Let Them Taste It First: Freemium Strategy for Vibe Projects
Freemium fails when the line between free and paid is drawn in the wrong place. This guide covers what the free plan must include, three ways to draw the line, upgrade triggers, the cost ledger for free users, and a 12-item pre-launch checklist.

He built a notes app called Sprout Notes — the sample product and people in this article are fictional, and all figures are illustrative examples or the recommendations of this guide, not measured data. In its first year the app was entirely free, users grew to 1,000 sign-ups, and the server bill grew with them. He gritted his teeth and shipped a paid plan: the free plan could hold only 20 notes, and anything beyond that cost money. New sign-ups halved overnight, and existing users were furious: "Yesterday I could write as much as I wanted, today the 21st note pops up a paywall — this is robbery." Three months later he changed the free plan back to unlimited notes and charged for "advanced features" instead. Sign-ups came back, but paid conversion was only 3.6% — out of 1,000 sign-ups, just 36 people paid.
The problem was never "whether to charge". It was where the line between free and paid was drawn. Freemium is hard not because of the price number, but because you have to decide: what does the free plan give, so that users start using it, get hooked, and then willingly pay — and what exactly does the paid plan add, so that paying feels reasonable? Draw the line where users care most, and they feel robbed; draw it where they do not care at all, and they never have a reason to upgrade. Products built by vibe coding get their first users almost entirely on "let me try it for free" — whether the line is drawn right decides whether the product survives until the day it charges money.

1. Freemium is not a free trial
Separate the two concepts people confuse most. A free trial is a countdown: 14 or 30 days, and when it ends you pay or lose access. Freemium is a permanent split: the free plan works forever, with limits on features, capacity, or seats, and you upgrade when it is not enough. One is "eat the whole meal first, pay afterwards"; the other is "the snack is free to taste, the main course costs money". This site already has a guide on trial expiry and downgrade, about how to keep people at the moment a trial ends; this guide is about something else: how to draw the line while the free plan exists permanently.
The boundaries with two neighbouring guides need drawing too. The pricing and paywall guide covers where and when the wall appears: when it pops up, how often, what the copy says; this guide covers what sits on each side of the wall: what free gives, what paid adds. The credits ledger guide covers how billing is implemented and reconciled; this guide covers the tiering strategy, and billing implementation comes afterwards. Do not mix the three questions: tiering first, then the wall's timing, then the ledger. Reverse the order and you get a product where the free plan has nothing and the paywall pops up everywhere.
Whether your product fits freemium comes down to one test: can users experience the core value without paying? A notes app can: writing notes and finding them cost nothing, and the more you write, the harder it is to leave. But if the core value is a one-off result — a one-time resume review, a one-time contract check — users leave after one use, and giving it away free is giving it away for nothing. Those products fit a free trial or pay-up-front better. Freemium only fits one kind of product: the longer people use it, the more they accumulate, and the costlier it gets to leave.
2. What goes in the free plan: three lines
The free plan is not a "crippled version" — it is part of the complete product. Decide what goes in it with three lines. First line: the core value must be free. For a notes app, the core value is "write it down and find it again", so writing notes, searching, and syncing all belong in the free plan. Lock the core value behind payment and users never get to find out whether the thing is useful to them — they sign up and leave the next day. The free plan has to make users say within the first week, "I cannot live without this now". If they cannot say that, the line is drawn wrong.
Second line: charge only for three kinds of things. The first is capacity: number of notes, storage space, file size. The second is collaboration: multi-person editing, shared spaces, permission management. The third is automation: scheduled reminders, automatic organisation, AI summaries. Why only these three? Because they share one trait: the deeper users go and the more attached they get, the more they need them. A brand-new user does not care about collaboration or automation — they care whether writing feels smooth. A user with 500 notes after three months starts caring whether they can edit with colleagues and organise automatically. Paid features should grow along the user's journey, not block the front door.
Third line: the free plan must contain an aha moment. The aha moment is the first time a user thinks "this is genuinely useful" — for a notes app, maybe finding a note from six months ago in three seconds of searching, or seeing the same note sync instantly between phone and laptop. The free plan must include at least one such moment, and it must happen within 7 days of signing up. Users who do not hit it in 7 days basically never come back. When designing the free plan, ask first: on which day, on which action, will users first feel this is useful? If you cannot answer, do not think about charging yet.
3. Three ways to draw the line: by amount, by feature, by seat
There are only three ways to draw the line, and you should pick one as the main one — do not mix all three. The first is by amount: the free plan allows 100 notes a month and 1 GB of storage, and beyond that you pay. It is simple, and users understand it at a glance; the downside is that it easily lands where it hurts most — the 101st note is often the user's most important one, and hitting a wall there invites the most complaints. Drawing by amount fits products where usage itself is the value: storage, note counts, export counts. A reference for setting the number: let 80% of free users never touch the limit, and only the most active 20% ever hit it. The 80% who never hit it are your word of mouth; the 20% who do are your paid candidates.
