one. Save the email versions, call outline, qualification rules, common objections, proposal language, and onboarding steps. When a change improves a stage consistently, update the playbook and note the date.

Do not automate the sequence until you can explain why each step exists. Automation multiplies a known process; it cannot decide whether the segment has pain. At this stage, ten minutes spent reading an account can save an hour on a call with someone who could never buy.

Sales calls should diagnose before they demonstrate

A first call should establish whether the buyer's situation matches your hypothesis before you show the product. Founders often demo too early because the product is where they feel competent. The buyer then comments on screens while the underlying purchase conditions remain unknown. You leave with praise and no decision.

Open by confirming why the person took the meeting and what changed recently. Reconstruct the current process, quantify the consequence using the buyer's own numbers, and identify everyone involved in a decision. Ask how the company has tried to fix the issue and why the attempt remained insufficient. Only then connect the relevant part of the product to that process.

Keep the demonstration narrow. Show the path from the buyer's input to the promised outcome with their use case or a faithful sample. Skip features that do not affect the decision. When a prospect asks for something missing, ask what happens without it and who needs it. That question separates a genuine blocker from a familiar feature they expected to see.

End with a mutual decision. The next step can be a paid purchase, a bounded pilot, a technical review, a meeting with another stakeholder, or a no. Avoid a free pilot with no owner, success measure, or decision date. Free removes price friction but often removes commitment too, leaving the founder to support an experiment the buyer never prioritized.

Afterward, write down three facts: the trigger that created urgency, the alternative the buyer uses, and the reason the deal advanced or stopped. Do not write great call. That note cannot improve the next one. Over several calls, repeated facts will either strengthen the segment hypothesis or tell you where it breaks.

Objections should change the playbook selectively

Repeated objections reveal gaps in targeting, message, product, or buying process, but they do not all deserve a feature. Classify the objection before acting. If the prospect never felt the problem, revisit the segment or trigger. If the person feels it but cannot explain your claim, fix the message. If the claim matters but the product cannot deliver the outcome, address the product. If the buyer agrees yet the deal stalls, inspect authority, budget, risk, and next steps.

Walk through one failure in detail. Suppose five payroll firms reply, three take calls, and all say reconciliation is painful. Two cannot share client data under their current security process. One wants the product but needs approval from a partner who never joins the next call. The founder responds by building three reporting features because prospects mentioned them during demos.

That response is popular because shipping restores a sense of control. It is also wrong. The evidence points to a trust and buying-process problem, not reporting. The founder should define what data a test requires, prepare a clear explanation of handling and access, ask the buyer to involve the partner before scoping work, and test whether a lower-risk sample can prove the claim. Building reports may make the demo busier while leaving every deal blocked.

Keep an objection log with the account, exact wording, stage, classification, response, and result. Promote an answer into the playbook after it works across several comparable opportunities. If each account raises a different objection, the segment may still be too broad or the product may be entering workflows with incompatible requirements.

Pricing objections need the same care. Too expensive can mean the problem costs less than you assumed, the buyer doubts the outcome, the wrong person is negotiating, or cash timing is tight. Do not discount before you know which one applies. A lower price cannot repair weak urgency, and it can turn an implementation-heavy customer into a loss even when the contract closes.

Metrics should locate the broken stage

Use metrics to find where the motion loses qualified buyers, not to decorate an update. Early numbers are small and noisy, so keep raw counts beside rates and review a rolling four-week view. One unusually warm introduction or one large contract can distort a percentage.

Track delivered first contacts, positive replies, meetings held, qualified opportunities, proposals, wins, activated customers, and retained customers. For each rate, keep the denominator explicit. Positive reply rate means positive replies divided by delivered first contacts. Win rate can mean wins divided by qualified opportunities or proposals; choose one and never switch quietly.

Read the sequence as a diagnosis. Few replies usually point to account selection, timing, deliverability, or message. Replies without attended meetings point to a weak ask or poor scheduling. Meetings without qualification point to the wrong segment or trigger. Qualified deals without proposals point to an unclear process. Proposals without wins point to value, trust, price, authority, or competition. Wins without activation mean the sale promised an outcome onboarding did not produce.

Set thresholds as decisions, not borrowed benchmarks. For example: after each batch of 30 delivered messages, continue the wording if at least three right-segment buyers reply positively; otherwise review the list and message before sending another batch. This is a starting rule for disciplined learning, not a universal industry rate. Your price, category, buyer seniority, and urgency will change the numbers.

Revenue matters, but it arrives late. Pair it with evidence that predicts repeatability: similar accounts respond to the same problem statement, comparable buyers advance through the same stages, sales cycles cluster rather than wander, customers pay a consistent price, and users reach the promised result without heroic founder intervention. A founder can close an outlier through trust or persistence. A motion produces comparable outcomes from comparable inputs.

Do not call the strategy successful because meetings increased. The test works when the full path works: the segment is findable, outreach earns conversations, qualified buyers make decisions, customers pay, and they use the product enough to receive the result you sold. If the top of the funnel looks healthy while customers fail, more outreach will magnify churn and support work.

Repeatability comes before a sales hire

Hire a salesperson after you can teach the motion and provide enough qualified accounts for that person to run it. A rep should improve execution, not discover which market the company is in while the founder steps away. The premature hire fails because the founder cannot tell whether poor results come from the person, the segment, the message, or the product.

Evidence of repeatability is behavioral. You can describe the segment without exceptions. The same trigger earns attention across several batches. Calls follow a stable qualification path. Buyers understand the claim without a founder giving a different speech each time. Pricing has a defensible range. New customers activate through documented steps. Loss reasons repeat often enough that the team can address them.

Do not wait for perfect data, because an early startup will never have it. Look for consistency across several sales cycles and for a playbook another capable person can follow while you observe. Give an early hire recorded examples, stage definitions, message versions, objection notes, and a clear account list. Stay involved in calls until the rep can reproduce the process and you can separate coaching needs from market evidence.

Change the strategy when evidence breaks a core assumption. If well-matched buyers do not recognize the problem, revisit the segment. If they care but cannot act, change the buyer or trigger. If they act but fail to reach the outcome, fix the product or narrow the promise. Do not change all three at once. A clean test preserves enough of the system to tell you what changed.

Founders who want outside scrutiny can bring their segment card, email, and stage metrics to peers before widening the campaign. Sisters is a free, invite-only community where women building businesses can ask experienced peers for direct feedback, find advisors, and join workshops on subjects including go-to-market and sales.

The first motion may remain manual longer than you expect. That is acceptable while each week produces customers and sharper documentation. The moment the work repeats without new learning, standardize it. The moment the evidence contradicts the segment, stop polishing the spreadsheet and change the assumption. Discipline here is cheaper than pretending six weak channels add up to distribution.