Your reviews already told you what's wrong. The information is sitting in your own account, free, timestamped, and written by people who watched it happen. Most owners never see it, because they check the number at the top and close the tab.
What do my reviews tell me about my business?
The number is the worst thing on that page.
Why the average can't warn you
A star rating is the mean of every review you have ever received. That's what makes it feel official, and it's the same thing that makes it useless as an alarm.
Averages are slow on purpose. Say you run a dental practice with 180 reviews at a 4.6. Your front desk loses a person in March, the schedule starts slipping, and ten people write one-star reviews about waiting forty minutes past their appointment. That drags you to a 4.4. You would need a genuinely catastrophic spring for the number itself to say anything, and by the time it does, you have stopped diagnosing a problem and started managing a reputation.
The text has no such lag. The first customer who writes "the front desk seemed overwhelmed" writes it the week it starts.
So the most valuable number on your listing is the one nobody watches. It's the date.
Complaints are data wearing a costume
Read fifty reviews in order and they sort themselves into a handful of causes, whether or not anyone involved meant to file a report.
"Waited forty minutes past my appointment." "Third time they've rescheduled on us." "Nobody on the floor, gave up and left." "Called four times, never got a person." Those come from a dental office, a lawn care crew, a hardware store, and a body shop. They are the same review. Every one of them describes a business trying to serve more demand than it has hands for, and not one of them uses the word staffing.
A customer who writes that the front desk seemed overwhelmed is reporting a ratio, whatever they think they're doing. They lack the vocabulary for it, so they describe what it felt like from the waiting room, which is the only instrument they have.
This is the part worth sitting with. Your customers are running a continuous, unpaid audit of your operations. They are terrible at naming what they found and very good at noticing it, and they publish the results where every future customer can read them.
The date is the diagnosis
Sort by date and the reviews stop being feedback and turn into a chart.
Somewhere in the run, the vocabulary changes. For two years people wrote about the work. Then, in some particular month, they start writing about the wait. Or the phone. Or that it's "not what it used to be," which is the most useful sentence a customer ever writes, because it means they have a baseline and you fell off it.
That month is the diagnosis. Something inside the business changed, and it changed shortly before the language did.
Usually it was a decision you made on purpose and for a sensible reason. Payroll looked high in February, so you cut a shift. Somebody left and you decided to cover the gap yourself rather than rush a hire you would regret. You dropped the Tuesday deep clean because nobody was complaining about Tuesdays. Each of those was defensible on the day you made it. The reviews are the receipt, and they arrive two months later addressed to somebody else.
Here is the part that goes unnoticed. None of those decisions removed any work. Cutting a shift doesn't cut the dishes, the ordering, the callbacks, or the floor. The work stays exactly where it was. It just moves onto whoever is left, and the person who is left is usually you. The cost didn't leave the business. It left the books.
That's why the payroll line can improve while the business gets worse, and why the first place the move becomes visible to anyone is a stranger's review about a wait.
You already have the instrument
There is a version of this where the fix is expensive. It usually involves buying a dashboard that turns your reviews into a sentiment score, which is a number about your numbers, several steps further from the thing that actually happened.
You don't need it. Open your Google profile, sort by newest, and read a year backward. It takes a Sunday afternoon. Read the words, watch for the month, and go find out what you did that spring. Nobody sells that because nobody can charge for it.
The reason this works is that a small business is far more legible than its owner thinks. Capability is cheap now and legibility is scarce, and most owners assume theirs is the scarce kind, unreadable from the outside without an expensive translator. Meanwhile the business has been publishing a public monthly report on its own internal condition, in plain English, for years, written by volunteers.
Where the reviews go quiet
The reviews will show you that the wait got long in March. What they leave out is how many people quietly stopped coming, what those people were worth over a year, and whether you could have served them if they had shown up.
That arithmetic needs numbers that live inside the business rather than on the listing, and it is a different job with a different set of tools. But the diagnosis half is free, and it has been sitting in your account this whole time, waiting for somebody to read past the first line.
Start with the month the language changed. You already know what happened that month. You just haven't connected it to the number you've been watching instead.