The second is by feature: the free plan writes, searches, and syncs; the paid plan adds AI summaries, version history, and offline mode. The upside is that the free plan is always enough, so nobody feels robbed; the downside is that "advanced" is in the eye of the beholder — a feature you think is premium may mean nothing to users. Drawing by feature fits products whose advanced features really are advanced: AI summaries, automation, analytics. Test it simply: read the paid features' names to five real users and ask whether they would pay for them. If fewer than three say yes, the feature cannot carry a paid plan.
The third is by seat: one person free, three or more people pay. The upside is that it binds naturally to how people use the product — one person using it is a personal tool, three people using it is a team tool, and team tools charging money is only fair; the downside is that individual users never have a reason to upgrade, so all paid conversion depends on team users. Drawing by seat fits products with strong collaboration: shared lists, team spaces, multi-person editing. You can combine the three ways, but only one can be the main line: main line by amount, features as decoration; main line by feature, capacity generous; main line by seat, the personal plan not stingy. Squeeze on all three at once, and users feel you are charging them everywhere.
4. Upgrade triggers: appear when users need it most
With the line drawn, the next question is when upgrade prompts appear. The worst approach is popping them up everywhere: once when the app opens, once while writing a note, once more in settings. Users are not persuaded by prompts — they are annoyed away by them. A good trigger has one standard: it appears at the exact moment users need the paid feature most. When a user writes their 95th note and is about to hit the 100-note limit, a message saying "5 more notes until the free limit — upgrade for unlimited notes" is helping; showing an upgrade prompt to someone who just signed up and has not written a single note is shooing them away.
Triggers come in three kinds. The first is capacity: remind once when usage reaches 80%, and again at 100%, and do not disturb them in between. The 80% number matters: remind too early and users wonder what you are talking about; remind too late and they have already hit the wall and are angry. The second is feature: when users click a paid feature like AI summaries, show a small box saying "This is a paid feature — available after upgrading", with a "See the paid plan" button, and do not jump straight to the payment page. They are curious, not buying — do not mistake curiosity for purchase intent. The third is scenario: the first time users invite a second collaborator or try to export all their data, those actions are themselves upgrade signals — mention the upgrade after the action completes, not by blocking the action beforehand.
Frequency needs a cap too: the same user gets at most 2 upgrade prompts a month. Past 2, a prompt stops being a reminder and becomes harassment — users either turn off notifications or uninstall. The number 2 is a recommendation; adjust it for your own product, but there must be a cap, written in the product spec, not left to feel. Finally, every upgrade prompt must offer a "Don't show again" option. Give users an exit from being asked, and the ones who stay are the ones who genuinely want to upgrade; refuse them even the refusal, and they will refuse with an uninstall.
5. The cost ledger for free users: do the maths before the romance
Free users are not free: every one of them costs you server money, storage money, and AI-call money. Nine out of ten freemium projects that die are not killed by angry users but by the bill — sign-ups grow faster, the bill grows faster, and in the end it is the bill that drags the project down. Once the tiers are set, do three sums. The first is storage: how much storage one free user takes on average, times the number of free users, times the unit price — that is the monthly storage cost. For anything beyond the free allowance, either cap upload size or turn large files into a paid feature; do not let free users store without limit.
The second is compute: AI summaries, image processing, full-text search all cost money per call. If the free plan includes any of these, cap them: 10 AI summaries a month, for example, then wait until next month or upgrade. An uncapped AI feature is a cash machine for abusers. The third is abuse: one person registering 50 free accounts, using the free allowance to run crawlers, using your service to serve someone else. The defences are unglamorous but effective: verify email at sign-up, cap registrations per device, cap call frequency for free accounts, and clear the data of free accounts that have not logged in for 90 days. The 90-day figure is a recommendation — send a reminder email before clearing, and give users 7 days to come back and keep their data.
One bill is often forgotten: support cost. Free users ask questions too — "why won't it sync" — and there are dozens of times more free users than paying ones, so once questions pile up, one person cannot answer them all. The fix is tiered support as well: free users go through self-serve channels — a help centre, FAQs, a community where users help each other; only paying users reach you directly, by email or live chat. That is not favouritism, it is arithmetic: your time is finite and must go to the people who pay. Write the help centre before the free plan ships, covering at least the three most common questions — sign-up, sync, and password recovery — with screenshots. Those three are 80% of free users' questions, and answering them in writing saves you half your weekly replies.
After the sums, set one number: the monthly cost cap per free user — say 0.5 yuan. Any free user past that number is either abuse or a line drawn in the wrong place. With that number, the free plan stops being "unlimited generosity" and becomes "generosity within budget". Romance is fine, but the maths comes first — freemium that cannot do its sums does not survive its first viral moment.
