SLA adherence looks strong. Parcels are arriving on time. Support ticket volume for "where is my order?" (WISMO) questions are low. And yet international repeat purchases are declining, and nobody on the operations side can point to what changed.
The dashboard isn't necessarily wrong. But it's measuring the shipment instead of the customer.
A parcel can meet every operational benchmark a carrier reports on and still be the final order a customer places with a brand. A successful delivery and a retained customer are two distinct outcomes, and most shipping KPIs were built to track only the first one.
Why operational reporting stops at delivery
Shipping metrics and dashboards were designed to answer one question well: Did the parcel arrive as specified? Carrier scorecards, SLA reports, and cost-per-shipment tracking all confirm that a defined operational commitment was met. None of those key performance indicators, however, were built to answer a different question that matters just as much to the business: What did the customer do after the parcel arrived?
That's not a criticism of the reporting. It's a description of its boundary. A carrier's job ends at delivery; a customer relationship (hopefully) doesn't. The measurement systems most brands rely on were designed around the carrier’s job, not the job left to whichever team happens to notice a retention gap months later, if at all.
There is also a timing problem. Shipping teams work within short time frames: An exception gets escalated, resolved, and closed, often within days. Customer behavior unfolds more slowly. A customer may not have another natural purchase opportunity for weeks or months. By the time a retention difference appears, the shipping incident that may have contributed to it has already disappeared from the operational queue.
Four KPIs that can hide silent churn
SLA adherence
Service level agreement adherence is measured against the carrier's contractual definition of on time, which may not match the delivery window the customer saw at checkout. Some carriers start the clock at first scan. Some measure against an internal benchmark that excludes certain markets or disruption periods. A carrier can report a strong SLA number while the customer's lived experience of that same shipment was slower, more confusing, or less predictable than what they were promised. The metric is accurate. It's just answering a narrower question than the one that predicts whether the customer buys again. This is one reason carrier evaluation needs to go beyond the headline SLA number; our 8 Questions to Ask Your International Carrier Before Peak Season walks through what to clarify before volume increases.
On-time delivery
On-time delivery confirms a parcel arrived within a stated window. As with SLA adherence, this metric says nothing about whether that window matched the promise made at checkout or how the customer felt in the run-up to delivery. A shipment can land inside the window, but if the tracking page sat frozen for more than a week, the customer might have experienced uncertainty the entire time, dissuading them from making a repeat purchase.
Cost per shipment
Cost per shipment stays stable right up until a disruption forces emergency rerouting at spot rates, and even then, the number describes what the business spent, not what the customer experienced. A program can hold cost per shipment flat for a full quarter while a subset of international customers quietly have a worsening delivery experience underneath that average. Cost discipline and customer experience are not the same thing, and tracking one closely doesn't tell you anything about the other.
WISMO ticket volume
WISMO ticket volume requires the customer to complain. Silent churn is defined by the customers who don't. A brand that watches WISMO volume stay low can read that as a sign the program is healthy, when it may simply mean that customers who had a bad experience decided it wasn't worth the effort to say so. That gap between operational confidence and customer knowledge is where silent churn lives.
What the aggregate hides
A company-wide metric can look healthy while the performance within a specific cohort is not. This is the part most operational reporting misses entirely: Averages flatten exactly the variation that matters.
According to ePost Global's 2025 International Shipping Insights and Trends Report, in high-risk destination markets, Delivered Duty Paid (DDP) shipments were over 30 times more likely to clear customs and be delivered successfully than Delivered Duty Unpaid (DDU) shipments, also referred to as Delivered At Place (DAP). A brand's overall delivery success rate can look strong while DDU customers in high-duty markets are having a materially worse experience, one that's large enough to matter but invisible in the aggregate number.
The same flattening happens with repeat purchase data. A company-wide repeat purchase rate can hold steady despite a narrow but valuable segment of international customers underperforming underneath it. This segment might consist of customers in a particular destination market, customers acquired during a disruption period, customers who received a delayed shipment, or those surprised by duties. Rather than ask whether international repeat purchase declined overall, determine whether customers who went through a materially different delivery journey behaved differently afterward than otherwise comparable customers did. If you’re trying to determine whether that disconnect is already happening, start with our post 5 Signs Your International Shipping Program Has a Shadow Loss Problem.
That comparison matters. A late shipment followed by a lost customer does not, by itself, prove that shipping caused the churn. Destination market, order value, product category, acquisition timing, duty treatment, and normal reorder cycle can all influence retention. The goal is to identify whether an exposed group behaves differently enough from a reasonable comparison group to warrant attention.
