Online Reputation Management: The Reviews Are Symptoms, Not the Disease
A multi-location services business weworked with had a reputation problem they were attacking through the standardplaybook. Reputation management software was monitoring reviews across majorplatforms, automated requests were prompting satisfied customers to leavepositive reviews, and a response team was replying to negative reviews within24 hours. The Google rating had inched up from 3.8 to 4.1 over twelve months.Leadership treated this as progress.
We looked at the underlying drivers. Thenegative reviews — the ones requiring response, the ones dragging the average —clustered around specific issues: missed appointments, rescheduling friction,unclear pricing communicated late in the process. These weren’t unknown issues.The contact center had been handling them all year. They generated complaints,escalations, and eventually reviews — but the reviews were the final, public,indelible expression of what had been a private problem for months.
The reputation management program wasworking downstream of the actual reputation. The reviews were symptoms. Thedisease was further upstream, in the conversations that no one was treating asreputation-relevant.
Online reputation, in the way most customers experience it, isn’t afunction of what your past customers wrote. It’s a function of what your pastcustomers’ experiences were like, expressed publicly through whichever channelthat customer used.
This sounds obvious and isn’t. The implication is that reputation isbuilt in the actual customer experience — most of which lives in interactionsthe customer never explicitly intended as “reputation events.” The phone callwhere they tried to reschedule, the chat where they asked about pricing, theemail where they reported a problem. These produce the satisfaction ordissatisfaction that later expresses itself as a review.
Reputation management programs that focus on the review layer areoperating at the end of a long causal chain. They can encourage positivereviews from satisfied customers, respond to negative reviews to soften theirimpact, and monitor sentiment trends in public data. None of this changes theunderlying experience that produces reviews in the first place. The reviewskeep coming, with the same distribution, because the experiences producing themhaven’t changed.
The actual reputation-building moments cluster in a small number ofhigh-stakes interactions:
The unexpected problem. Something hasgone wrong from the customer’s perspective. The interaction that followsdetermines whether they’ll later describe the company as “great when somethingwent wrong” or as “they really dropped the ball.” This is the highest-leveragereputation moment most companies don’t measure or coach for.
The pricing or contract surprise. A costthe customer didn’t expect. A clause that becomes relevant in a way they didn’tanticipate. The conversation where the company explains it is the conversationthat determines whether the surprise becomes a review.
The recovery from poor service.Something has already gone wrong — a missed deadline, a quality issue, abilling error. The recovery conversation either restores the relationship orseals its negative trajectory. Recovery conversations are disproportionatelypredictive of reviews.
The end of relationship. Thecancellation conversation. The churn conversation. How these are handledproduces more reviews — positive and negative — than most customer experienceteams realize.
Each of these is happening in your contact center. Almost none ofthem are being analyzed as reputation events. The reputation management programis reading the reviews these conversations eventually produce, six to twelveweeks later, when nothing can be done about them.
A reputation program that takes the upstream view organizes aroundfour motions.
Identify the high-stakes conversation types. Not every contact is reputation-relevant. The four categories aboveusually cover most of the variance. Flag these conversations specifically forcloser attention.
Measure outcomes on those conversations specifically. Standard CSAT and FCR don’t capture reputation impact. A customercan be FCR-resolved and still produce a negative review because the experienceof getting there was painful. Reputation-relevant measurement looks atperceived effort, perceived fairness, and perceived care.
Coach the conversations that produce reputation moments. Conversationanalytics can identify which agents handle high-stakes interactions well andwhich don’t. The differences are coachable. Most reputation problems trace backto a small number of patterns in how difficult moments are handled.
Close the loop with the review layer.When negative reviews appear, trace them back to the originating conversations.The pattern reveals which call types and which conversation moments areproducing the most reputation damage. This becomes the priority list forupstream intervention.
A common shortcut in reputation management is to solicit reviewsaggressively from customers identified as satisfied — through CSAT surveys,recent positive interactions, or NPS responses. This works in the short term:average ratings rise. It fails in the medium term because it doesn’t change theunderlying experience producing the negative reviews. The dissatisfiedcustomers continue to leave reviews unsolicited. The solicited positive reviewsdilute their impact temporarily, but the trajectory of the underlying signaldoesn’t change.
Worse, aggressive solicitation can produce its own reputation cost.Customers can tell when they’re being prompted asymmetrically — positivecontact triggering review requests, negative contact triggering nothing. Thisperceived manipulation, once detected, becomes its own negative narrative.
The durable approach is to fix the experiences producing the reviewsyou don’t want, not to drown them in solicited reviews you do want.
1. Trace 20 recent negative reviewsback to the originating interaction. When did thiscustomer first contact you about this issue? What happened in thatconversation? The pattern will reveal the upstream source.
2. Identify the high-stakesconversation types in your business. Problemreports, pricing surprises, recovery conversations, cancellations. Estimatevolume for each. These are your reputation-building moments.
3. Listen to 10 problem-reportingcalls. How do agents handle the moment when thecustomer first expresses dissatisfaction? The variance between agents isusually large and entirely coachable.
4. Measure agent performance onrecovery conversations specifically. Calls thatfollow a known prior issue with the customer. These predict reviewsdisproportionately and almost no QA program measures them as a category.
5. Stop optimizing the review layeralone. Continue monitoring and responding toreviews, but recognize that the underlying signal is what determines thelong-term trajectory. The conversations producing the reviews are the actualreputation work.
The Google rating that went from 3.8 to4.1 wasn’t reputation improvement. It was the visible surface of an underlyingexperience that hadn’t changed, with a thicker layer of solicited positivereviews on top. Real reputation work happens further upstream, in conversationsthe customer never thought of as reputation events, six to twelve weeks beforeany review gets written. That’s where the reviews are being made, and that’swhere the reputation actually lives.
The AI terminology chaos is real. Your "divide and conquer" framework is the clarity we needed.
Finally, a clear way to cut through the AI hype. It's not about the name, but the problem it solves.
