If you have read anything about post-purchase experience in the last decade, you have seen a WISMO statistic. It is usually a clean round figure, presented without hesitation, and attributed to a large and respectable-sounding name. We tried to follow those attributions back to primary research. For the commonly repeated figures examined here, we could not locate a published underlying study with a sample, instrument and methodology.
What WISMO is and why the number matters
WISMO stands for "where is my order". It covers every inbound contact asking about the whereabouts, status or expected arrival of an order already placed, by email, chat, phone, social, marketplace message or self-service ticket. It is the most repetitive category of contact most eCommerce support teams handle.
The number matters because it is load-bearing. It sizes support teams, justifies tracking software, sets deflection targets and argues for logistics investment. Vendors quote it in pitch decks, buyers quote it back in business cases, analysts quote it in market sizing. If it is wrong, everything stacked on it inherits the error.
Ship24 sells tracking infrastructure, so a large WISMO number is commercially convenient for us. That is precisely why it is worth saying plainly that we do not know what it is, and neither does anyone else.
The figures in circulation, and where the trail leads
The claims cluster into three bands, which is itself revealing:
- WISMO accounts for 10 to 25% of customer contacts.
- WISMO accounts for 25 to 40% of inbound support volume.
- WISMO reaches 50 to 60% during peak season.
These are attributed to names including ClickPost, Shopify, Salesforce and ReadyCloud. Follow any of them and the pattern repeats. The citation leads to a blog post, which either states the figure with no citation or cites another blog post, which cites a third, which cites the first. Nowhere in the chain is there a study: no sample size, no survey instrument, no time period, no merchant profile, no definition of what was counted.
To be fair to the companies named, most are not claiming to have run a study. They are writing marketing content and reaching for a figure already in the air. The problem is not fabrication, it is circulation: a number repeated often enough acquires the appearance of evidence without acquiring the substance.
What we are not saying: that WISMO is small, or that it does not matter. Anyone who has run a support queue knows the volume is real and spikes when delivery performance dips. The point is that we could not find a defensible published measurement of its size in the sources reviewed.
Why missing methodology is not a technicality
The plausible answers span a range so wide that the question is meaningless without the fine print.
The definition of "contact"
Ticket, conversation, message, session or customer? One customer who emails, then chats, then calls about the same parcel can be counted as one or four. A bot session that resolves without a human can be counted or excluded. Teams counting deflected self-service lookups report a far higher WISMO share than teams counting only human-handled tickets, with no difference in the underlying reality.
Channel scope
WISMO is concentrated in low-effort channels. Customers who would never phone about a parcel will happily send a one-line chat message. A measurement including chat and social shows a higher share than one built on email and phone. Neither is wrong. They measure different things.
Season and merchant profile
Peak and off-peak are not variations around a mean, they are different regimes: transit times lengthen and question volume rises while other contact types stay flat. Order value, category, delivery speed, international share and carrier mix move the number too. Blend the seasons and aggregate the merchants and you get a figure describing no actual merchant at no actual time of year.
The denominator
The quietest problem and the largest. A percentage of contacts is not a percentage of orders, yet both get quoted as "the WISMO number". They can differ by an order of magnitude and answer different questions. Sizing a support team needs the first. Judging whether your logistics is causing the contacts needs the second.
The same problem, in the statistics next door
Estimated delivery date accuracy and conversion. Shoppers demonstrably care about delivery dates. Narvar's 2025 State of Post-Purchase research, a vendor-run survey of 3,461 US online shoppers, found 73% say a visible estimated delivery date influences purchase decisions and 40% will not buy at all when no date is shown. Baymard Institute, independent and large-sample, finds "delivery was too slow" accounts for 21% of checkout abandonments, behind extra costs at 39%. What does not exist is a controlled study of the effect of accuracy specifically. The circulating figures are vendor case studies, such as a "13-point conversion lift", or unverifiable attributions like a "Forrester 13 to 25%". One merchant's before-and-after is not a controlled result.
The "41% of major US checkouts show only a shipping speed" claim. Attributed to Baymard by third parties, but we could not locate it in Baymard's own published material. That does not make it false, but an attribution you cannot verify in the source is not usable.
Delivered but not received. No carrier or regulator publishes a rate for parcels marked delivered that never arrive. Every circulating figure comes from consumer self-report, which cannot separate theft, misdelivery, premature scanning or a parcel left somewhere unexpected. That is evidence of consumer experience, not a carrier performance metric.
Last-mile cost. Claims such as "53% of shipping cost" and "$17.20 per failed delivery" circulate without traceable methodology, and should not be repeated as fact.
A worked example: what transparency looks like
Package theft shows the difference cleanly, because here one source does the work properly and others do not.
Security.org's 2025 annual report surveyed 3,307 US adults in October 2025, census-weighted on age, sex and ethnicity. The methodology is stated. It finds 1 in 4 Americans, roughly 64 million people, have ever had a package stolen, and 3%, roughly 9 million adults, in the last three months. It estimates 37 million packages stolen per year at roughly $8.2 billion, average value $222, up around 10% year on year.
