Microsoft's State of Global Customer Service Report Explained

Every CX deck has one slide that refuses to die: the one with Microsoft's 96% stat on it. You have seen it in a dozen boardrooms, usually projected by someone making the case for a service-desk budget they already know they will not get in full. The report behind that slide is nearly a decade old. It is still one of the most quoted pieces of research in the industry. That gap — between a report's age and its staying power — tells you more about customer expectations than the stat itself ever could.

Microsoft's State of Global Customer Service Report is a multi-country consumer research study, most recently run at scale in 2017, that measured how people feel about service quality, channel preference and brand loyalty. Its best-known finding is that 96% of consumers worldwide say customer service is an important factor in their choice of loyalty to a brand, and that a majority reported higher service expectations than they'd held just a year earlier. The report has not been refreshed on the same scale since, yet its central argument — that service had stopped being a cost centre and started being a loyalty engine — has only become more true.

What is Microsoft's State of Global Customer Service Report?

It was a consumer research study published by Microsoft in 2017, surveying consumers across multiple countries on their attitudes toward customer service: what channels they preferred, how their expectations had shifted, and how service experiences connected to whether they stayed loyal to a brand or walked. It sat alongside Microsoft's broader push into Dynamics 365 and service-cloud software, which is worth naming honestly — the research doubled as market validation for a product line. That commercial motive doesn't invalidate the findings. It does mean the report should be read as a snapshot of consumer sentiment, not an independent academic study.

Two findings carried the report into permanent citation status. First, 96% of consumers said customer service was an important factor in their loyalty decisions — not a nice-to-have, a precondition. Second, a majority said their expectations for service had risen in just the preceding year. Taken together, those two numbers reframed service from a support function into a loyalty mechanism, which is exactly the argument customer loyalty strategy has been built on ever since.

What did the report actually find?

Strip away the marketing framing and three structural claims remain, each of which has aged well:

• Service is a loyalty lever, not a cost line. Consumers directly connected a poor service interaction to whether they'd buy again — loyalty was conditional on service, not separate from it.
• Expectations compound rather than plateau. Respondents reported wanting more from service than they had the year before, regardless of how much service quality had actually improved in that time.
• Channel proliferation outpaced channel consistency. Consumers wanted the option to move between phone, chat, self-service and social support — and expected the context to travel with them, not reset at every handoff.

None of these are shocking in isolation. What makes the report useful is that it put numbers behind feelings practitioners already suspected, which is precisely what good research is for — it doesn't need to be new to be valuable, it needs to be confirmed.

Why does a report from 2017 still get cited today?

Because it described a structural shift, not a seasonal trend. Trend reports — including more recent ones like the Zendesk CX Trends Report or Salesforce's State of the Connected Customer research — track what's changing year to year: AI adoption, channel mix, generational preference. Microsoft's report, almost by accident, captured something closer to a law: service quality and loyalty are structurally linked, and that link doesn't depreciate the way a channel-adoption stat does.

This is the uncomfortable truth most CX leaders avoid saying out loud: a stat about rising expectations doesn't need updating every year, because the mechanism behind it — not the number — is permanent. Expectations always rise. That's not a 2017 phenomenon. It's a feature of how humans process experience, and it's where behavioural economics does more explanatory work than any survey can.

A service report doesn't go stale when the mechanism it describes is structural rather than seasonal — and expectation inflation is structural.

What's the behavioural economics explanation for rising expectations?

Psychologists call it hedonic adaptation: humans recalibrate their baseline to whatever they've just experienced, so yesterday's delight becomes today's minimum. Apply that to service and Microsoft's "expectations are rising" finding stops being a survey result and becomes a prediction. Every time a bank resolves a dispute in one chat message instead of three phone calls, that becomes the new floor — not just for that bank, but for every brand the customer deals with next. Service doesn't compete within categories anymore; it competes against the best interaction a customer had anywhere, that week.

