Take the ReplyResearch Customer Support Black Hole Survey
CXReachabilityResponsiveness

What companies mean when they talk about “deflection”

Eva Stefanidou's avatar Eva Stefanidou
4 weeks ago
12 min read

In customer-service and customer-experience circles, deflection has a special meaning. It usually means a customer who might otherwise contact a human support agent is steered somewhere else first.

That route might be a help article, chatbot, app flow, phone menu, customer portal, community forum, automated form, or other self-service route.

Inside the industry, this can sound neutral or technical. A company might talk about “case deflection”, “call deflection”, “ticket deflection” or “self-service containment”.

Why the word lands awkwardly

For someone outside CX, the word lands differently. In ordinary English, deflection sounds like avoidance. We deflect blame, deflect questions, deflect responsibility.

So when a company says it is improving deflection, a customer might wonder what is being improved: the route to help, or the company’s ability to keep people away from it?

That tension is useful because it reveals the gap between the company’s map of the service and the customer’s experience of trying to use it.

The useful version exists

There is a good version of this, and it’s worth starting there. Sometimes we don’t want to phone anyone.

We don’t want to wait in a queue to reset a password, check a refund, download an invoice, change an address, track a parcel, or read a returns policy.

If the answer is clear and the task is simple, self-service can be faster, less awkward, and more convenient than speaking to a person.

Self-service is not the same as deflection

In that situation, the customer probably doesn’t think of themselves as having been “deflected”. They think: that was easy, I sorted it, I didn’t need to call.

The positive experience is not being turned away from support. It is getting the thing done.

That distinction matters because the industry word can blur two different outcomes: self-resolution, where the customer solves the problem, and contact avoidance, where the customer simply doesn’t reach a person.

The metric and the experience differ

From the company’s side, deflection can look attractive because human support costs money.

Every phone call, email, live chat, complaint case or manual intervention takes staff time. Most organisations have a reason to reduce avoidable demand.

That does not make deflection automatically bad. A good help centre can stop people needing to queue, and a good app flow can let someone make a simple change outside office hours.

The problem is what gets counted

The problem starts when the organisation measures avoided contact more clearly than solved issues.

A company may know fewer tickets were created. It may be less sure whether people got an answer, gave up, tried again elsewhere, complained through another route, or decided the process was too much trouble.

That is why “deflection” can become a slippery success metric. It may record that contact did not happen, without proving that help did.

A better question

The uploaded research notes make the central distinction plainly: resolution means the customer’s issue was actually solved, while deflection means the customer did not reach a human-support channel. Those are not the same thing.

So the better question is not only: did the organisation avoid a ticket?

It is also: did the person get the problem dealt with, in a way they could understand and rely on?

Why customers suspect cost-cutting

This is why deflection can sound awkward outside the CX world. It appears to name the company’s success before it names the customer’s.

A customer doesn’t usually arrive looking for a channel strategy. They arrive because a refund hasn’t appeared, a bill looks wrong, a delivery has gone missing, an appointment has vanished, or the service has failed.

From their side, the important question is not which channel absorbed the demand. It is whether anyone, or anything, helped.

When convenience sounds like cover

When companies use deflection as a positive word, they may be describing a real efficiency gain.

But they may also be exposing a priority customers already suspect: that the contact itself is being treated as something to be reduced, not necessarily answered.

That is why “this gives you more freedom” can land badly when someone is already stuck. The customer may hear something closer to: we made it harder to reach a person, and now we’re calling that convenience.

When self-service becomes contact theatre

At ReplyResearch, this is the kind of pattern we would describe as contact theatre.

Not because every automated route is fake, but because some systems create the appearance of help without a reliable route to someone who can actually deal with the issue.

The form exists, the chatbot exists, the help page exists, the acknowledgement email exists. From the outside, it can look official enough.

The route may not do what it appears to do

If the route doesn’t resolve the problem, escalate cleanly, or say what happens next, the visible process may be doing less than it appears to do.

The customer might still have to come back through another route. Or they may wait because they believe the issue is now inside the system.

Or they may stop trying, which can be easy to misread as success if the organisation is mostly watching contact volume.

Silence is not resolution

This is one danger of treating deflection as a headline win.

Silence is not the same as resolution. Fewer contacts may mean the system is working well, but they may also mean people have been sent round the houses until they run out of patience.

That matters because a system can look quieter from the inside while feeling harder to use from the outside.

The evidence is not especially comforting

The research notes cite Gartner’s 2024 finding that only 14% of customer-service and support issues were fully resolved in self-service. Even for issues customers described as “very simple”, only 36% were fully resolved that way.

That does not mean self-service should be scrapped.

It does suggest that many organisations may be overclaiming what self-service can reliably do, especially if they are counting avoided contact as though it were the same thing as a completed outcome.

The risk is sharper when choice is limited

The risk is sharper where the customer has little choice.

In a normal market, we may be able to leave a company that makes support too hard.

With banks, utilities, telecoms, public bodies, landlords, insurers, platforms, or essential local services, leaving may be slow, costly, unrealistic or impossible.

When blocked contact becomes harm

When the issue is serious, blocked contact is not just irritating.

