I publish numbers for a living. 183M+ Google impressions across client campaigns. 27M+ AI citations. Those figures are true, and I am going to argue that by themselves they tell you almost nothing about whether PR worked for any one person.
That is the reason so many owners finish a PR engagement and say it was a waste of money. Often the work was fine. The scoreboard was broken. Nobody agreed up front on what a win looked like, so the report arrived full of numbers and the owner felt nothing change.
My position: PR feels wasted when you count activity, and it feels worth it when you count a change in what a stranger finds and believes about you. Here are four counts I would stop trusting, and what I would use instead.
1. The clip count
A report that says "12 placements this quarter" sounds like progress. Twelve of what? A trade title your buyers read, or a site nobody visits? A page that names you in the first paragraph, or one that lists you in a roundup of forty names?
Volume is the easiest number to produce and the hardest one to feel. I wrote about the same trap from the publisher side in Why one great article beats ten forgettable ones. A count treats every mention as equal, and readers do not.
Ask for the list of pages instead of the total. Open five of them. If you would not send the link to a prospect, the count is padding.
2. Ad value
Some agencies still convert coverage into the cost of buying the same space as an ad, and report that dollar figure as the result. The PR measurement industry walked away from this years ago. The Barcelona Principles, first agreed in 2010 and updated to version 3.0 in 2020, state that advertising value equivalents are not the value of communication, and the industry body behind them, AMEC, does not accept campaigns evaluated that way into its awards, according to a summary from the Open University.
Some practitioners still use ad value as one input inside a larger analysis, and I will not claim the debate is closed. As a headline result, though, it answers a question nobody asked. A reader who trusts you more after a feature has not paid you a cost of ad space. They have changed their mind about you.
If a proposal promises a big dollar figure for coverage, ask what decision it helps you make. Usually there is none.
3. Impressions
I keep 183M+ in my own materials, so I am not going to pretend impressions are worthless. They show that a page was served and had a chance to be seen. They say nothing about whether the right person saw it, read it or acted.
Treat impressions as a floor. They tell you the work reached someone. Then ask what those someones did that they had not done before.
4. The one big name
The mirror image of the clip count is a single logo. A famous masthead feels like proof of a good engagement, and a PR firm can point at it for a year. I covered that pull in Stop asking for Forbes and start asking for a reason. A logo is one data point. It tells you almost nothing about whether anything around it changed.
The count I use instead
Pick one question that can be answered before you start and again ninety days later. Mine is this: when a stranger searches for you, or asks AI about you, what do they find, and does it match what you want them to believe?
That question is checkable and it belongs to you, not to a vendor.
Here is a small example from my own work. For one client we saw 253.5K AI citations in 60 days, starting from zero. Look at that sentence closely. It has a starting point, a window and an end point. It says what the baseline was. A number without those three parts is a decoration, and I try not to publish my own numbers without them. When I see a PR report with no baseline, I assume the author does not have one.
So the structure I want in any engagement, mine or anyone else's, is a baseline, a window and a named change.
- Baseline: what a search for your name and ask AI about you return today. Save screenshots with the date.
- Window: a fixed stretch, usually 60 to 90 days. Anything shorter is noise. Anything with no end date cannot fail, which is why vendors like it.
- Named change: one or two specific things that should be different. A fact that now appears on a credible page. A description of your business that is accurate where it was wrong. A mention on a page a buyer would read.
Why owners get burned
I have a theory about why this keeps happening, and it is partly the owner's fault. Buying PR feels like buying a result, so people skip the part where they define it. Agencies are happy to fill that silence with whatever number looks best. Nobody is lying. They are answering a question that never got asked.
There is also a patience problem I will own. Press is a slow asset. A feature does not pay out in a week. It sits on a credible page and keeps doing its job whenever someone looks you up, which I described in Old press about you still shapes what AI says today. If you judge it by next Friday, nearly all PR will look like a waste. Judge it by what a stranger finds in ninety days and the picture gets clearer in either direction.
That cuts both ways, and I want it said plainly. Some PR is a waste. A report with no baseline, no named change and no link you would be proud to share is a bad sign, and you are allowed to walk away from it.
Do this before you spend another dollar
If you are running PR now, or thinking about it, take thirty minutes this week.
- Search your name and your business name in a private window. Screenshot page one.
- Ask ChatGPT or Claude who you are and what you do. Screenshot the answer.
- Write down the two things you wish a stranger found. Be specific enough that a friend could check them.
- Put a date ninety days out in your calendar and repeat steps one and two.
Then show those four items to anyone who sends you a PR report. Their report either explains the gap between your first screenshots and your second ones, or it has nothing to say about it. That tells you what you needed to know.



