How one test saved a sacred cow from slaughter.

We recently put one of our oldest sacred cows on the chopping block: proration. 

Before any butchering began, though, we tested our assumptions. Our goal was to find out if proration actually improves driver retention, revenue, and sentiment.

Why? 

Because testing is learning, and learning helps us understand what to improve. This is central to our company thesis that the first step to optimizing an asset is understanding it.  But we don’t pick what to optimize (or how) based on hunches. We rely on the results of continual test design, execution, and analysis—whether it’s as simple as an updated checkout flow or as complex as a multi-site pricing rollout. 

Even strongly held assumptions don’t get special treatment. Our commitment is to optimizing, not simply assuming what works and never changing. 

Transitional sentence - here’s why we did it and what we learned. 

Is proration an outdated assumption?

Most parking operators charge for an hour and pocket the difference if you leave early. We don't, and we never have. We prorate every session to the minute because that’s fair. 

The non-prorated model of charging drivers, which requires them to guess how many hours they need and pay upfront or for a full hour, is a holdover from mechanical parking meters created in the 1930s. It’s a skeuomorphism that nobody in the parking industry has questioned. 

When AirGarage started, we assumed that prorating down to the minute was an obvious upgrade from the industry norm, and we built our pricing around that assumption.

Proration has a real cost, though. It complicates billing, it creates edge cases in customer support, and by design it can reduce revenue per session. If drivers weren't noticing it or responding to it, we were paying for a feature that wasn't earning its keep.

What we wanted to learn.

There are basically three ways to handle an hourly parking session that ends early:

• Charge full price and keep the difference. What most competitors do.
• Prorate the difference down to the minute. What we do.
• Charge full price and credit the next session. What we wanted to test.

Our Ops team wanted to know whether proration was really changing driver behavior, or whether we were refunding drivers who didn't care. Three things were unclear:

• Whether drivers even realized proration was happening
• Whether proration was affecting their satisfaction or their likelihood of coming back
• Whether we could improve retention and revenue at the same time by charging the full hourly rate but crediting drivers for unused time on their next visit

Hypothesis and method.

Our hypothesis was that crediting the next session would increase return visits without hurting driver satisfaction. We also wanted to understand whether non-prorated billing, the industry norm, would hold up on sentiment if it turned out drivers didn't actually care about proration one way or the other.

Our decision ultimately rested on the following outcomes:

• Retention: how many drivers came back
• Sentiment: how driver reviews changed and why
• Revenue: gross and per transaction revenue

The test ran across three locations over eleven weeks and captured 3,414 unique drivers. Each driver was assigned to one of three groups: prorated hourly billing as the control, non-prorated hourly billing, or non-prorated billing with an account credit for unused time.

What we learned.

Learning: Credits backfired, causing a drop in repeat visits. 

• First-time drivers in the credit group repeated at 13.6% compared to 21.7% in the control group.
• Telling a driver "you paid for more time than you used" only called attention to an overcharge, and it was not perceived as a perk. This finding is supported by Prospect theory, which posits that the pain of a loss is roughly twice as intense as the pleasure of an equivalent gain. 
• Based on the data gathered during our experiment, 92% of the credit value was issued to drivers who never returned.

Learning: Proration creates the best first impression.

• Prorated first-time drivers repeated at 21.7% compared to 16.6% for non-prorated drivers. 
• This finding didn’t quite meet statistical significance, but it was directionally strong and consistent across all three locations. 
• Interestingly, proration made no measurable difference for drivers who had previously parked with us.

Learning: Non-prorated billing makes more money but hurts the experience. 

• Non-prorated billing produced 13% more gross revenue and an additional $1.40 per driver.
• Unsurprisingly, driver sentiment dropped as a result. One-star reviews doubled, from 9.7% to 19.5%, and the "too expensive" tag (a rating feature) appeared three times as often. 
• While 13% is a desirable revenue lift, the effect on satisfaction could also meaningfully lower driver lifetime value, making it more harmful in the long run.

So is proration the right answer?

We don’t want to end on a cliffhanger, but the short answer is that it depends. Your driver mix, asset type, and whether you’re optimizing for short-term revenue or long-term value all play a role in determining whether proration can help. For now, we're pulling credits, keeping proration, and running a larger follow-up that controls for the price difference between prorated and non-prorated segments.

Though our initial results didn't give us a clearly universal answer, they did clarify questions for the next round of learning. Ultimately, that’s the process of testing. It’s slow, painful work to peel back assumptions and surface a solution that works for everyone involved: 

• Drivers enjoy best-in-class experiences
• Partners increase revenue
• We strengthen our operating model 

When those incentives line up, everyone wins. And the findings can benefit more than the facilities where tests were run. This is the compounding effect of running experiments across our network and making changes everywhere the results should apply. 

The potential gains from finding a new way is why we're willing to put even our most cherished assumptions on the chopping block.

Testing, learning, and iterating is just one part of what separates a modern parking management approach from a traditional one. We break down those differences in more detail in The 5 Levels of Parking Management.