6. The funnel from free to paid: what healthy numbers look like
Freemium's results end up in one funnel. This guide works through it with a fictional example: 1,000 people sign up; 420 activate within 7 days — writing their first note and seeing sync work; 180 come back at least once in the following week and become weekly active users; 36 end up paying. The paid conversion rate is 3.6% — out of every 100 sign-ups, 3.6 pay. Each of the three numbers, 420, 180, and 36, has an action attached: if activation does not rise, the onboarding is the problem; if weekly active does not rise, the product gives people no reason to return; if paid does not rise, the line is drawn wrong. A funnel is not for reporting results — it is for locating the fault.
The three conversion rates also relate to each other: sign-up to activation is 42%, activation to weekly active is about 43%, weekly active to paid is 20%. Fix whichever step drops hardest first. Many products jump to blaming "low paid conversion" when activation already collapsed — 1,000 sign-ups but only 200 activations, and no amount of payment optimisation helps because the pool is empty. Only when activation and weekly active are healthy and paid alone is low does the line itself take the blame. Fix the funnel top-down: fill the top first, then work down.

7. The five easiest mistakes
The first mistake: the free plan gives too much, and the paid plan has nothing left to sell. Unlimited notes, unlimited storage, free AI summaries — and the paid plan is left with nothing but a "support us" label, so conversion is naturally zero. The free plan should be generous, but not so generous that the paid plan has no reason to exist. The test: cover the paid feature list, look only at the free plan, and ask yourself, "If I were a user, what reason would I have to upgrade?" If you cannot answer, you gave too much.
The second mistake: the free plan gives too little, and users leave before they get hooked. Five notes, one synced device — users hit the wall on day three and uninstall on day four. The free plan has to let users get hooked before talking about upgrades. The test: can the free plan carry a user through one complete aha moment? If not, you gave too little.
The third mistake: the line keeps moving. Unlimited notes today, 100 notes tomorrow, 50 the day after — existing users feel cheated. Once a line is published, it is a promise. If it must change, it can only move towards generosity: raise the limit, move features down to free — never tighten it. Before tightening anything, give 30 days' notice and grandfather existing users: they keep their old free allowance, and only new users follow the new rules.
The fourth mistake: upgrade prompts everywhere. As said above, at most 2 a month. One more detail: in the month right after users pay, do not send them a single upgrade prompt. People who already paid do not need to be sold to again — selling to them only makes them think "this product only cares about money".
The fifth mistake: watching only conversion and ignoring what free users say about you. Many free users will never pay, but they recommend friends, write reviews, and answer questions for you in groups. They are your acquisition channel, not a cost burden. Judge freemium by more than the 3.6% conversion rate — also look at how many new sign-ups free users bring in. If free users bring nobody new, the model cannot grow no matter how high conversion goes.
8. The 12-item pre-launch checklist
Before the free plan ships, run through these 12 items and tick them all. First, the core value works fully in the free plan — the aha moment is reachable without paying. Second, the line has only one main way of being drawn: by amount, by feature, or by seat — pick one of the three. Third, 80% of free users never touch the limit; only the most active 20% do. Fourth, the paid features' value can be explained to five real users, and at least three of them nod. Fifth, upgrade triggers appear only at the moment users need them most, never blocking the front door. Sixth, the same user gets at most 2 upgrade prompts a month, with a "Don't show again" option. Seventh, there is a monthly cost cap per free user, and breaching it triggers an abuse check. Eighth, per-call features like AI have a monthly usage cap. Ninth, sign-up verifies email, and registrations per device are capped. Tenth, free accounts idle for 90 days get a reminder email and 7 days' grace before their data is cleared. Eleventh, once published, the line is never tightened — changes only move towards generosity, and existing users are grandfathered. Twelfth, the funnel's three numbers — activation, weekly active, paid — are reviewed weekly, and whichever step drops hardest gets fixed first.
One honest note to close: freemium is not "free first, figure out charging later" — the free plan is part of the product itself. Let users taste it first; once they taste it, paying follows naturally. If they taste nothing, no amount of upgrade prompts on the wall will help. Make the free plan worth using first — only then do you have the standing to talk about charging.
Comments (0)
Related articles
localStorage only stores strings and has no indexes. IndexedDB is a real database in the browser: structured data, far larger capacity, indexes, and transactions. Following MDN's Using IndexedDB, IDBDatabase, and IDBObjectStore: the five-step routine, opening, upgrades, keys, indexes, transactions, add versus put, get, cursors, ranges, and multi-tab upgrades.
You already call functions every day; this guide names what you have been doing. Following MDN's Functions, Arrow function expressions, and Rest parameters: declarations versus expressions, parameters and arguments, default values, rest parameters, spread syntax, return values, arrow function shorthand, the this trap, callbacks, and scope.
One page should stand up on a phone, a tablet, and a computer. This guide follows MDN's Responsive web design, Using media queries, and the viewport meta reference: fluid HTML, the viewport tag, fluid grids and max-width, images that never burst their container, media query syntax, breakpoints set by content, mobile first, responsive type, testing capability instead of width, and a pre-launch checklist.