Customer Experience Statistics
Most CX statistics you'll find online are undateable. No year, no methodology, and a "source" that turns out to be a blog citing another blog citing a conference slide from 2016.
We know, because that's what this page used to be. So we rebuilt it, and we threw a lot away.
Every statistic below carries a year and a link to the organisation that actually ran the research. If we couldn't trace a number to a primary source, it isn't here — no matter how often you've seen it quoted.
ℹ️ The standard we hold every stat to
To appear on this page, a statistic must have (1) a publication year, (2) a named organisation that conducted the research, and (3) a working link to that organisation's own report or release. Predictions are badged Forecast so nobody mistakes one for a finding.
Sources here include Qualtrics XM Institute, Forrester, Gartner, McKinsey, PwC, Zendesk, Salesforce, Deloitte, Genesys, Twilio, Intercom, ServiceNow, Pew Research, the American Customer Satisfaction Index, and the UK Institute of Customer Service.
85% of CX leaders say customers will drop a brand that cannot resolve their issue on first contact.
Zendesk, 202674% of consumers are frustrated when they have to repeat information to different agents.
Zendesk, 2026US customer satisfaction has flatlined at 76.9 out of 100 and has not materially improved since 2017.
ACSI, 2026Australians spent 113.5 million hours on hold in 2025 — down 10 million hours year on year.
ServiceNow, 2026The Statistics We Deleted
Before the new numbers, the old ones. These are the most-quoted CX statistics on the internet — and every one of them failed our source check. You will see them on hundreds of agency blogs, in vendor decks, and quite possibly in your own board papers.
We removed them all. Here's why, so you can stop using them too.
Every citation trail ends in a blog citing another blog. There is no study. The defensible equivalent is McKinsey's finding that replacing the value of one lost customer can require acquiring three new ones.
Traces to Reichheld & Sasser (Harvard Business Review, 1990), where cutting defections 5% raised profits 85% in one bank's branch network. The 95% figure appears in no Bain source. Thirty-six years old and still quoted as current.
From Walker's "Customers 2020", published in 2013. The deadline passed six years ago. It is still being quoted in the present tense.
Not traceable to any published Gartner document. The deadline passed six years ago and it was comprehensively wrong. This is the single most-recycled fake stat in the industry.
From PwC's Experience is Everything (2018), now quoted everywhere with the date quietly removed. The current, traceable version: 36% of UK customers say they'd pay more for excellent service (Institute of Customer Service, 2026).
From an Esteban Kolsky presentation. No underlying study was ever published. The real, sourced version is in the library below: fewer than 3 in 10 customers now tell a company what went wrong (Qualtrics XM Institute, 2026).
Attributed to a book ("Understanding Customers", Ruby Newell-Legner). No published dataset exists anywhere.
And two forecasts that have now quietly failed
Gartner, 2022: "By 2026, conversational AI in contact centres will reduce agent labour costs by $80 billion." It is now mid-2026. Gartner's own December 2025 survey found only 20% of service leaders had reduced headcount at all because of AI — and Gartner now predicts half of those who did cut will rehire by 2027.
Gartner, 2021: "By 2025, proactive customer engagement will outnumber reactive engagement." Gartner's own later research found just 13% of customers could recall receiving any proactive service at all.
This is why every prediction in our library is badged Forecast. A forecast is a hypothesis with a marketing budget, not a finding.
A Note on What "CX" Means Here
Two quite different things both get called "CX", and this page carries both — so it's worth being clear which is which before you read the leaderboard below.
The library above is broad customer experience: brand, company and consumer research from around the world, measuring how people feel about the organisations they deal with across every channel.
The leaderboard below is narrower, and it's ours. ACXPA's Australian Call Centre Rankings measure one specific thing — how well individual Australian contact centres handle real customer interactions, mystery shopped against the Contact Centre CX Standards on 80+ metrics. Not a survey, not self-reported: actual calls.
ℹ️ So the rankings aren't a verdict on a brand's whole CX
A company can top the call centre rankings and still frustrate you on its website — or the reverse. The rankings tell you one thing the broad surveys can't: on measured evidence rather than opinion, which Australian contact centres are genuinely good on the phone. Read them alongside the global library, not as the same kind of number.
Learn more about the rankings >
Data accurate as of July 2026
The CX Statistics Library
Every statistic is dated and linked to the organisation that ran the research. Filter by topic or region, search the full text, or hide the forecasts to see only observed findings.
