As of 29 October 2026, a study covering 10,440 online mystery-shopping journeys across 329 retailers has quantified something most service teams treat as intuition: bad execution costs far more than good execution earns. Being hard to contact reduced customers' likelihood of returning by 6.6 percentage points, while being exceptionally easy to contact earned back only 3.2. The asymmetry matters for any brand selling through digital channels, because it means the return on service investment is not symmetric. Preventing a failure is worth roughly twice as much as creating a delight, and most operating budgets are still built as if the opposite were true.
Why the downside is twice the upside
The asymmetry is a well-known pattern in service research, and the mechanism is straightforward. A good experience confirms an expectation the customer already had, so it produces little change. A bad experience breaks an expectation, and the customer generalises from that single failure to the whole brand. The result is that a small number of broken journeys can outweigh a large number of adequate ones.
For consumer demand planning, this changes the priority order. If the goal is to lift repurchase, fixing the worst journeys produces more movement than adding a new service channel. Most retailers measure average satisfaction, which hides exactly the tail that drives the loss.
What "hard to contact" actually means online
Contact difficulty is rarely a single broken link. It is usually a chain: the help entry point is buried, the chat bot loops, the form asks for an order number the customer cannot find, and the email reply arrives after the return window. Each step is survivable on its own; together they produce the 6.6 point loss.
The bot loop problem
Automated assistants reduce cost per contact but raise failure cost when they cannot escalate. A customer who repeats the same question three times to a bot is more frustrated than one who waited on hold, because the bot creates the impression of being ignored rather than queued.
Order data as the first fix
The cheapest improvement is usually making order context visible at the point of contact. If the agent or bot already knows the order, the customer skips the identification step, which is where most abandonment happens. This is a data integration task, not a staffing task.
Store and online service are one journey
For retailers operating both channels, the asymmetry compounds. A customer who buys online and returns in store experiences two service systems, and a failure at the handover is attributed to the brand rather than to either channel. That is why buy-online-return-in-store programmes live or die on the store side of the journey.
The channel mix in Asia makes this more acute, as brand builders balancing heritage with local relevance increasingly run marketplace, direct-to-consumer and physical stores in parallel. Each channel adds a handover point, and each handover point is a place where the 6.6 point loss can occur.
| Journey stage | Typical failure | Cost signal |
|---|---|---|
| Pre-purchase | Buried help entry | Abandoned sessions |
| Order issue | Bot loop without escalation | Repeat contacts per case |
| Return | Unclear policy or address | Dispute rate |
| Cross-channel | Store cannot see online order | Repeat visits |
How to measure the tail, not the average
Average satisfaction scores are structurally blind to the asymmetry. A retailer with 90 per cent good journeys and 10 per cent broken ones can post a healthy average while losing a meaningful share of repeat customers. The measurement fix is to report the distribution, not the mean.
Three practical metrics capture the tail: the share of contacts requiring more than one attempt, the share of returns completed without a second contact, and the repurchase rate of customers who had any service contact at all. The third is the closest proxy for the 6.6 point effect in a live business.
FAQ
Why does a bad service experience cost more than a good one earns?
Because a good experience confirms an existing expectation, while a bad one breaks it and is generalised to the whole brand. The study measured 6.6 points lost against 3.2 points gained.
Is the fix more staffing or better data?
Usually better data. Making order context visible at first contact removes the identification step where most abandonment happens, which is an integration task rather than a hiring one.
Why do average satisfaction scores miss this?
Because they blend the tail into the mean. Reporting the distribution, especially the worst decile, shows the failures that drive lost repeat purchase.
Does the cross-channel journey make it worse?
Yes. Every handover between online and store adds a point where a failure is attributed to the brand rather than to one channel.
