Why we won't build an 'available now' feed

Every few days someone writes to ask for the same thing. Could we add a filter for puppies that are ready to go? A tab for what is available this week? An alert when something new appears?

The people asking are not chancers. They are usually families who have been looking for months, who have been polite to breeders who never replied, and who would quite like the search to have an end. It is a reasonable thing to want, and we are not going to build it. It seems only fair to explain why, because refusing an obvious feature without a reason is just stubbornness.

What availability actually measures

Start with what the word tells you.

A puppy that can leave this week is a puppy without a home already arranged. In the part of the market we are trying to support, that is an unusual state of affairs. A breeder producing one litter a year, from a bitch they have health-tested and waited for, has generally spoken to more people than there are puppies long before the puppies exist. Homes are matched during the pregnancy, sometimes before the mating. By eight weeks the litter is spoken for and there is nothing to advertise.

So availability, as a live property of the market, is not evenly distributed across breeders. It concentrates. It is produced reliably and in volume by people whose business model depends on producing it — because a puppy sitting unclaimed at nine weeks is, for them, a cost, and the whole apparatus of ready now, viewings this weekend, last one left exists to clear that cost quickly.

That is not a moral point about every seller with a puppy free. Litters are bigger than expected. Buyers pull out; families' circumstances change a fortnight before collection; a good breeder can absolutely find themselves with a puppy and no home for it. But those are exceptions arriving one at a time, and a feed does not sort by exception. A feed sorts by supply, and supply belongs to whoever makes the most of it.

The sort order is the product

There is a thing that happens to marketplaces, and it happens whether or not anyone intends it.

Whatever you sort by becomes what sellers optimise for. Sort by price and you get a race downwards. Sort by response time and you get sellers who answer instantly and think later. Sort by availability and you get more availability — which in this trade means more puppies bred speculatively, without a home in mind, on the calculation that the feed will find one.

We have been careful about this elsewhere. We do not rank breeders, because a ranking is an instruction about what to become. An availability feed is the same instruction wearing different clothes, and it is aimed at the one behaviour we would least like to encourage. The breeders who plan two years out, who turn away four buyers in five, who will hold a puppy back rather than place it wrongly — those are precisely the people who would never appear in it. We would have built a product whose front page systematically excluded the breeders it exists to find.

The fortnight problem

There is also the buyer to think about, and here I am less certain of my ground, because I am arguing against a thing people genuinely want.

I wrote yesterday about what happens after a breeder turns you down — that the two or three weeks afterwards are when people make the worst decisions of the whole process, because the emotional decision has already been taken and only the dog is missing. An availability feed is a machine for meeting people at exactly that moment with exactly the thing they should not act on. It would be, without any exaggeration, the most useful feature on the site for anyone whose interests run against the buyer's.

I do not think a marketplace can wash its hands of this. The design decides who gets found and when, and a design that surfaces immediacy to someone in a hurry is making a choice on their behalf, whatever the disclaimers say.

What we are building instead

The honest answer to the underlying problem — I have been looking for months and I don't know where to start — is not a feed. It is planning.

So the things we are working on point the other way. Listings that say when a breeder expects to breed next, rather than only what exists today, so that a buyer can join a list eleven months out and know they have. A way to register interest with a breeder that does not require a deposit and does not commit anybody to anything, because a deposit is not the right instrument for expressing interest and never was. Breed-level views, so that someone drawn to a numerically small breed can see the handful of people breeding it properly and understand that the wait is the norm rather than a personal misfortune.

None of that is as satisfying as a page of puppies with faces on it. It is slower and it converts worse and it will not trend. But it describes the market as it actually is for the breeders we want on the platform, and it gives a buyer the one thing the available-now page cannot: a reason to still be pleased in nine years.

The trade we are making

I want to be plain that this costs us something. Availability is the highest-intent search there is. A feed would bring traffic, and the traffic would be people ready to act. We are choosing not to have it, and there will be months where that choice is visible in the numbers and invisible in everything else.

We are betting that a marketplace which is honest about waiting will eventually be trusted more than one that promises not to make you. If we are wrong about that, it will be a slow and quiet sort of wrong, and it will still have been the right thing to build.

In the meantime, if you write to us asking for the feature, you will get a version of this note back. It is not a no on principle. It is a no because we can see, fairly clearly, what the feature would do.

— Rene