Mirror reads what you promise and what people live, in public. Then it shows you exactly where the two part ways. Every finding is verified against its source.
“The less obvious vulnerability is not whether people find the staff caring; it is whether that care survives a handoff, discharge or administrative problem.”
Blind spot
Help after the bill. Surprises before it.
The hospital says
Free financial screening after the bill
Reviewers say
Unexpected hospital-level charges
Strength
Personal attention. Publicly confirmed.
25/33
public mentions of staff are positive
Claim to prove
Praise is not proof.
“World-leading specialists”
Praise found · comparative outcomes not found
Peers
Who answers. Who doesn’t.
This hospital18.6%
Peer A86.7%
Peer B0%
Share of public reviews with an owner reply
Action
Every difficult discharge. One named owner.
Confirm mobility support, transport, the receiving service and the next contact. Then close the handoff.
From a real Roadyom Mirror reading of a California hospital, October 2026, built from public sources. We don’t name the hospital or its peers.
The reading
The real thing. Open it now.
No sign-up. No sales call. This is a real reading, as a customer receives it. Every section, every figure.
Enter your website and email. Mirror does the rest. No meetings, no questionnaires, no waiting.
01
Collect
Your website and channels. Public reviews, videos and comments. News coverage.
02
Read
Every text is read three times. Only what holds each time stays.
03
Cite
Every finding links to its public source.
04
Deliver
A private link and an email, usually in about 30 minutes.
Behind every review is a person who walked through your door.
Mirror listens at scale. And leaves every name out.
The method
Two mirrors. One truth.
Your image isn’t what you say about yourself. It isn’t what one angry review says. It’s the distance between the two. Mirror measures that distance, automatically, in every reading.
Expression
What you say
Your website
Your own videos
Your stated claims
The gapblind spots · claims without proof · strengths you under-use
Perception
What people live
Public reviews
Videos by others and their comments
News coverage
Inside the reading engine
Read like a researcher. Checked like an auditor.
3×
independent reads of every text
The majority holds. When the reads disagree, the text is set aside as unstable. Nothing is guessed. The safeguard comes from published research on machine annotation.
10
dimensions of institutional life
Service, staff, safety, access and waiting, price and value, trust and ethics, expertise, communication, leadership, community. Every text lands where it belongs.
9
kinds of speech told apart
Praise is not promotion. A question is not a complaint. A lived experience is not an opinion. Sarcasm is detected and quoted, so irony isn’t read as praise.
100%
of quotes matched to the source
Every quote behind a finding is checked against the collected public text. Evidence that doesn’t match is removed. A finding left without evidence is dropped.
95%
confidence interval on every share
Positive, negative, mixed. Every figure carries its margin of error. A peer difference counts as real only when the data supports it.
30
stable outside voices, minimum
Below that, the reading says INSUFFICIENT EVIDENCE. It never invents a verdict.
Every finding. Traced to its source.
No black box. If public evidence can’t support a point, it isn’t in the reading. When the sample is thin, the reading says so.
Weakness · real reading
A difficult exit can undo a caring encounter.
Two reviewers describe discharge and referrals between locations left without an owner — where the hospital promises connected care.
Quoted public reviewsEach quote linked to its source
Sample read 3 times · 86 public items · 1 unstable item set aside
Peers · real reading
The hospital read58.1%
Comparable hospital A76.5%
Comparable hospital B30.8%
Positive share of public perception · real reading, October 2026
Know where you stand.
Mirror proposes comparable institutions and reads them the same way. Every figure carries its margin of error. So you see a real difference, or that there isn’t one yet.
Not a snapshot. A monthly mirror.
On Monthly and Annual, Mirror reads again every month. You see how your figures moved: positive and negative share, low-star reviews, replies. You see whether each shift goes beyond the margin of error. And every earlier reading stays at your link.
Month 1①First reading
Month 2②What changed
Month 3③History kept
A real reading
One hospital. Seventeen minutes.
What Mirror read and produced for a hospital in California. Public sources only.
70public reviews
30public videos
86public items read
10findings, each cited
5ranked actions
2peer hospitals compared
“The clearest actionable benchmark: the hospital replies to 18.6% of public reviews; a comparable hospital replies to 86.7%.”
Principles you can check.
Institutions, not people.
Mirror analyzes organizations. It doesn’t track or profile individuals. It doesn’t infer religion, sect or political views.
Public sources only.
No bypassed logins. No private data. No reviewer identities in what we collect.
Honest about limits.
A reading describes patterns in a public sample. It isn’t a survey or a representative measure of public opinion. And it states its sample size.
Choose your Mirror.
Start with a free preview. Choose a plan to open the full reading.
Full reading
$149one time
One complete reading of your institution, kept at your private link.
For a decision, a launch or a board meeting that needs an outside view now.
Prices in USD, before tax. Secure checkout by Paddle. Every payment fully refundable within 14 days. Refund policy · Terms
Questions, answered.
What’s in the free preview?
The verdict and one blind spot in full, each with its sources. You also see how many other findings the reading holds. The preview starts at once when your email is on your institution’s domain. Otherwise it starts after a short check, or the moment you choose a plan.
How is the full reading different from Monthly?
The full reading is one complete reading, today. Monthly gives you a new complete reading every month. A table shows how your measured figures moved since the last one, and marks each shift as clear or within the margin of error. Every earlier reading stays at the same link. Annual is Monthly paid yearly, about two months cheaper.
How long does it take?
Usually under an hour. The real reading on this page took seventeen minutes.
What if there isn’t enough public information about us?
Mirror needs at least 30 stable outside voices. Below that, the reading says so plainly. It doesn’t guess. You see this in the free preview, before you pay anything.
Which sources does Mirror read?
For what you say: your website and your own videos. For what people say: public reviews, videos by others, their comments and news coverage. Public sources only. No logins, no private data, no reviewer identities.
In which language is the reading?
Readings are delivered in English for now.
Can I cancel or get a refund?
Yes. Cancel a subscription anytime from the billing link in your reading. Every payment is fully refundable within 14 days, no conditions.