💡 Hit Copy on any stat to grab it with its full citation, ready to paste into a deck or a board paper.
85% of CX leaders say customers will drop a brand that cannot resolve their issue on first contact.
74% of consumers are frustrated when they have to repeat information to different agents.
US customer satisfaction has flatlined at 76.9 out of 100 and has not materially improved since 2017.
Australians spent 113.5 million hours on hold in 2025 — down 10 million hours year on year.
51% of Australians say their biggest AI concern is that it cannot understand their issue; 67% say it falls short on anything nuanced.
91% of customer service leaders report pressure from executives to implement AI.
Half of the companies that cut service staff because of AI are predicted to rehire for similar work under new job titles by 2027.
Nearly 1 in 5 consumers who used AI for customer service got no benefit at all — a failure rate about four times higher than AI use in general.
Half of consumers say they would prefer to give their business to brands that do NOT use generative AI in consumer-facing content.
82% of senior service leaders invested in AI — but only 10% say they have reached mature deployment at scale.
64% of consumers prefer companies that tailor experiences to their individual needs — up 2.5 points year on year.
Only 39% of consumers trust organisations to use their personal information responsibly.
Only 3 in 10 customers will tell a company directly what went wrong after a bad experience — an all-time low.
30% of consumers now say nothing at all after a bad experience and simply switch — up 9.2 points in five years.
For every 10 poor experiences a company delivers, 5 result in the customer reducing or eliminating their spend.
36% of UK customers say they are willing to pay more for excellent service.
88% of customers expect faster response times than they did a year ago.
86% of consumers say responsiveness and accurate resolution strongly influence their purchase decisions.
The UK Customer Satisfaction Index stands at 78.3 out of 100, up 1 point year on year.
The average Australian spends 9.3 hours a year resolving customer issues, down from 11.1 hours.
Three-quarters of Australians now prefer to try self-service before calling a customer service representative.
76% of consumers would choose a company that lets them share text, images and video in one thread without starting over.
Generative AI cost per resolution is predicted to exceed US$3 by 2030 — higher than many offshore human agents.
95% of consumers expect an explanation when a decision affecting them is made by AI.
74% of consumers now expect 24/7 service, because AI has made it possible.
62% of service teams say their metrics improved after implementing AI — rising to 87% among teams with mature deployments.
64% of service leaders report higher agent productivity from AI, and 39% report a lower cost per contact.
48% of companies with mature service capabilities already use agentic AI, against 24% of low-maturity peers.
Misuse of personal data is now the top consumer concern about AI-automated interactions, cited by 53% — up 8 points in a year.
32% of consumers are uncomfortable with their data being used for personalisation in any common form.
46% of consumers would share more data if organisations were transparent about what is collected; 45% want to be able to delete it easily.
67% of consumers expect brands to tailor support based on prior interactions; 81% want agents to continue the conversation without backtracking.
Only 15% of consumers post on social media after a bad experience — down 4.7 points in five years.
82% of CX leaders say promptable analytics let them unlock insights in seconds; 87% say AI is already improving their data and analytics.
High-maturity CX organisations track AI-driven metrics at triple the rate of low-maturity peers — 66% track automation success rates against just 21%.
Poor customer experiences put close to US$3 trillion of global sales at risk each year.
34% of consumers reduce their spending after a negative experience, and a further 13% stop spending entirely.
Brands that excel at customer experience grow revenue at twice the rate of brands that do not.
Nine in ten executives believe customer loyalty to their brand has grown. Only four in ten consumers agree.
Poor customer service costs UK organisations an estimated £7.3 billion every month in lost productivity.
25% of US brands saw their CX score decline in 2025 and only 7% improved — a fourth straight year of decline, and an all-time low.
35% of Australian and New Zealand consumers stopped doing business with a company in the past year after bad service.
60% of service agents fail to promote self-service to customers — and 12% make explicitly negative comments about it.
Service teams estimate 30% of customer service cases are currently handled by AI.
Only 20% of service leaders report reducing agent headcount because of AI. Most say staffing held steady even as volumes grew.
More than 40% of agentic AI projects are predicted to be cancelled by the end of 2027 — escalating costs, unclear business value.
86% of consumers say human interaction is moderately or very important to their experience of a brand.
Just 39% of organisations using AI can attribute any EBIT impact to it — and most of those say it is under 5% of EBIT.
88% of consumers are more likely to buy when engagement is personalised in real time — but only 44% of brands say they can execute at that level.
