You may be able to test a business offer without buying inventory or software, but operating with literally zero cost is not guaranteed. A designer, for example, might sell one paid prototype before ordering materials or building a complete product. The practical goal is to prove that a real customer has a problem and a credible path to payment before committing to a large expense.
Start by listing four facts: the cash you can risk, the skills or assets you already have, the first customer problem you could solve, and the local requirements you must meet. Registration, permits, taxes, payment processing, insurance, delivery, and support can all affect the cost. The U.S. Small Business Administration launch guidance is a useful starting point, but requirements depend on your location and activity.
If you cannot meet a required operating or delivery cost, revise the offer or timing rather than calling the launch cost-free.
Use skills or assets you already have to test one paid offer before taking on fixed costs. “No money” is a constraint, not an exemption from operating obligations.
Choose an offer with low upfront cash exposure
Compare the cost you cannot recover if the idea fails, not just the advertised cost of starting. A service uses time instead of inventory, but still requires customer acquisition and delivery. Resale can use items you already own, although suitability and marketplace rules matter. A digital product avoids physical shipping, but still takes work to create, sell, support, and maintain.
| Path | First offer | Main exposure | Evidence to seek |
|---|---|---|---|
| Sell a skill | Defined, paid service pilot | Time, acquisition, delivery | Agreement on scope, price, and timing |
| Resell owned items | Listings for suitable items | Time, platform terms, transaction costs | Relevant inquiry or completed sale |
| Test a digital product | Paid pilot or clearly described preorder | Creation, fees, support, refunds, marketing | Specific requests, preorder, or paid pilot |
Choose the path with the smallest irreversible expense. Do not commission a full course, template library, or inventory run after one vague compliment. A compliment is feedback, not evidence that someone will pay. For a digital product, state what is ready now, what remains to be delivered, and the delivery and refund terms before accepting payment.
Validate demand before building or buying
Run the first test manually. The founder owns the decision, and a spreadsheet or offer register can serve as the initial system of record. Keep each request or transaction as its own row rather than recording only a total. A request, a conversation, a preorder, and a completed purchase are different kinds of evidence.
A useful offer record can include offer_id, request_id, source, date, stated problem, next step, payment status, and decision. One illustrative record is:
{
"offer_id": "pilot-001",
"request_id": "req-004",
"source": "prospect conversation",
"date": "2026-10-12",
"stated_problem": "Needs a one-page product site",
"next_step": "Review fixed-scope pilot offer",
"payment_status": "not paid",
"decision": "follow up on stated date"
}This is a proposed editorial record, not a vendor schema. If a request or transaction has a stable external identifier, store it. If not, define a suitable composite key. Enforce uniqueness with a database constraint or transactional upsert when concurrent writers are possible. A read-then-insert check alone can create duplicates. Keep raw requests, payment events, and aggregate summaries at different grains.
A request shows interest, while a payment shows a transaction. Do not let an aggregate response count, a compliment, or a repeated page view stand in for a sale. Label every summary with its offer, evidence type, and date range.
Make a lean plan and set a cash boundary
Before accepting work or seeking funding, write a one-page working plan covering the customer, problem, offer, price, delivery steps, expected costs, and next test. Separate must-pay operating costs from optional upgrades such as premium branding, paid software, or a larger product run.
If you need funding, state the exact amount, what it will buy, and which next milestone it enables. The SBA’s business planning guidance discusses funding needs and financial projections. Detailed projections matter more when seeking financing, while early validation can use a leaner plan. If the next experiment can run without borrowing, applying for a grant, buying software, or placing a bulk order, run that smaller experiment first.
Choose funding by its actual terms, not its label
Funding routes are not interchangeable. A grant generally does not require repayment, but eligibility and application cycles are specific. A loan must be repaid and requires underwriting. Accelerators commonly provide selected support such as mentoring, technical help, workshops, or credits rather than unrestricted cash. Equity funding exchanges ownership for capital.
The SBA says it does not provide general grants to start or expand ordinary small businesses. Its grant activity focuses on specific areas and resource programs. Read the SBA grant scope before treating an opportunity as relevant.
