Free Users to First 100: A 30-Day Referral Spec

You have a free tier, a handful of users who actually use it, and no paid growth budget. The fastest way to turn that into your first hundred signups is a 30-day referral loop that pays in extra usage, not cash. OpenAI began a ChatGPT Free referral offer in India, Indonesia, and Mexico on August 18.

The offer closed on September 17, creating a 30-day participation period. It is small enough to run by hand: identify engaged free users, message them directly, reward extra usage, cap stacking, and add guardrails before launch.

Start with the users who already act like advocates

Shortlist the free users who send feedback, hit limits, or mention the product to a colleague. Aim for a list you can message by hand, not a dashboard you need to build before launch. For the first hundred, manual outreach is the growth tactic that works because it does not need scale. Start with a note you can send in one sitting. Do not wait for a polished page.

Message each person directly. Say why you picked them, what the reward is, and what you want them to do. A note that names their recent behavior is easier to trust than a public post. Ask for one person, not a list, so the loop feels personal rather than promotional. If someone asks for a bigger reward, hold the line. The test is whether the product is worth passing along.

Run the loop as a spec, not a promotion

• Target engaged free users.
• Give the referrer doubled usage for seven days per qualifying referral.
• Cap at three referrals or 21 days.
• Restrict to the same country or same cohort, and ban self-referral.
• Offer no cash.

Guardrails are not bureaucracy. They keep one user from taking the whole program, keep the reward scarce enough to matter, and keep the loop from becoming a spam channel. A hard stop also makes the loop easier to explain in a short note.

Pay in product, not cash

Product-denominated referral rewards align incentives with retention and preserve cash margins, unlike cash referral rewards. Cash referral incentives remove friction but are expensive and attract gaming, so they require caps, activation qualification, and fraud monitoring. If your free tier already has a storage limit, ChatGPT Free users can store up to 500MB in Library. That kind of product surface is a natural place to put a referral reward. The user gets more of the thing they already use, and you do not pay out of pocket.

Dropbox's referral program gave both the referrer and the new user 500MB of extra storage. That storage reward was denominated in storage, a product-usage incentive with near-zero marginal cost that deepened usage. The program expanded its user base from 100,000 to 4 million users over 15 months.

Measure the loop like a retention problem

Track activation and retention, not just signups. Keep the table small enough to review by hand: referrer, referred user, activation date, and whether the referred user came back. Use it to decide who gets the next personal note. Review it often enough that the table tells you which referrer is producing users who stay, not just users who click.

Dropbox's referred customers showed 18% higher retention and 25% higher average spend. The mistake is treating the referral as a signup event. What matters is the referred user doing the thing that made the referrer a fan. If the referred user does not activate, fix the onboarding before you send more invites.

Run the loop until the referred users have had time to come back. Do not chase a bigger number while the first referred users are still learning the product. If the loop works, the next batch of invites comes from those users, and you spend your time onboarding them.