Prove It Before You Touch a Single Rate

Raising prices, or even just restructuring how you present them, is one of the riskiest levers in Self Storage. Get it wrong, and you spook customers, tank conversions, and spend months second-guessing the decision.

It’s also the lever with the most upside. A rate change costs nothing to make. There’s no fit-out, no new signage, no extra staff. Whatever it adds falls almost entirely to the bottom line, and whatever it costs you comes off the same place. That asymmetry is exactly why it’s worth being careful with, and exactly why so many operators put it off.

The fix isn’t guessing harder. It’s testing properly, before the change goes live everywhere.

You’re not bad at this. Everyone is bad at this.

There’s a comforting piece of research here, and operators rarely hear it. Microsoft ran thousands of controlled experiments across its products and published what happened. Only about a third of the ideas tested improved the metric they were designed to improve. Another third did nothing at all. The final third actively made things worse. These weren’t random ideas; they were changes proposed by product experts who did this for a living. Google and Bing report an even lower hit rate, somewhere between 10 and 20% of experiments producing a positive result. Netflix works on the assumption that roughly 90% of what they try is wrong.

The point isn’t that these companies are incompetent. It’s that human beings, including very experienced ones, are poor at predicting how other human beings will behave when a price or a layout changes. Ronny Kohavi, who built Microsoft’s experimentation platform, has made the same observation for two decades: teams are consistently bad at assessing the value of their own ideas, and expertise doesn’t fix it.

So if you’ve ever changed a rate, watched enquiries wobble, and quietly wondered whether it was the right call, that’s not a knowledge gap. That’s the normal condition of anyone making pricing decisions without a control group.

The famous example nobody in your team would have backed

In 2012, a Microsoft engineer suggested a small change to how ad headlines displayed in Bing search results. It sat in the backlog for more than six months. Nobody rated it. It looked like a cosmetic tweak. When someone finally built it and ran it as an experiment, revenue jumped 12%. That single change was worth more than $100 million a year in the US alone, and it didn’t hurt the user experience at all.

The lesson for storage isn’t “go change your headlines”.  It’s that the size of a change has almost nothing to do with the size of its result. The way a rate is framed, what sits next to it, whether the promo is shown as a dollar figure or a percentage, whether the cheapest unit is listed first, these are all the sorts of changes that get dismissed as cosmetic and then turn out to move real money.

You can’t tell which ones from a meeting room. You can only tell by running them.

Why “we tried that last year” isn’t evidence

Most operators have already run what they consider a pricing test. They put rates up 8% in March, watched move-ins for six weeks, and formed a view.

The problem is there was no control. Between March and April, your market also changed. A competitor down the road ran a promo, or didn’t. Google shifted how it ranked local results. It was a wet month, or a school holiday period, or the local rental market moved. You weren’t comparing your new price against your old price. You were comparing your new price against a version of the market that no longer exists.

A proper test removes all of that. Both variants run at the same time, to the same market, under the same conditions, with traffic split between them. Anything happening in the outside world hits both sides equally. The only difference left is the one you introduced, which means the difference in results is attributable to it.

Why this matters more in storage than most industries

Three things make storage pricing unusually hard to guess.

1. Price sensitivity isn’t uniform. A customer looking at a 3m x 3m unit for a house move behaves differently to a business taking a small unit for stock.
2. What reads as good value at one facility reads as expensive at another five kilometres away. A single rate decision applied across a portfolio is really a dozen different decisions bundled together, and they won’t all go the same way.
3. Occupancy and price interact constantly. Sitting at 92% occupancy usually means you’re underpriced, and sitting at 78% doesn’t automatically mean you’re overpriced. Static rates can’t respond to either. Operators who are still running fixed street rates are competing against those who adjust weekly.

And the booking journey does a lot of the selling. In storage, most of the pricing conversation now happens online without a human involved. How the rate is presented, what it’s compared to, what’s bundled with it, and where it appears in the flow all shape whether someone books. All of that is testable, and quite often you can lift revenue without changing the rate at all.

That last point is worth sitting with. The safest pricing test is frequently one where the price doesn’t move.

The 5-step framework

Here’s how it works in R6 Automate’s Experiments tool.

1. Set your control. Start with what customers see today: your current pricing presentation or rate. This is Variant A, the benchmark everything else is measured against. It’s not the “old” version; it’s the reference point, and it keeps running for the life of the test.
2. Build your challenger. Create Variant B: the new price, structure, or presentation you want to test. Change one thing. If you move the rate and restructure the tiers and reword the promo at once, a win tells you nothing about which part won, and a loss tells you nothing about what to fix. Nothing goes live for everyone yet. It runs alongside the control.
3. Set your guardrails. Before launch, define the boundaries: which devices are included (desktop, mobile, or both), how traffic splits between variants, the minimum uplift you need to see before it counts, which facilities take part, and the schedule it runs on. Deciding the minimum uplift in advance matters more than it sounds. It stops the test becoming an argument about whether 1.4% is meaningful after you already know which variant produced it.
4. Let real customers decide. Once live, visitors are automatically and evenly divided between both experiences. You’re not surveying opinions or running a focus group, and there’s a reason that distinction matters: what people say they’d pay and what they actually pay are different numbers. You’re watching real booking behaviour instead. Traffic to each variant, conversions achieved, and conversion rate are tracked until the result reaches statistical significance. Resist the urge to call it early. A variant that’s 20% ahead on day three is usually just noise, and stopping a test at the moment it looks best is the single most common way operators end up rolling out a change that does nothing.
5. Promote the winner. When a clear winner emerges, roll it out to 100% of traffic with a single click. No rebuilding the booking journey, no manual rollout across sites. Just a data-backed decision, live instantly. Then start the next one.

Where to start

If you’ve never run one, don’t open with a rate rise. Start somewhere the downside is small, and the learning is useful: how the promo is framed, the order units appear in, whether you show a monthly rate or a weekly one, whether a “most popular” marker changes what gets picked.

Give it enough traffic to mean something. A site with a handful of bookings a week will take longer to reach a reliable answer than a busy portfolio, and that’s fine; it just needs to run longer rather than be judged sooner.

And expect the wins to be modest. Most successful tests shift results by less than 20%. The operators who get real value out of this aren’t the ones who found one enormous win; they’re the ones who ran twelve tests a year, won four of them, and compounded the results.

Worth knowing: fewer than 1% of websites run any kind of controlled testing. Almost nobody in this industry is doing this properly, which is precisely why it’s still worth doing.

That’s five steps, and not one of them involves changing your rate before you know it works. Run the experiment, learn from the behaviour, implement the winner. Then test the next idea and repeat.

To find out more, chat to the R6 Automate team today.