Only 45% of consumers feel understood by the brands they deal with — while 83% of business leaders claim to deeply understand their customers.
94% of Australian consumers have stopped purchasing from at least one company after a poor service experience.
For Australians, the top drivers of service excellence are information accuracy (91%), knowledgeable representatives (84%) and consistency across channels (79%).
47% of Australian and NZ customers say they would switch to a competitor because of slow or inadequate service.
Only 30% of ANZ customers prefer service from a human over a machine — yet 51% say current chatbots do not understand their questions.
11% of all customer experiences are bad — and 47% of those lead the customer to cut their spending.
Companies rated highest by consumers for CX outperformed their industry stock index; the lowest-rated fell 23 percentage points behind.
American organisations risk US$973 billion of sales by delivering bad customer experiences — the largest exposure of any country.
52% of consumers stopped buying from a brand because of a bad experience with its products or services.
In Asia Pacific, 37% of brands saw CX scores fall and just 5% rose.
Nearly 60% of ANZ consumers would switch away from a favourite brand after just two to five bad experiences — against 53% globally.
83% of ANZ consumers agree that "a company is only as good as its service" — against an Asia-Pacific average of 76%.
Only 35% of customers whose last interaction was by phone are willing to adopt a generative AI assistant instead.
Live chat, self-service portals and knowledge management are expected to overtake phone and email as the most valuable service technologies by 2027.
Service teams project AI will handle 50% of all customer service cases by 2027.
Agents using AI spend 20% less time on routine cases — about four hours a week freed for complex work.
Agentic AI is predicted to autonomously resolve 80% of common service issues by 2029, cutting operational costs by 30%.
58% of consumers are only somewhat or not at all comfortable using AI tools to engage with brands.
23% of organisations are scaling an agentic AI system somewhere — but in any single business function, no more than 10% have scaled AI agents.
51% of service leaders say security concerns have delayed or limited their AI initiatives.
Half of US adults say AI in daily life makes them feel more concerned than excited. Only 10% are more excited than concerned.
71% of consumers abandon a purchase when the experience does not feel relevant to them.
61% of consumers do not believe brands use their data in their best interest. Just 15% absolutely trust brands with it.
84% of consumers want control over their own personalisation settings.
Companies that unify their service channel data are 1.4x more likely to achieve a very successful AI implementation.
Only 34% of Australian consumers believe companies prioritise service excellence — though that is up from 28% a year earlier.
77% of Australians prefer phone support for complex enquiries, while 46% favour digital self-service for simple ones.
Australians share negative service experiences far more than positive ones — 90% pass on a bad experience versus 76% a good one.
54% of Australian consumers reported mostly positive purchasing experiences, with only 2% mostly negative.
55% of ANZ customers expect improved speed and efficiency to be the top benefit of AI in customer service.
Lack of empathy is the top service frustration for ANZ customers, cited by 39%.
77% of service agents report heavier and more complex workloads than a year ago, and more than half report burnout.
Only 14% of customer service issues are fully resolved in self-service.
73% of customers use self-service at some point — yet nearly 9 in 10 journeys that start there end up resolved across other channels.
A poor service experience is the second most common reason customers stopped buying from a brand, cited by 43%.
69% of service decision makers say agent attrition is a major or moderate challenge.
69% of agents report difficulty balancing speed and quality in their interactions.
Even for issues customers call very simple, only 36% resolve fully in self-service.
69% of consumers expect consistent interactions across departments, and almost 60% prefer fewer touchpoints.
72% of consumers trust companies less than a year ago, and 65% feel companies are reckless with customer data.
The share of service organisations tracking revenue generation as a KPI has nearly doubled since 2018, from 51% to 91%.
Companies leading on customer experience achieved more than double the revenue growth of CX laggards.
Replacing the value of one lost customer can require acquiring three new ones.
No statistics match those filters.
Go Deeper Into Australian Data
The library above is the world's research. These are closer to home — Australian contact centre data, some of it ACXPA's own, some of it work we contribute to.
Australian Call Centre Rankings
Which Australian call centres are actually any good — mystery shopped against the CX Standards, by sector, updated quarterly. The public leaderboard is free; ACXPA members get the deeper, exclusive ranking data behind it.
Contact Centre Best Practice Report
Australia's deepest contact centre study — attrition, absenteeism, salaries, AI adoption, agent sentiment. Produced by Smaart Recruitment; ACXPA contributes the mystery shopping chapter. Our article summarises the key findings.
