A two-week, two-location experiment in booking friction — with the numbers most operators never get to see.
Every operator who takes reservations has had this argument — with a business partner, a GM, or themselves at 1am:
"The credit card requirement is killing our bookings."
It feels true. You watch guests get three steps into checkout, hit the card form, and vanish. Every abandoned booking looks like money the card field stole from you.
Two weeks ago, we stopped arguing about it and tested it. At one of our locations, we removed the credit card requirement entirely — book with just your name, email, and phone. Our other location kept the card-on-file policy, unchanged, as a control.
The card-free store's bookings jumped 65% in two weeks.
We're putting the card requirement back anyway.
Here's the whole story, because the numbers underneath the headline are the part nobody publishes.
The setup: a natural A/B test
We run multiple locations of the same business on the same booking platform, same offerings, same price points, same reminder emails. That gave us something most operators never have: a real control group.
- Location A: card requirement removed. No charge at booking — just contact info. No cancellation or no-show fee possible (there's no card to charge).
- Location B: unchanged. Card saved at booking, standard late-cancel/no-show fee.
Same weeks. Same weather. Same marketing conditions. One variable.
What went up
The friction removal worked exactly the way you'd hope:
- Confirmed bookings: +65% (26/week → 43/week)
- Checkout completion: 90% → 97% — almost nobody who reached the final step bailed
- More guests started checkout, too — the "no credit card required" promise pulls people in earlier in the flow
If we had stopped measuring here — and most operators do, because this is where the dashboard smiles at you — we'd have declared victory and rolled it out everywhere.
What went down (a cliff)
Then the no-card cohort's reservation dates arrived. Same 7-day windows, both stores:
| No-card store | Card store (control) | |
|---|---|---|
| Cancellation rate, before | 4% | 12% |
| Cancellation rate, after | 17–24% | 7% |
| No-show rate, before | 0% | 2% |
| No-show rate, after | 14–16% | 3% |
Read that again: the store that removed the card went from zero no-shows to roughly one in six reservations ghosting — in two weeks. The control store, same brand, same weeks, same reminder sequence, stayed at 2–3%.
That's the cleanest causal read you'll ever get outside a lab. It wasn't seasonality. It wasn't our reminders breaking. It was the card.
And the no-shows weren't random Tuesday-afternoon slots. They clustered on Friday and Saturday evenings — the exact inventory that sells out, the slots where an empty table has the highest opportunity cost, the reservations that blocked other paying guests from booking.
The math that actually matters
Here's the arithmetic every operator should run before touching their booking policy:
43 bookings/week × ~33% cancel/no-show = ~29 kept reservations
The old policy kept 26.
We booked 65% more to host roughly the same number of guests — while giving up every dollar of cancellation-fee protection, prepping stations for parties that never arrived, and holding prime weekend slots for ghosts.
A booking is not revenue. A kept booking is revenue. The card field wasn't blocking demand; it was filtering for commitment. Turns out a $0 hold on a credit card is one of the cheapest commitment devices ever invented.
The trap that almost got us: the "conversion doubled" illusion
One more finding, and honestly it's the one that justifies the whole experiment.
During the same window, we also paused our Meta ads. Website traffic dropped — and our booking conversion rate nearly doubled. It would have been irresistible to credit the card removal.
Except the control store's conversion doubled too. Same lift, no policy change.
Killing the ads removed a layer of low-intent traffic from the denominator, and the conversion rate rose at both stores mechanically. The card removal's real contribution to conversion was a few points at the final checkout step — real, but small.
If you change two things at once, your dashboard will happily let you believe the wrong one worked. Without a control location, we'd have attributed the doubling to the card change and scaled a policy that was quietly costing us a third of our reservations.
What we're doing instead
We're not swinging back to maximum friction and pretending nothing happened — the demand lift was real. The next test is the middle ground: a small deposit at booking, credited toward the experience. Lighter than a full card-on-file flow, but it restores the one thing that mattered: skin in the game.
The five takeaways
- Friction isn't your enemy. The wrong friction is. A card form at checkout filters out exactly the bookings you didn't want.
- Measure kept reservations, not bookings. Any policy change that lifts bookings can still lose you revenue.
- No-show damage arrives on a lag. Bookings change day one; no-shows show up when those dates arrive, 1–3 weeks later. Judge no policy in its first week.
- Never change two things at once — and if you must, keep a control. Our second location was the only thing standing between us and a very confident, very wrong conclusion.
- Your worst no-shows take your best slots. The cost isn't spread evenly — it lands on Friday and Saturday nights.
We ran this experiment on our own booking platform, which is why we could see every stage of it — funnel step conversion, cohort no-show rates, per-slot ghost bookings — without stitching together spreadsheets. That platform is now ArtistryHost, the booking system we're opening to other experience-based businesses. If you're running classes, tastings, or events and your booking tool can't answer "what's my no-show rate on bookings made without a card?" — that's the gap we built it to close.