When Should You Replace Pay-As-You-Go With a Subscription? The ClassPass Pivot
Short answer: Switch only when customers already repeat the same purchase, the subscription makes that repetition easier and expected usage leaves positive contribution after supplier costs. Run a paid cohort for three billing cycles before replacing pay-as-you-go. Keep the old model until retention, use and contribution show that recurring billing improves customer behaviour rather than disguising weak demand.
Subscriptions look attractive because revenue recurs. The bill can recur even when value does not, which produces cancellations, refund requests and a business constantly replacing disappointed members.
The correct question is not whether customers say they like a membership. It is whether they repeatedly perform the behaviour that the membership is meant to support, at a cost the business can afford.
ClassPass found repeat demand inside misuse
Payal Kadakia originally launched Classtivity as a search and booking marketplace for classes. A contemporary TechCrunch account says booking volume was too low to be worth disclosing. People browsed, but the listing model did not make them attend.
The company then tested a Passport: ten classes at different studios over 30 days for $49. The same report says the beta produced more than 20,000 reservations. Yet the one-off offer did not create the studio loyalty the company expected. Customers began using different email addresses to buy another Passport.
That behaviour could have been treated only as abuse. Kadakia treated it as evidence of a repeated job. In a founder interview about the pivot, she described testing the monthly ClassPass with 50 people in June 2013 at $90 for ten classes. By the third month, its revenue had overtaken the Passport product.
Accounts vary between 95% and 99% when describing how many surveyed users said they wanted to repeat the experience. That difference is a warning. Stated interest was not the decisive evidence. Paid renewal and actual attendance were.
My view is that you should not invent a subscription to repair irregular demand. Use it when repeat behaviour already exists and the present transaction makes repetition unnecessarily difficult.
Use the Repeat-Use Revenue Gate
The Repeat-Use Revenue Gate tests four conditions in order.
| Gate | What must be true | Measure | |---|---|---| | Repeat | Customers make the purchase again without prompting | Repeat-purchase interval and rate | | Convenience | Membership removes a real transaction burden | Activation and use versus pay-as-you-go | | Capacity | Usage remains deliverable at the promised price | Cost per active member and peak utilisation | | Retention | Customers stay after the first novelty cycle | Paid renewal and cohort contribution |
A subscription that fails the repeat gate should not proceed. Discounts can produce a busy first month without creating a recurring need.
Start with the job, not the billing interval
Define what customers are trying to repeat. ClassPass was not merely selling cheaper classes. It was selling regular access to variety. The earlier search engine helped someone find a class, while the subscription supported an ongoing exercise habit.
Look at transaction histories, not broad survey answers. How many customers buy twice within the natural cycle? What triggers the second purchase? Where does payment or choice create friction? Interview repeat buyers and customers who intended to return but did not.
If the job occurs twice a year, monthly billing is probably a mismatch. An annual plan, bundle or simple repeat-order process may fit better.
Set usage boundaries before the price
A fixed subscription transfers usage risk from the customer to the business. Your price must cover the distribution of use, not the average alone.
Model inactive, typical and heavy members separately. Include supplier payments, fulfilment, payment fees, support and any peak-capacity premium. Heavy use may be acceptable when it improves retention, but it cannot be ignored.
Limits can be expressed as credits, visits, units, locations or fair-use rules. They should match the customer's job rather than create a puzzle. A limit that customers discover only after joining will undermine trust.
The guide to [pricing a new product or service](/business-finance/how-to-price-a-new-product-or-service/) explains why competitor prices cannot replace your own contribution calculation.
Test paid renewal, not free enthusiasm
Recruit a small cohort with a clear price, allowance and cancellation route. ClassPass used 50 people for its early monthly test. Your number should be large enough to include different usage levels but small enough to serve safely.
Run at least three natural billing cycles. In each, record members starting, paying, activating, using their allowance, renewing, pausing and cancelling. Ask cancelling customers what they expected and what they actually used.
Do not count a card that renews without use as proof of value. Dormant members can make early margins look excellent, then leave together when they notice the charge.
Subscription, renewal, cancellation, tax and consumer rules vary by country and sector. Make continuing charges clear and obtain qualified local legal and tax advice before launching.
Worked example: MoveMix Studios
MoveMix sells access to independent fitness studios in Bristol. These figures are illustrative.
Under pay-as-you-go, 120 customers take an average of 2.2 classes a month at £14 each. MoveMix retains £3.50 after paying the studio. Monthly contribution before support is 120 x 2.2 x £3.50 = £924. Payment and customer support cost £180, leaving £744.