The metrics that reveal what happened after delivery
If shipment-level metrics answer whether the parcel arrived, a different set of metrics is needed to answer whether the customer relationship survived the delivery. Four are worth building into any international program.
Promised-to-actual delivery variance. This isn’t the carrier's own SLA adherence. Instead it’s the gap between the delivery window shown to the customer at checkout and what actually happened. This metric is tied to how the customer experienced the promise, not how the carrier measured its own performance against it.
Repeat purchase rate by delivery experience. Compare customers whose first international shipment arrived as promised against those who ran into a delay or a duty surprise. The size of that gap shows whether customers exposed to a worse delivery experience are behaving differently afterward, and if so, whether the difference is large enough to investigate.
Time to second purchase by delivery cohort. A degraded delivery experience doesn't always produce immediate churn. It can simply push the second order out past when it was expected, quietly lengthening the sales cycle for that cohort without ever showing up as a lost customer on paper.
Silent failure rate. This assesses how many customers who had a measurable delivery problem never generated a support ticket. It’s the metric that most directly measures silent churn itself, and building it requires deliberately joining delivery data to support data rather than assuming a clean SLA report means nothing went wrong.
How to connect shipping data to customer data
Most brands discover the gap between shipping data and customer data well after silent churn has begun. That’s because they fail to ask and answer the question “Of the customers who had a poor international delivery experience in the past 12 months, what percentage placed a second order?”
The data does exist, but it lives in disparate systems that were never built to talk to each other. Delivery outcome sits in the shipping platform. Purchase behavior sits in the eCommerce platform or the CRM. Logistics can see the disruption. Customer service can see the complaints that got filed but not the ones that didn't. Nobody owns the joint between the two.
Building that connection doesn't require new data collection. It requires pulling delivery outcomes (on time, late, duty-surprised, missing) and joining it to 90-day repeat purchase rate by cohort, using data the business already has sitting in separate systems. Once that join exists, the gap between operational success and customer retention stops being a hypothesis and becomes a number a team can act on.
The strongest test is simple: Can you identify customers who received a materially different shipping experience, compare them with similar customers who did not, and follow both groups through the relevant next-purchase window?
If you can't, you might know exactly how the shipping incident ended without knowing whether the customer relationship ended with it.
A successful delivery isn't necessarily a retained customer
At ePost Global, we treat the connection between delivery experience and retention as something to monitor continuously, not something to audit after a quarter goes badly. Delivery performance data on its own tells a brand whether its network is working. It doesn't tell a brand whether its customers are coming back, and in our work with international brands, those two answers diverge more often than operational reporting alone would suggest.
A clean SLA report is not proof that an international program is healthy. It's proof that the carrier did its job. Whether the customer relationship survived that delivery is a separate question, and most reporting stacks were never built to answer it.
The starting point is knowing where the disconnect might already be. Take the Cross-Border Optionality Check to see where delivery experience and customer retention have started to diverge in your own program.
For the full model connecting shipping structure to customer loss, see the Shadow Loss Chain. For the underlying concept, see Shadow Loss.
FAQ: Shipping KPIs and customer retention
Why can my SLA and on-time delivery numbers look fine while international repeat purchases decline? Delivery numbers can look fine despite declining international repeat purchases because they measure whether the carrier met a contractual delivery commitment, not whether the customer's experience matched what they were promised or whether they came back afterward. A shipment can hit every operational benchmark and still be the last order a customer places with a brand.
What is a silent failure rate, and how do I measure it? A silent failure rate is the share of customers who had a measurable delivery problem, such as a delay or a duty surprise, but never generated a support ticket about it. Measuring it requires joining delivery outcome data to support ticket data so that you can see the customers who had a problem and said nothing.
How do I connect delivery data to customer retention data? Pull delivery outcome by shipment (on time, late, duty-flagged, missing) and join it to repeat purchase rate by cohort, using a follow-up window that reflects the business's normal purchase cycle. Most brands already have both data sets. The gap is that they sit in separate systems that were never built to talk to each other.
What's the difference between shipment-level and customer-level shipping metrics? Shipment-level metrics, such as SLA adherence and cost per shipment, measure what happened to the parcel. Customer-level metrics, such as repeat purchase rate by delivery experience, measure what happened to the person who received it. Both matter, but only one of them shows the effect of the delivery experience on the customer relationship.
Can a shipping program look successful and still be losing customers? Yes. That's the entire premise of silent churn. A program can post strong delivery metrics quarter after quarter while a specific cohort of customers, often concentrated in a particular market or delivery outcome, quietly stop reordering. The aggregate numbers never catch it, because they were never designed to look for it.
