Need and limit comparison Authored hypotheses, not conversion forecasts
KELVIN ZEE
INDEPENDENT RESEARCH · FREE TIER DESIGN
01 · THE THESIS
02 · WHERE IT APPLIES
03 · USE IS NOT NEED
04 · FOUR LIMIT STRUCTURES
05 · THE BOARD
INDEPENDENT RESEARCH
A free tier has to makethe paid need legible.
A volume meter can show heavy use and still miss a customer who uses the product rarely but faces a high-consequence moment.
SCOPE
Where this thesis applies
✓Self-serve paid conversion
Education
Ecosystem adoption
Collaboration
Network growth
A public-service commitment
Even inside a conversion-led product, the right limit depends on service cost, competitive context, customer trust, and whether a user can understand the paid value before the limit appears.
HYPOTHESIS
Usage intensityis not need intensity.
FOUR LIMIT STRUCTURES
None is universally better. Each selects for a different customer state.
MODEL
The preferred structure changesbecause the moment of need changes.
Ringed: the most revealed need for that user, in the model below.
KELVIN ZEE · INDEPENDENT RESEARCH
A free tier shouldsurface need, notjust meter use.
20 PAIRINGS IN THE MODEL BELOW
Brief reel, at your own pace
The thesis
A free tier has to make the paid need legible. A volume meter can show heavy use and still miss a customer who uses the product rarely but faces a high-consequence moment.
Where it appliesSCOPE
The thesis applies where the free tier's primary job is to create self-serve paid conversion, not education, ecosystem adoption, collaboration, network growth, or a public-service commitment. Even then, the right limit depends on service cost, competitive context, customer trust, and whether a user can understand the paid value before the limit appears.
Use is not needHYPOTHESIS
Usage intensity is not need intensity.
Four limit structures
A volume cap limits how much, a feature lock limits what, a history limit limits permanence, and a time trial limits how long. None is universally better. Each selects for a different customer state.
The boardMODEL
The preferred structure changes because the moment of need changes. The model below rings the most revealed need for each user, across 20 pairings.
The argument in 30 seconds. Each claim is labelled as scope, a hypothesis, or the model.
A free tier built for self-serve conversion has to make the paid need legible. A volume meter can show heavy use and still miss a customer who uses the product rarely but faces a high-consequence moment.
The model on this page compares five user situations with four common limits. Its scores are authored assumptions designed to expose the tradeoffs. They are not observed conversion rates or retention outcomes.
DECISION FRAME
Usage frequency is not need intensity
A limit converts when it appears at a moment that matters. Each row is a user situation and each column a limit. The ringed cell is the limit that scores best for that user, and it moves as the moment of need moves. Pick any cell to send it to the model.
User situation ↓Free tier limit →
Volume capLimits how much.
Feature lockLimits what.
History limitLimits permanence.
Time trialLimits how long.
Casual userOpens the product now and then. Never comes close to any limit.
Routine userSteady moderate use, week after week, with no single high-stakes moment.
Power userHeavy sustained use. Reaches any ceiling quickly and knows it.
Rare but high stakesBarely uses the product for months, then needs it badly for one thing.
Active evaluatorComparing this against alternatives right now, deliberately.
↘The ringed limit changes row by row: the moment of need decides, not how much the product is used.
Rare but high stakes · History limitRevealed need 88 of 100. Fires at the exact moment the product matters most to them. The cost: the moment it fires is also the moment they are least forgiving.
Each cell is one of the model's twenty authored pairings, shown as an illustrative index from 0 to 100, not a measured conversion or retention rate.
Where this thesis applies
This argument applies when the free tier's primary job is to create self-serve paid conversion. It should not be generalized to products where free access exists mainly for education, ecosystem adoption, collaboration, network growth, or a public-service commitment.
Even inside a conversion-led product, the right limit depends on service cost, competitive context, customer trust, and whether a user can understand the paid value before the limit appears.
Usage intensity is not need intensity
Volume caps focus on how much a customer uses. That works for sustained heavy users because they reach a ceiling quickly. It can fail for a rare, high-stakes user because low frequency hides the importance of the moment.
A history limit operates on a different axis. The product can remain fully useful until the customer needs an older record. The limit appears at the moment the missing capability matters. A feature lock and a time trial create different moments again.
The design question is therefore more specific than how much to give away. It is which need the paid tier resolves, and whether the customer encounters that need while the value is still clear.
Four limit structures
Volume cap: limits how much. It is legible and reaches heavy users, but it can turn the paid decision into a contest over quota size.
Feature lock: limits what. It works when customers know the missing capability exists and can connect it to their job.
History limit: limits permanence. It can reveal value when a customer returns to recover something important.
Time trial: limits how long. It forces a clean decision, but it also removes the long tail of users whose need arrives later.
None is universally better. Each selects for a different customer state.
How to read the model
Choose a user type and a limit. The 20 pairings compare conversion signal, continued use, trust cost, and revealed need. Move the rare, high-stakes user across the four limits. Then move the active evaluator across the same limits. The preferred structure changes because the moment of need changes.
The scores are pairwise judgments. They are useful for finding disagreement and designing tests. They are not calibrated forecasts.
INTERACTIVE MODEL
Compare a user state with a free-tier limit
Change the user situation and limit to see which need becomes visible, and what the design gives up.
Illustrative indices. These are Kelvin-selected scenario assumptions, not measured conversion rates, retention outcomes, or probabilities.
THE BRIEF'S EXPERIMENT
Move the rare, high-stakes user across the four limits. Then move the active evaluator across the same limits. The preferred structure changes because the moment of need changes.
Or start from a pairing
Rare but high stakesBarely uses the product for months, then needs it badly for one thing.
Volume capA monthly quantity ceiling. Limits how much.
Showing Volume cap for Rare but high stakes. Conversion 7, Still using at 90 days 64, Trust cost 11, Revealed need 8.
ILLUSTRATIVE INDICES / 0 TO 100
This pairing
Does the limit make the paid need obvious, and at what cost?
Conversion7/100
Higher means more of this group moves to a paid plan.
Still using at 90 days64/100
Higher means they are still active, whether or not they paid.
Trust cost11/100
Higher means the limit reads as extraction rather than a fair boundary.
Revealed need8/100
Higher means the limit makes it obvious what the paid tier is for.
Trade-off map
All twenty pairings by revealed need and trust cost. The dashed line joins this user's four options; click one to choose it.
What this buys you
Costs nothing, because this group never approaches the ceiling.
What it costs you
Structurally cannot reach the most motivated buyer the product has.
All four limits for rare but high stakes
Limit
Conversion
Still using at 90 days
Trust cost
Revealed need
7
64
11
8
18
62
19
24
Most revealing
71
69
44
88
9
26
63
12
WHERE THIS THESIS APPLIES
SCOPEThis argument applies when the free tier's primary job is to create self-serve paid conversion.
NOT FORProducts where free access exists mainly for education, ecosystem adoption, collaboration, network growth, or a public-service commitment.
DEPENDS ONService cost, competitive context, customer trust, and whether a user can understand the paid value before the limit appears.