Worst Hold Times in Australia
Which industries keep you waiting longest, and why the wait is usually a deliberate business decision rather than an accident.
Contact Centre CX Standards
The framework the rankings are assessed against — what "good" actually means, defined and published rather than asserted.
Want to know where your contact centre sits against all of it? That's what independent benchmarking is for.
Put the Numbers to Work on Your Own CX
Statistics are only useful if they change something. If these have you wondering where your own customer experience actually stands, start here.
Turn the Numbers Into a Business Case
Help Us Keep It Current
Good CX research comes out constantly, and no single team catches all of it. If you've published a study, or you've read one worth adding, send it our way — this page is better as a shared effort than a closed one.
Submit a stat or report
Have a dated, primary-sourced statistic we're missing — or a whole report? Send it over. Vendors and researchers with current data are especially welcome; it's a straightforward way to get credible work in front of a CX audience.
Flag something that looks wrong
Spotted a figure that's been superseded, or a source that's changed its numbers? Tell us and we'll check it. We'd rather be corrected than carry a stat that's quietly gone stale.
ℹ️ How this page is maintained
The statistics live in a single data file, and we monitor for new and superseded research on an ongoing basis — AI-assisted, human-checked. AI helps us scan for fresh data and flag figures that may have aged; a person verifies every stat against its primary source before it goes in or comes out. If we get one wrong, tell us and we'll fix it.
These statistics are here to help you make a case, not to sell you anything. Every figure links to the organisation that produced it, so you can cite the source directly — you never have to take our word for it.
Frequently Asked Questions
Why are there fewer statistics here than on other CX stats pages?
Because most of theirs don't survive a source check. This page previously carried 155 statistics — 135 of which had no year attached at all, and many of which traced back to content-farm aggregators recycling each other. We rebuilt it from scratch with a hard rule: a year, a named research organisation, and a working link to that organisation's own report. Only what cleared that bar is here. We'd rather publish a set you can defend in a meeting than a bigger one you can't.
How do I cite a statistic from this page?
Don't cite us — cite the source. Every statistic links to the organisation that actually conducted the research, and the Copy button gives you the stat, the source, the year and the URL in one click, ready to paste. If you're building something more substantial, ACXPA Members can select multiple statistics and export a fully-cited Stat Pack as a print-ready PDF.
What does the "Forecast" badge mean?
It means the number is a prediction, not an observed result — and you should treat it very differently. We learned this the hard way: we used to republish Gartner's 2022 forecast that conversational AI would cut $80 billion from contact centre labour costs by 2026. It's now 2026, and Gartner's own research found only 20% of service leaders have reduced headcount at all. Forecasts are hypotheses with a marketing budget attached. We badge them so you can decide how much weight to give them.
How often is this page updated?
Continuously. The statistics live in a single data file rather than being hard-coded into the page, so a new figure can be added the day it's published — which is exactly why the old version rotted and this one shouldn't. If you spot a statistic that has been superseded, or a source that has changed its numbers, tell us and we'll fix it.
Do you have Australian-specific CX data?
Yes. Filter the library to Australia for the third-party research, then go to the Australian Call Centre Rankings — ACXPA's own, built from real mystery-shopped customer interactions against the Contact Centre CX Standards, not a survey. For the broader industry picture, the Contact Centre Best Practice Report (produced by Smaart Recruitment, with ACXPA contributing the mystery shopping chapter) is the deepest study of the Australian contact centre workforce. The public leaderboard is free; members see the deeper ranking detail.
Can I filter or link to a specific set of statistics?
Yes. The filters update the page URL as you use them, so a filtered view is shareable — you can send a colleague a link that opens straight on, say, AI & Technology statistics for Australia. You can also search the full text of every statistic, and hide forecasts entirely to see only observed findings.
Summary
The customer experience statistics industry has a sourcing problem. The most-quoted numbers in the field — the 5x acquisition cost, the 25–95% retention profit, the 2020 predictions still written in the present tense — are circular, expired, or were never real to begin with.
What's left when you strip those out is smaller, but it's true, and you can take it into a meeting without it falling apart.
The picture it paints is not a comfortable one: satisfaction is flat or falling in the US and UK, a quarter of brands are going backwards, consumers are increasingly not bothering to complain — they simply leave — and the AI that was supposed to fix all of it has a failure rate in customer service roughly four times higher than its failure rate everywhere else.
Which is, in fairness, the most useful thing a statistics page can tell you.