The other 2025 estimates are not refinements of it. They differ from Security.org and from each other by roughly four to six times.
| Estimate | Packages stolen per year | Value |
|---|---|---|
| Security.org 2025 (n=3,307, census-weighted, methodology published) | ~37 million | ~$8.2 billion |
| Omnisend 2025 | 228 million | $12.8 billion |
| SafeWise 2025 | not stated in the same terms | $37 billion |
These cannot be reconciled by argument, only by methodology. Packages or incidents? Households or adults? Ever, last year, or last quarter? Declared value, replacement cost or retail price? Weighted or not? Each choice can move the answer several-fold, which is exactly the spread observed.
To be fair to the other two: we were able to locate Security.org's sample, weighting and field dates, and we were not able to locate equivalents for the others. That is a statement about what is readily findable, not a claim that no methodology exists. The point stands either way. Where the working is not published alongside the number, a reader cannot locate the disagreement, let alone resolve it.
This is the WISMO problem with the numbers filled in, except that package theft has at least one credible measurement to anchor against. WISMO has nothing.
What a defensible WISMO measurement would require
| Circulating claim | What can actually be verified |
|---|---|
| WISMO is 10 to 25% of customer contacts | No primary source. No definition of "contact" |
| WISMO is 25 to 40% of inbound support volume | No primary source. No channel scope, no denominator |
| WISMO hits 50 to 60% at peak | No primary source. No period definition |
| Attributed to ClickPost, Shopify, Salesforce, ReadyCloud | Attributions lead to blog posts, not studies |
| Accurate EDDs lift conversion by a stated amount | No controlled study. Vendor case studies only |
| "41% of US checkouts show only a shipping speed" | Attributed to Baymard by third parties, not in Baymard's material |
| A carrier-side delivered-but-not-received rate | Does not exist. All figures are self-report |
| Last mile is 53% of cost; $17.20 per failed delivery | No traceable methodology |
| Delivery too slow drives 21% of checkout abandonments | Verifiable. Baymard, independent, large-sample |
| 74% of US shoppers had a late delivery last year | Verifiable. Narvar 2025, n=3,461, vendor-run, disclosed |
| ~37 million US packages stolen a year, ~$8.2 billion | Verifiable. Security.org 2025, n=3,307, census-weighted |
A publishable WISMO measurement needs six things stated up front:
- Definition. What counts, including edge cases: pre-dispatch enquiries, delivery-date changes, returns-in-transit questions, repeat contacts on one order.
- Denominator. Contacts, orders, shipments or customers. State which, and do not switch.
- Channel scope. What is in, what is out, and whether self-service and bot-resolved sessions count.
- Period. Exact dates, with peak and off-peak reported separately rather than blended.
- Merchant profile. Size band, category, cross-border share, delivery speed and carrier mix.
- Sample. How many merchants, how selected, how representative. A figure drawn only from merchants who already bought post-purchase software describes that population, not eCommerce.
Anything short of that is an anecdote with a percentage sign on it, which is not disqualifying, but should be labelled.
Conclusion: what the industry should do about it
Three changes, none of them difficult:
- Stop laundering figures through citation. Link to the study, not to the blog post that linked to the blog post. If there is no study, say the number is an estimate and say whose.
- Publish your own data with its methodology attached. Vendor-run research is legitimate, and Narvar's post-purchase work is a fair example: it says it is vendor research, states its sample and dates, and can therefore be weighed. Vendor research is a problem only when disguised as independent or published without its workings.
- Report ranges and conditions, not single figures. WISMO share almost certainly varies by merchant, channel and season. A range with stated conditions is more honest than a false point estimate.
There is a business reason to care beyond tidiness. Pitney Bowes' 2026 US Parcel Shipping Index recorded 23.1 billion US parcels in 2025, with the fastest growth by far in alternative carriers such as OnTrac, Veho, UniUni and SpeedX, up 127% to 1.8 billion parcels, and it names delivery tracking and reliability as that segment's weak point. There is still no common data model to lean on: UPU standards such as S10 and EMSEVT cover the postal network only, and the large private carriers each use proprietary event codes. More parcels, more carriers and weaker tracking in the fastest-growing segment is a recipe for more WISMO, not less.
We are publishing this because we intend to measure WISMO properly, not to score points off other vendors. The companies whose posts carry these figures are mostly repeating something they inherited, and the phenomenon they describe is real. But an industry running on a number nobody has measured is deciding on vibes, and the tooling business, ours included, has benefited from that. When Ship24 publishes a WISMO figure, it will come with the definition, the denominator, the channel scope, the period, the merchant profile and the sample. Until then, the honest answer to "what does WISMO cost?" is that no defensible industry-wide figure has been published in the sources we reviewed.
Sources & methodology
- Universal Postal Union (UPU) Standards programme and catalogue Primary source.
- Narvar State of Post-Purchase 2025 Company research.
- Baymard Institute Cart abandonment research Independent research.
- Security.org Package Theft Annual Report 2025 Survey research.
