Most press is sold by the count. Ten placements, twenty placements, fifty. It is an easy number to put in a proposal and an easy number to put in a report, and I understand why buyers like it. You can see it going up.
I run a company that works with a network of 1,600+ publications, so you might expect me to say more is better. I don't. My view is that one great article beats ten forgettable ones, and that buying by the count is the most common way people waste money on PR. The rest of this post is the argument, and a test you can run on your own coverage this week.
What "forgettable" actually means
A forgettable placement is not a bad one. It is a placement nobody would miss if it vanished tomorrow.
It has your name in it, but the piece is not really about anything. It reads like a lightly edited announcement. It could have been written about your competitor with a find and replace. A reader finishes it and cannot say what you do better than anyone else, or why they should believe it.
Ten of those do not add up to a reputation. They add up to ten pages that say roughly the same thin thing. I wrote about the same trap from the other side in Why I stopped caring about vanity placements. A vanity placement feels good on the day it runs, and then it does nothing for you the rest of the year.
A great article is the opposite. It is specific. It contains a claim, a story or a piece of expertise that is yours. An editor stood behind it. When a stranger reads it, they come away knowing something about you that they could repeat to someone else. That last part is the real test. If a reader could not repeat what they learned, the article did not do its job.
Why this matters even more with AI search
There was a time when the audience for a placement was a person glancing at your homepage. In that world, a long row of logos had a certain effect on a skimmer, and count mattered a bit.
Now the audience includes ChatGPT, Claude, Perplexity and Google AI Mode, and they read the page instead of the logo. I cannot tell you how each tool weighs what it reads, because the companies do not publish it, and I would be suspicious of anyone who says they know the formula. What I can point to is outside research on where the tools get their material. Muck Rack's Generative Pulse research, updated in May 2026, analyzed more than 25 million links from ChatGPT, Claude and Gemini answers. It found that earned media accounted for 84% of citations, and paid and advertorial content for 0.3%.
Muck Rack sells to PR teams, so keep in mind that it has an interest in the finding, and its definition of earned is broad. It does not prove that any single article causes any single citation. But it does say something that matters for the quantity question. The tools lean on what independent publications have written about you. They do not lean on volume you generated yourself.
Think about what that means for ten forgettable pieces. If they are thin, near-identical and say nothing specific, a tool reading them has very little to work with. It gets ten versions of "this company exists." If instead there is one piece where an editor put real words around what you are known for, the tool has something it can actually repeat. I made the larger argument in What 27 million AI citations taught me about trust. A citation is borrowed trust, and thin coverage has little trust to lend.
The number I do not want you to misread
Here is a number from our own work. Across client campaigns, we have seen 253.5K AI citations in 60 days for one client, starting from zero. It is a real result, and I am proud of it.
I am also going to be careful about what it proves, because a number like that is exactly how people get sold on volume. It does not mean that more articles produce more citations in a straight line. It does not mean sixty days is a promise for anyone else. What it shows me is that a client with almost no presence can change quickly once credible sources start saying specific, consistent things about them. The word doing the work there is credible, not many.
I should also say what I got wrong early on. I used to think of coverage like inventory, something you stack up so there is more of it. The longer I looked at what actually moved how people and tools talked about a client, the less that held up. The pieces that mattered were few, and they had one thing in common. A reader could tell who the person was and why they were worth listening to.
A test you can run this week
Pull up every piece of coverage you have, and I mean all of it. Then do four things.
- Read the first two paragraphs of each piece as a stranger. Write one sentence about what you learned about the person or company. If you cannot write it, mark the piece forgettable.
- Check whether it could be about anyone else. Swap in a competitor's name in your head. If nothing breaks, the piece is generic.
- Check who stood behind it. Was it chosen by an editor, or placed by someone with a budget? I covered why that distinction matters in Paid placements are fine. Pretending they're earned isn't. Neither is wrong, but they are not worth the same, and you should know which pile each piece is in.
- Ask AI about yourself. Ask ChatGPT or Perplexity who the best option in your field and city is, then ask what it knows about you. Compare its answer to your best piece of coverage. If it can repeat what that article says, the article is working. If it says nothing, you have learned something more useful than any report.
When you finish, you will likely have a short list of pieces that pass and a long list that do not. That ratio is your real score. The long list is not a loss, since it cost what it cost already. But it tells you where the next dollar should go, and it should not go toward a bigger pile.
If you take one decision from this post, take this one. The next time someone quotes you a number of placements, ask them to show you the one piece you would be proud to send to a stranger. Then ask them to explain why the other nine matter.
If you ran that test on your own coverage today, how many pieces would pass?