MoveMix tests a £59 subscription including six classes. It recruits 50 members. Forty-four pay for month two and 39 pay for month three. Month-three usage divides as follows:
| Member group | Members | Average classes | Studio cost at £9 per class | |---|---:|---:|---:| | Light | 10 | 2 | £180 | | Typical | 21 | 4 | £756 | | Heavy | 8 | 6 | £432 | | Total | 39 | 3.9 | £1,368 |
Month-three revenue is 39 x £59 = £2,301. Studio cost is £1,368, payment fees are £69 and support is £210. Contribution is £2,301 - £1,368 - £69 - £210 = £654.
The subscription cohort contributes £654 divided by 39 = £16.77 per member. A comparable group of 39 pay-as-you-go customers contributes 39 x 2.2 x £3.50 = £300.30 before its share of support and payment cost.
The subscription appears better, but 11 of 50 members have left within two renewals. MoveMix does not replace pay-as-you-go. It keeps both options and sets a gate: at least 70% of the original cohort must remain after six months, with average contribution above £14 per active member and no studio exceeding agreed spare capacity.
That decision sacrifices a clean recurring-revenue story. It preserves real customer choice while the evidence matures.
Watch the supplier side
In a marketplace, a membership can delight buyers while damaging suppliers. A low customer price may be funded by studios, practitioners or other providers receiving less for popular capacity.
Separate genuinely spare capacity from sessions the supplier could have sold at full price. Track the member's effect on total supplier revenue, repeat direct customers and peak availability. A supplier who earns incremental income on Tuesday morning may lose money on Saturday's full class.
Agree payment timing, cancellation handling and capacity controls before expansion. If the model works only while suppliers misunderstand its economics, it does not work.
Decide after three cycles, expand after six
During week one, identify customers who have already purchased twice and the interval between purchases. In week two, price a limited membership from low, typical and heavy usage.
Recruit the cohort in week three, then measure activation and delivery. Review paid renewal after each of the next three billing dates. Keep pay-as-you-go available as a comparison.
After three cycles, stop if the subscription does not improve use or contribution. Continue the test when the result is promising but retention is immature. Expand only after six cycles show that customers stay, suppliers benefit and heavy users remain affordable.
Related guides
Frequently asked questions
How many repeat purchases prove that a subscription could work?
There is no universal count. Look for a repeat pattern across a cohort, not one enthusiastic customer. As a working guide, identify at least 30 customers who have bought twice within the natural usage cycle, then examine how many buy a third time without a large discount. A high-value specialist service may need fewer observations. The important evidence is consistent timing and purpose. If repeat purchases happen for unrelated reasons or only after promotions, recurring billing may not improve the customer's job.
Should I offer a free trial before charging?
Only when customers need to experience the service before understanding its value and your trial can be delivered cheaply. A paid first month gives cleaner evidence of willingness to pay. If you use a free trial, measure conversion, activation, cancellation and contribution after the first paid renewal. Make the charging date and cancellation process prominent. Consumer and subscription rules vary, and automatic conversion can create legal and trust problems. Obtain local advice and never rely on customers forgetting to cancel as the economic model.
What if heavy users make the subscription unprofitable?
First confirm that heavy use is the cause, rather than supplier pricing or weak membership fees. Model contribution by usage band. You can raise the price, introduce credits, exclude peak capacity or offer a higher tier, but changes must be clear and fair. Do not describe a service as unlimited while quietly preventing legitimate use. Heavy users may also be your strongest advocates, so calculate retention and referrals before restricting them. Test the revised boundary with new members rather than unexpectedly removing value from existing ones.
Can I keep pay-as-you-go alongside the subscription?
Yes, and in most early tests you should. The transactional option creates a useful comparison and serves customers with irregular demand. Watch whether the subscription genuinely increases total use or merely moves frequent buyers onto a discount. Keep pricing simple enough that customers can choose without a spreadsheet. If one option persistently loses money or causes confusion, remove it after giving customers notice. The goal is not to force everyone into recurring billing. It is to match each repeat pattern with an economically sound way to buy.
Which retention period matters most?
Use enough cycles for the initial enthusiasm and any introductory offer to expire. Three paid renewals can reveal immediate mismatch, while six months gives stronger evidence for a monthly consumer service. Annual subscriptions need activation and interim usage measures because renewal takes too long to guide an early decision. Compare cohorts that joined under the same promise and price. A blended retention figure can hide deterioration. Track whether people who join later retain better after you change onboarding, allowance or positioning.
Should I switch if recurring revenue is more predictable but less profitable?
Not automatically. Predictability has value because it can improve purchasing and staffing, but a consistently loss-making subscription simply makes the loss recur. Quantify the operational saving from predictable demand, then include it in contribution. Also model the cash cost of refunds, supplier payments and tax. A modestly lower margin may be acceptable if retention is strong and service capacity becomes easier to plan. Set a minimum contribution floor before the test so the attraction of recurring revenue does not move the decision rule afterwards.
Comments (0)