Written by Ankush Gupta
Most companies I work with have two records of why people leave. One is the exit interview file, which is calm, procedural, and full of phrases like “seeking new challenges.” The other is the public review page, which is specific, occasionally furious, and names a manager.
When those two documents contradict each other, HR tends to treat the exit interview as the reliable one because it was collected properly. In my experience that instinct is backwards, and it is the single most expensive assumption in employer branding.
Why the exit interview is structurally unreliable
Nothing about an exit interview is designed to produce honesty.
The person answering still needs a reference. They may want to be rehireable. They are also sitting across from the function that signs off their final settlement, in the same week they want to leave without friction. So they do the arithmetic. Candour has no upside, and everyone in the room knows it.
Timing makes it worse. The exit interview happens at the moment of maximum politeness. The public review usually gets written weeks or months later, after the person has landed somewhere else and has nothing left to protect. That distance is exactly what produces specificity.
So the two documents are not two accounts of the same event. They are one account given under pressure and one given after the pressure lifted.
What the public version actually contains
Read the reviews for a company you know well and you will notice they rarely complain about the thing HR measures.
Almost nobody writes about the benefits package. They write about a specific manager, a promotion that was promised in a review cycle and never happened, a reorganisation nobody explained, or a workload that grew quietly after someone left and was never backfilled. These are operational failures with names and dates attached.
That level of detail is useful precisely because it is uncomfortable. Exit interview summaries roll up into categories like compensation or career growth, which are broad enough that nobody has to act on them. A review that says the same manager pushed out three people in eight months cannot be rolled up.
The pattern I look for is repetition across time. One angry review is a person. The same specific complaint appearing from four people who left in different quarters is a process.
Responses are read as evidence, not as damage control
Most companies treat review responses as a chore, and it shows. You know the reply. “Thanks for your feedback, please contact HR.” That is worse than saying nothing, because it proves nobody read the review.
Candidates notice the difference. Glassdoor puts it at roughly 62 percent of job seekers saying their view of a company improved once they saw the employer reply. Tone is not what moves that number. A specific concern getting a specific answer is.
The responses that work look like this. Acknowledge the particular thing raised. Say what changed, or say plainly that nothing has changed yet and why. Do not dispute the person’s experience, and do not hint at who they are, which is a mistake I still see from companies trying to signal that the review is unfair.
What I tell clients to do with this
Put the exit interview themes and the public review themes side by side for the last two years. Not summarised, quoted. The gap between the two lists is your actual blind spot, and it usually takes an afternoon to build.
Where the public version raises something the exit interviews never surfaced, treat that as a finding about your exit process, not just about the complaint. If leavers do not tell you things they will tell the internet, the interview is not collecting data. It is collecting compliance.
Fix the operational item before touching the review page. Suppression is available and it does not work here, because the same conditions keep generating new reviews. I turn down this work when a company wants the page cleaned while the manager stays.
Then respond to the older reviews as well as the new ones. Candidates read down the page, and a company that engaged seriously two years ago reads very differently from one that started last month.
The uncomfortable part
None of this is a communications problem. A review page is a lagging indicator of management quality, and the writing on it is mostly accurate. That is why cleaning it up rarely holds.
The companies that improved their ratings in my client base did it by changing something specific and letting the reviews catch up, which took months. The ones that treated it as a content exercise are still having the same conversation.
Your leavers already told you what is wrong. They just waited until they were safe to say it.
Author Bio:
Ankush Gupta is the Fractional CMO of FameNinja, an online reputation management agency. He has spent more than eight years across reputation management, digital PR and brand marketing, working with founders and companies on search result repair, review management and crisis communications. He writes about how public perception of a company is formed, contested and misread.