The end of the filter panel: how renters actually search

Almost nobody who lands on a rental site is looking for a category. There is a birthday on Saturday, twenty children are coming, the yard is fifteen by fifteen, and everything has to be packed up by six in the evening. There is no way to express that in a filter panel — and even if there were, nobody fills in four separate dropdowns. What follows is the scene everyone knows: three tabs open, two of them say “call for availability”, the third goes empty as soon as a date is picked, and the customer reaches for the phone.

A filter panel is really a translation interface. It asks the customer to translate the day in their head into fields a database understands — and it makes the customer do the translating. In rental that job is unusually hard, because the same product is filed under a different name at every company: “combo bouncer” here, “inflatable play set” there, just “castle” somewhere else. Not knowing which word to type, the customer clicks a category, finds a hundred products, and still has no idea which of them is free that day.

A sentence carries more than a form

“I need a bounce house for 20 kids in Garland this Saturday” says more than a form could hold. There is a date in it (this weekend), a place (Garland and around), a capacity (twenty children, not twenty adults), a category, and an unstated but obvious expectation: delivery. An assistant’s job is to pull those apart and turn them into a query the inventory understands.

The distinction that matters: the assistant does not merely understand the sentence, it turns that understanding into a real query. The category resolves from the catalogue’s taxonomy. The capacity is read from the product’s required attributes. The date goes against actual calendar occupancy. The price comes from the tariff. So the answer is not a guess — it is what the database says right now.

The result list should be short, not long

Classic search measures its own success in result count: “412 results found.” For someone with a birthday party on Saturday, 412 results is not an achievement, it is a chore. The assistant does the opposite: it narrows first and leaves three options standing. All three are genuinely free that day, all three fit twenty children, all three deliver to that address.

The “3 of 442” line above the cards is deliberate. It tells the customer not that the inventory is small, but that the narrowing has been done. That is where trust starts: a short list means either very little stock or somebody looked on your behalf. You have to show it is the second.

Comparison, without three tabs

People open three tabs because they want to compare. How big is the price gap, which one is larger, which delivers free, which lets me cancel 24 hours out. Those four questions sit inside almost every rental decision, and no product page ever puts them side by side.

Having the assistant build the comparison is not only convenient, it is honest. If the table says “delivery: $45”, that number comes from the tariff; if it says “free cancellation: 48h”, it comes from the cancellation policy. Not from marketing copy.

Detail is where the decision happens

Once the options narrow, people go into detail and always look at the same things: how many people, how many feet, which socket, how long is setup, what is the total. If those five answers are not written on the product, the customer picks up the phone — and that phone often rings unanswered.

Showing the deposit inside the breakdown matters. In rental, the surprise in the total is almost always the deposit; when it appears at the payment step, drop-off climbs noticeably. Writing it up front does not delay the cancellation — it prevents it.

The renter is not searching for a category; they are trying to solve a day. The interface’s job is to translate that day into inventory.

What this asks of your catalogue

This approach has exactly one cost, and it is not technical — it is data discipline. The assistant can only filter on what has been written down. A product with no capacity gets dropped by the sentence that says “twenty children”. A product with no dimensions cannot safely be offered to someone whose yard is 15×15. With no setup time, “packed up by six” cannot be answered at all.

• Capacity — headcount, children or adults depending on the category
• Footprint — whether it fits the space it will stand in
• Power requirement — how many sockets, which voltage
• Setup and teardown time — it belongs to the customer’s plan
• Delivery coverage — which districts, at what fee

None of this is new information; it is what you already say on the phone. The only difference is that once it is in the form, it is answered before anyone asks.

And when the sentence is not clear

Not every sentence is precise. Someone who writes “we need something for the weekend” has not decided yet; someone who says “for 20” has not said whether those are adults or children. At that point there are two bad options: show results on a guess, or greet the customer with a form. Both land in the same place — lost trust.

The right behaviour is the third one: ask for the missing piece with a single question. If the date is vague, ask about the date; if the capacity is vague, ask about the capacity. One question, because three in a row make the customer feel they are filling in a form after all, and the whole advantage of a conversation is gone.

The no-results case deserves the same honesty. If nothing is free on the requested date, the right answer is “I have nothing for that day”, followed by a concrete alternative: the day before, a nearby district, or a smaller unit. The only thing as damaging as an empty list is describing a product that does not exist.

The name for this behaviour is simple: the assistant only states products, prices and dates that have a counterpart in the catalogue. An invented name or a rounded-off price may please the customer in the moment, but it comes back to you on delivery day.

One thing that is easy to measure

You do not need an elaborate dashboard to tell whether this is working. Watch a single number: how many customers reached a booking without calling. The expensive part of the filter panel was never its design — it was that it forced the customer onto the phone, and that the phone rang after hours.