Why your puppies aren’t selling
A health-testing breeder with a litter still sitting at ten weeks is almost never doing anything wrong with the dogs. In my experience the problem is upstream of the dogs, in the ten minutes a buyer spends before they ever write to you. I went and measured those ten minutes. Here is what I found, including on my own kennel.
First, what this is not
If you typed that phrase into a search box, most of what came back was about puppy mills — pieces written for buyers, warning them away from people like you. This is not that. This is one breeder writing to another about the specific, unglamorous reasons a well-bred litter sits, and none of them are about the dogs.
The market has not collapsed. It has split. The programs that health-test and hold a waitlist are still placing at full price; the volume end is sitting on unsold litters. The uncomfortable part is that a serious program can end up sitting alongside the volume tier for reasons a buyer never says out loud — because the buyer never got as far as talking to you.
Where the buyer went before they found you
A buyer used to type a breed and a state, open a dozen kennel websites and email three of them. Now many ask an AI assistant, or open Good Dog, and have a shortlist before they have read a single breeder’s page. The assistant answers with names. And you cannot see the answer it gave.
That last part bothered me enough to go and look. On 2 September I asked three assistants — Claude, ChatGPT and Perplexity, each with live web search on — the five questions a buyer actually types, in twelve states. That is 180 answers, and I read every source they used to write them: 5,076 citations across 919 websites, an average of 28.2 sources per answer.
Two things from that reading matter here. The assistants do name breeders, freely; the anxious version of the question is settled. And the shortlist they name is built from a very short list of websites. Ten sites supplied 1,809 of the 5,076 sources. One of them, gooddog.com, was quoted 571 times in 84 of the 180 answers. The American Kennel Club was named in the text of 146 of the 180 — about four in five. The full count is in the Breed Study; the short version is that whether you are named is decided by a handful of sources, most of which are not you.
The six reasons, and what I could measure on each
1. The buyer asked somebody else first, and the answer named somebody else
This is the one you cannot see from your own chair. A buyer in your state asked which breeders do OFA testing, got a list of six names, and wrote to three of them. You were not on the list, so nothing happened, and nothing happening leaves no trace. A quiet inbox looks the same whether nobody is buying or whether nobody was told you exist. The only way to know which it is, is to ask the same question the buyer asked and read what comes back.
2. Health testing you assert but do not publish, dog by dog
“All our breeding dogs are fully health tested” is a sentence. A dam’s registered name next to her hips, elbows, eyes and heart, with the clearance numbers, is a fact. A buyer can take either on trust. A machine can only quote the second one. When I read the sources the assistants leaned on, the pattern was the same every time: the sites that get quoted state a specific fact about a specific named dog or breeder in a specific place, in a form that can be lifted out without interpretation. A paragraph of reassurance is not that, however true it is.
3. The browser-tab title with no town or state in it
Every one of the five buyer questions has a state in it. Open your homepage and read the words in the browser tab. If it says Home, or your kennel name and nothing else, then the one line a machine reads first says nothing about where you are. Buyers do not read that line. Everything that assembles a shortlist for them does.
4. The waitlist form that goes to a dead inbox
I would not have believed this one if it had not happened to me. A form on one of my own sites was silently dropping submissions this summer, and I found out only because I sent a test through it. No error on the buyer’s side, no error on mine. From where I sat it looked like a quiet month.
5. Pricing that reads as the other tier
Buyers read a price as a signal before they read it as a cost. A figure set well below what the health-testing programs in your state ask does not read as generous. It reads as the volume tier — the tier every puppy-mill article just told them to avoid — and a serious buyer moves on without asking why. Underpricing good dogs does not make them easier to place. It makes them harder to trust.
6. Photos with no dates on them
A page of puppies with no dates reads as a page that always has puppies. That is the volume signal again, and a careful buyer picks it up in seconds. A dated litter page — this dam, this sire, whelped on this day, these clearances — reads as a program.
What I found on my own kennel
Our programs were in that study on the same footing as everyone else’s. Nothing told the assistants we exist. The result:
- 3 of the 180 answers named one of our programs. All three were Oregon answers.
- 6 answers cited our own pages as a source — 15 citations across 8 different pages of ours — and those six answers were in Oregon, California and Ohio.
Read those two lines together. Our Cavalier health-testing write-ups were being quoted to buyers two thousand miles away as reference material, and nothing anywhere recommended the kennel that wrote them. The pages were good enough to be an authority and not legible enough to be a breeder. That gap was open on my own sites first, and it is why the six items above are the six items.
As for our litters: they are usually fully committed before whelping. Occasionally one or two puppies are still unspoken for on whelping day — that is the honest version. No puppy has stayed available past eight weeks in the last two years.
I am not going to tell you the website did that. I do not know that, and anyone who tells you they know is guessing. What I can tell you is what the machines did with the pages, because I measured it, and that the pages the machines read are the same pages a buyer reads.
What to check tonight
Five things. Ten minutes, phone in hand, no login needed.
- Read your browser tab. Open your homepage and look at the words in the tab, not on the page. Do they contain your breed and your state?
- Send yourself a waitlist enquiry. Fill in your own form from your phone, as a stranger. Did it arrive, and where?
- Pick one dam. Can someone who has never emailed you find her registered name and every clearance, with the numbers, on one page of your site?
- Find your newest photo. Can a buyer tell when it was taken, and which litter it belongs to?
- Ask one assistant the buyer’s question. Open ChatGPT, Claude or Perplexity with search on, and ask: who are the best [your breed] breeders in [your state]? Read who it names. Then read the list of sources it used, which is usually the more interesting half.
If number five comes back with your name in it, good — that is real. If it does not, you are in ordinary company, and now you know which kind of quiet your inbox is.
If you would rather I ran number five properly, I will. Three assistants, the five buyer questions, your breed and your state, and every transcript sent back to you word for word, including the answers where you were not mentioned. It costs nothing, there is no score and no grade, and I do not phone people who have not asked me to. It is described on the free check page.
Mark Griffin runs Direct-Me LLC and, with his family, a boarding and daycare resort, a Golden Retriever program and a Cavalier program in Oregon. He also sells marketing services to breeders, which is a plain conflict of interest and the reason the counts above are stated with their method rather than summarised. Every figure on this page comes from the 180 recorded answers described in the Breed Study, where the method and its limits are set out in full.