This is also where loss aversion earns its keep. Research from Kahneman and Tversky established that losses are felt roughly twice as intensely as equivalent gains — which is why a customer who has become accustomed to fast resolution doesn't just prefer it going forward; they feel its absence as a loss, not merely a missed upgrade. A brand that improves service and later lets standards slip isn't returning to neutral in the customer's mind. It's triggering a loss. That asymmetry is the real engine behind Microsoft's loyalty stat: customers don't leave because service was bad in absolute terms. They leave because it got worse relative to what they'd come to expect.

There's a third mechanism worth naming: the peak-end rule, Daniel Kahneman's finding that people judge an experience largely by its most intense moment and how it ends, not by its average quality. This explains why channel-switching friction does disproportionate damage. A customer can tolerate a mediocre chat exchange if the resolution at the end is clean and fast. But force them to repeat their account number for the third time across three channels, and that friction becomes the ending they remember — regardless of how well the rest of the interaction went.

Where does the report get misused?

The 96% stat gets quoted to justify budget. It rarely gets used to justify action. That's the gap worth naming honestly: citing a statistic about loyalty is not the same as closing the gap between expectation and delivery. Richard Thaler's concept of sludge — friction deliberately or carelessly built into a process — is a useful lens here. Most companies that quote Microsoft's expectations stat still make customers repeat information across channels, re-authenticate at every handoff, or navigate an IVR tree to reach a human. That's sludge, and no amount of quoting a loyalty statistic offsets it.

Harvard Business Review's influential 2010 research on customer effort and loyalty made a related point: reducing customer effort predicts loyalty more reliably than trying to delight customers with surprise gestures. Microsoft's report and that HBR research point at the same target from different angles — loyalty is won less by grand gestures and more by removing the friction customers didn't ask to carry in the first place.

McKinsey's ongoing work on customer experience as a growth driver makes a similar case: companies that lead on experience outperform laggards on revenue growth over time, and the mechanism is consistency, not spectacle. Its CEO guide to customer experience is worth reading precisely because it treats CX as an operating discipline, not a marketing layer — which is the same argument Microsoft's data was pointing toward in 2017, even if the report itself was framed as consumer sentiment research rather than an operating model.

How should CX leaders apply these findings today?

Quoting the report is easy. Acting on its mechanism is harder, and it requires treating expectation inflation as a planning input rather than a mood. Here's how to do it properly:

1. Audit your channel handoffs for repeated information. Every time a customer has to re-explain their issue, you are manufacturing the exact sludge that erodes the loyalty Microsoft's research measured. Map every handoff point and eliminate re-authentication and repeated context wherever technically possible.
2. Design the ending deliberately. Given the peak-end rule, the final thirty seconds of a service interaction carry disproportionate weight in how it's remembered. Script and train for strong resolutions, not just competent middles.
3. Track expectation drift, not just satisfaction scores. A flat CSAT trend can mask rising expectations — hold the line and you're actually losing ground. Pair satisfaction metrics with a direct measure of what customers now consider table stakes.
4. Benchmark against the best experience your customer had anywhere, not just your category. Customers compare your service response time to the fastest brand they've dealt with recently, not your nearest competitor. Build that comparison into how you set internal targets.
5. Close the gap between the stat and the operating model. If leadership quotes Microsoft's loyalty finding to justify an investment, make sure the investment targets effort reduction and consistency — not a new feature that doesn't touch the friction points customers actually name.

Feeding these findings into a structured voice of customer programme turns expectation drift from an anecdote into a tracked metric — which is the difference between citing research and operationalising it. And if you're not sure where your organisation currently sits on the maturity curve the report implies, a structured CX maturity assessment is a faster starting point than another slide with the 96% stat on it.

What the report still gets right, nearly a decade on

Reports age. Mechanisms don't. Microsoft's 2017 research happened to capture a mechanism — expectation inflation driven by hedonic adaptation, amplified by loss aversion when standards slip — that doesn't need a refreshed survey to remain true. That's the real lesson for anyone building a customer experience function in 2026: stop waiting for next year's report to tell you expectations have risen again. They have. They always will. The only open question is whether your service model is built to keep pace with a baseline that never stops moving, or whether you're still running the operating model you had when 96% first showed up on a slide.