SBA Microloans are made through designated nonprofit intermediaries and can be up to $50,000. SBA says the average microloan is about $13,000, although rates and terms vary by intermediary. Proceeds cannot be used to repay existing debt or purchase real estate. Check the Microloan requirements and discuss fit with an intermediary. SBA Lender Match may connect a business with interested lenders, but it is not approval or a loan guarantee. Applicants still apply directly to lenders through their processes. See SBA Lender Match.
Match the expense first, then check the business stage, geography, industry, founder eligibility, funding type, and permitted use. A program that mentions small businesses may still exclude an idea-stage applicant, a particular expense, or a specific location.
- Open the official publisher’s page, not only a search result or roundup.
- Record whether the program is open, closed, recurring, retired, historical, or uncertain.
- Capture the current cycle, deadline, official URL, page title, and checked date.
- Match geography, business stage, industry, and applicant eligibility.
- Confirm the funding type and whether it allows the exact proposed use of funds.
- Save the supporting passage instead of relying on a copied summary.
- Send missing or conflicting terms to a named human reviewer.
Keep the record at the right level of detail. One opportunity-cycle record represents one program cycle. One page-observation record represents one retrieval of an official page at one time. One citation record represents one supporting passage. A proposed opportunity-cycle key could combine the publisher, normalized program name, and cycle identifier. Repeated page checks should create new observations, not duplicate opportunities. If multiple processes can write records, enforce the key with a database uniqueness constraint or atomic upsert.
Simple rules are preferable to AI for checking whether a deadline has passed, identifying an explicit retired label, or matching a stated location requirement. AI may draft a summary from a supplied official page or flag missing fields, but a human should confirm status, cycle, eligibility, and source passage. Keep applicant names, demographic details, financial documents, and credentials out of AI screening unless privacy terms and permissions allow their use. Store only the information needed for matching.
Consider three status examples. FedEx identifies its U.S. Small Business Grants Program as a retired legacy program that operated through 2024, so it should not be presented as an open annual application. Check FedEx’s current program page. Google for Startups Accelerators select technically oriented startups and provide support such as mentoring and technical work; they are not a general source of startup cash. Check the official accelerator overview for fit. Glossier’s grant FAQ describes eligibility for qualifying U.S.-based Black-owned beauty businesses selling physical products. Confirm current cycle dates and terms before relying on it.
Deliver the first offer, then reinvest from evidence
Deliver the smallest complete promise you sold. Record actual time and costs, then ask what was useful or missing. The first delivery can show whether the offer is repeatable, whether the price covers the work, and which bottleneck a next investment should solve.
For a digital download, choose a platform based on how you want to sell and fulfill. Etsy documents instant-download and made-to-order digital listings in its Seller Handbook. Shopify documents file or link delivery and automatic or manual fulfillment through its Digital Products app. Shopify also documents access settings, order management, and file limits for its current app, subject to plan and store-status conditions. Gumroad supports digital products and certain service-style products, with restrictions; see its guidance on selling services.
Before publishing, confirm that you have rights to distribute every file, test the purchase and delivery path, check access to external links, and state delivery and refund terms. No physical shipping does not mean no operating cost. Platform and payment fees, taxes, customer support, refunds, and marketing can still apply.
Track product_id, version_id, order_id, delivery status, refund status, actual delivery cost, and customer feedback. If an asset changes, record the new version and decide how existing buyers will be handled. A product version is not the same row as an order, delivery event, or refund event.
A spreadsheet can remain the record system while requests and orders are manageable. When repeated follow-up or handoffs make that difficult, consider whether CRM systems fit the workflow. This is a workflow option, not evidence that a ready-made integration exists. The next spend should address a specific observed bottleneck, not an assumption that a more elaborate setup will create demand.
A practical first test
Imagine a pre-revenue U.S. designer seeking $1,500 for prototype materials. Before applying anywhere, the designer should check whether the opportunity accepts pre-revenue businesses and funds that exact expense. The opportunity may instead require an operating business, a particular founder group, a research purpose, exporting activity, or a different industry.
If no suitable grant fits, compare a smaller prototype test, customer commitments, preorders with clear delivery and refund terms, or a loan for which the business may qualify. This example is hypothetical. It does not imply that any named program will fund the applicant.
Your next step
Today, write down your available cash, one customer problem, and one offer you can deliver using skills or assets you already have. Check unavoidable local requirements, then show a clearly scoped paid offer to relevant prospects and record what they do. Spend more only when the evidence identifies a specific, solvable bottleneck.