A customer complains about how a call went, and I have no proof of what was actually said. Now what?

This page may contain affiliate links from which Blur Studio earns commissions.

This situation is frustrating precisely because there's no fix for the call that already happened. Whatever was actually said is gone the moment the call ends unless you had something capturing it, and a one-off complaint with nothing to check it against leaves you choosing between distrusting a customer or distrusting your own staff, neither of which is a good position.

Fixing this going forward, not the call that already happened

The structural fix is recording calls by default, not reactively after a complaint. On KrispCall, a cloud phone system for business calls and texting, recording requires Advance plan or above ($40 a user a month, $32 annual); it isn't included on the entry Starter plan, which is worth knowing before you assume a cheaper plan already covers this. Once you're on Advance, you can turn on Auto Record Calls for a specific number under Settings, My Numbers, so every call on that line is captured without relying on anyone remembering to hit a button.

Live monitoring is a second, complementary layer: rather than reviewing after the fact, a manager can listen to calls as they happen on lines where quality is a known concern. KrispCall lists this kind of capability among its features, but I couldn't confirm the exact plan tier or detailed mechanics, since the relevant pages block automated access and no dedicated help article turned up. If this specific feature matters to your decision, confirm it directly with KrispCall, ideally on a trial, rather than assuming it behaves like another provider's version. Listening in is also how teams run coaching during live calls.

The consent side of this, both directions

Recording and monitoring both typically require telling the people on the call, in most places this isn't optional. A standard disclosure, in your call script or an automated message before the call connects, usually covers this for ongoing recording and monitoring, but the exact rule depends on where you and the caller both are, and you need the strictest rule that applies if your calls cross borders. This cuts both ways in a quality dispute too: a recording your staff didn't know was happening, if the law in your area required disclosure to them specifically, can create its own separate problem on top of the quality complaint you were trying to resolve.

Step by step: setting this up properly

1. Confirm the consent rule for recording and monitoring where you and your callers are based.
2. Turn on auto-recording for any customer-facing line where quality complaints have happened before or are plausible.
3. Add the disclosure to your call flow, either in the greeting message or as something staff say at the start of a call.
4. Review a sample of calls weekly, not just the ones tied to a complaint. A handful of random calls a week surfaces a pattern before it becomes a string of complaints.
5. Use live monitoring selectively on lines or staff where there's a specific, known concern, rather than constantly, which both respects staff and keeps the practice focused on where it actually helps.

A worked example: a recurring complaint pattern

Say a property management business gets a complaint that a tenant was "spoken to rudely" about a late rent payment. With recording running, the manager pulls the actual call and hears the staff member was firm but not rude, which settles the specific complaint quickly and fairly. But reviewing a weekly sample of other calls from the same staff member also surfaces a real pattern, a tendency to sound short with tenants who ask the same question twice. That second finding wouldn't have come from the one complaint alone; it only shows up because calls were being sampled regularly, not just pulled reactively when something went wrong. That's the real value of routine review over complaint-triggered review: it catches the pattern before it becomes five complaints instead of one.

Common mistakes with quality monitoring

The most common one is only reviewing calls after a complaint arrives, rather than sampling routinely. Reactive-only review means you only ever see the problem that already escalated; it never catches the pattern building toward the next one. Build a habit of pulling a handful of random calls every week, not just the ones someone already complained about. Those same calls also work as training material for new staff.

The second is monitoring or recording without telling staff specifically, relying only on a customer-facing disclosure. Staff who discover after the fact that they were being listened to without being told directly tend to feel surveilled rather than supported, which damages trust more than the monitoring itself ever would, regardless of how it's actually used.

The third is treating every flagged call the same way. A call where someone was genuinely rude deserves a direct conversation; a call where the tone was simply a little flat after a long day deserves a lighter touch. Reviewing calls without distinguishing between these two categories tends to make staff overly cautious on every call rather than specifically correcting the behavior that actually needs fixing.

A second worked example: when recording exonerates staff

Say a customer calls to dispute a quote, insisting the price agreed over the phone was lower than what showed up on the invoice. Without a recording, this becomes an unresolvable disagreement, likely settled in the customer's favor by default since a business without proof of its own side tends to lose the benefit of the doubt in these disputes. With the call recorded and the disclosure properly given at the start, pulling the actual audio settles it definitively, and in this case it turns out the staff member quoted correctly and the customer misremembered. Recording protects staff just as often as it protects customers, which is worth stating plainly when introducing it to a team that might otherwise read a new recording policy as a sign of distrust rather than a tool that cuts both ways.

When this isn't the right fix

If quality complaints are genuinely rare, a full recording-and-monitoring setup across every call might be more infrastructure than the problem justifies, especially for a very small team where a manager already overhears most calls naturally in a shared office. Recording and monitoring earn their cost once complaints are recurring enough that "he said, she said" is becoming a real pattern, not a one-off, or once your team is remote or spread across locations where a manager can't naturally overhear anything.

How I'd approach this

If quality review has become a regular enough task that manually sampling and listening to calls each week is eating real time, I'd build a lightweight system that flags calls matching risk patterns, certain keywords, unusually short calls, repeat callers, so a manager reviews a short, relevant shortlist instead of picking calls at random. A simpler first step is getting call summaries without relistening to each recording. That's a genuinely useful custom layer on top of your existing phone system, worth building once manual sampling has become the actual bottleneck rather than a quick weekly habit.