The uploaded notes also refer to the US Consumer Financial Protection Bureau’s warning that deficient chatbots preventing access to live human support can lead to law violations, diminished service and other harms.

Its concern was about consumer finance, but the principle travels: if an automated front door stops people reaching help when the issue needs a person, the contact route itself becomes part of the harm.

We still need a way back in

In practice, a service is not only the formal process. It is also the way we get help when that process fails.

That is true whether the service is a council collection, a bank account, a benefits claim, a tenancy repair, a healthcare appointment, or a consumer complaint.

We follow the rules, use the routes we’re given, and assume that if something goes wrong there is still a way back in.

Self-service needs a live edge

Self-service can be part of that. It can take pressure off staff and make simple tasks easier for us.

But it needs to be backed by a live route when the simple route isn’t enough.

Otherwise, the extra work comes back to the person trying to get help. They have to search, repeat themselves, decode the system, and work out which route is actually being watched.

The burden should not fall on us

It is hard to see why that burden should fall mainly on the customer.

That is especially true when the organisation designed the route and knows which channels are real, current and staffed.

This does not mean every broken route proves bad intent. It may come from an old page, a changed process, a legacy inbox, an undertrained bot, or a team that is stretched too thin.

Innocent explanations still matter

Making room for innocent explanations keeps the criticism fair.

A broken route may be accidental, but the customer’s problem does not disappear because the failure was accidental.

From our side, the issue is simpler: we were told to use a route, and the route did not get us to anyone or anything able to help.

What a better test would ask

A better test of deflection would start from the outside, with the route people actually use.

What happens when someone follows the instruction? Does the page answer the question? Does the chatbot understand the problem? Is the form picked up?

Is there a clear escalation route? Does the customer know when to expect a reply? If the automated route fails, is there a visible way to reach a person without starting again?

Start with the lived route

Those questions are less tidy than a deflection rate, but they are closer to the customer’s experience.

They also fit the ReplyResearch rule of testing from the outside: the key question is not what channels the organisation says it provides, but what happens when someone actually uses them.

That outside view matters because the official map can look complete while the lived route is broken.

The question deflection leaves behind

The immediate word is a CX term, but the wider issue is about whether organisations are really there when people need them.

If deflection means helping us solve simple problems quickly, there is not much to object to.

Many people would rather avoid a phone queue if the answer is clear and the task can genuinely be completed.

What is really being improved?

If deflection means counting avoided contact as success while unresolved problems sit with the customer, then the term deserves more scrutiny.

It may be measuring how well the organisation has reduced demand on itself, rather than how well it has helped the person who came looking for support.

That is why the word sounds so revealing. Companies may call it deflection when the customer does not reach support. From the customer’s side, especially when the problem is still there, it can feel much closer to being turned away.

Sign reading 'Nok Nok Footnote Zone' next to Charging Bull sculpture on city street
A sign designates a footnote-only zone near the Charging Bull statue in NYC

Disclosure: The diagnostic tools referenced below were developed by NokNok, a specialist in online responsiveness tool design.

This Footnote Zone uses NokNok’s diagnostic toolkit to examine how deflection, self-service containment, and contact-avoidance systems can be tested from the outside to distinguish genuine resolution from routes that merely reduce visible demand.

  • Email Finder: When organisations steer customers toward help centres, portals, forms, bots, or automated routes, direct contact options can become harder to find, outdated, or unclear, leaving users unsure which routes are real, current, and staffed. Email Finder scans an organisation’s website and related public-facing materials for published email addresses, then reports structural deficiencies, discrepancies, missing contact routes, and other contactability gaps.
  • Reply Radar: The article warns that fewer contacts may be misread as success when customers have actually given up, tried another route, or waited inside a system that is not being actively monitored. Reply Radar deploys targeted test emails and quantitatively measures reply rates, latency, response consistency, and related responsiveness benchmarks to test whether human-support routes are functioning reliably.
  • Compliance Sniffer: Deflection becomes especially problematic when automated routes produce vague acknowledgements, empty reassurance, evasive responses, or answers that fail to resolve the customer’s actual issue. Compliance Sniffer analyzes incoming responses for objective quality, clarity, relevance, escalation, and compliance benchmarks, helping identify whether the response meets basic communication expectations rather than merely appearing official.
  • Mystery Shopper: The article’s central concern is that the official service map can look complete while the lived route is broken, with customers pushed through chatbots, forms, help pages, and escalation paths that do not reliably lead to help. Mystery Shopper executes a comprehensive end-to-end responsiveness UX audit, testing how a real user experiences the organisation’s contact, response, and escalation pathways from the outside.

Disclosure: The diagnostic tools referenced in this Footnote Zone were developed by NokNok, a specialist in online responsiveness tool design. ReplyResearch may use NokNok tools, resources, or analysis when preparing coverage, while retaining responsibility for its editorial decisions, including what topics to cover, what sources to cite, and how stories are presented. Read the full ReplyResearch Collaborative Disclosure Policy

Sources and relevant reading for What companies mean when they talk about “deflection”

Eva Stefanidou's avatar

Eva Stefanidou

View all posts