2026 Last Mile Bulky Item Report: Small loads on short notice

Someone needs a couch moved tomorrow. The pickup address is a third-floor apartment. The legs may need to come off. The customer cannot help carry it.

The item count is one. The job still needs a plan.

That gap between a small load and the work required to move it runs through Laborhutt’s 2026 request data. Half of the requests that explicitly named furniture or appliances listed one unit. Among requests with usable dates, 59.1% asked for service the same day or next day.

This report examines those requests alongside recorded prices, online booking behavior and delivery conversations. It looks at what customers ask for, what the structured fields capture, and what still has to be worked out before a vehicle arrives.

The database analysis covers January 1 through September 30, 2026. Records were extracted on October 14. PostHog and conversation samples have the separate reporting periods explained below.

These findings describe demand observed on Laborhutt. They are not national averages, a survey of all furniture buyers or a measure of completed delivery performance.

The findings at a glance

The figures answer different questions. A requested date does not establish availability. A quote does not establish payment. A reference to an elevator does not establish that a building has one.

Those distinctions matter when interpreting the results.

Half the named-item requests listed one unit

The core sample contains 459 requests naming at least one couch, mattress, bed, chair, table, dresser, desk, television, refrigerator or washer/dryer category. Those requests came from 408 customer accounts.

Of the 459 requests:

Unit” needs a careful definition. It is the quantity entered on the request. A washer/dryer category can represent a set, and a request naming furniture can also include boxes or other items.

The result is similar when repeat customers receive less weight. Looking only at each account’s first eligible named-item request, 209 of 408 accounts, or 51.2%, listed one unit.

The broader database sample contains 988 eligible requests, including generic size buckets and other loads. In that group, 526 requests, or 53.2%, listed one unit.

Generic entries help describe the size of the requested load. They cannot reliably tell us whether someone was moving a couch, an appliance or another object.

For operators, the finding points toward a service requirement: small loads deserve a booking process that captures the work around the item. Knowing that a customer has one couch does not establish the vehicle, crew or carrying time needed.

Couches were the most frequently named category

Couches appeared in 188 requests, making them the most frequently named category in the core sample. Mattresses appeared in 135.

Categories overlap. A request containing a couch and mattress appears in both rows, so the percentages do not sum to 100%.

These are platform observations, not estimates of national product demand. Laborhutt’s geographic concentration also affects the sample: New York and California account for 190 of the 459 named-item pickup locations, or 41.4%.

The results are unweighted. They should not be presented as a geographically balanced U.S. benchmark.

Most usable requests asked for service today or tomorrow

Among 457 named-item requests with usable dates, 177 asked for same-day service and 93 for the following day.

Together, that is 270 requests, or 59.1%.

Percentages may not total 100% because of rounding. Calendar dates use the pickup location’s recorded timezone. Two requests with dates outside the accepted range were excluded.

This measures the date customers submitted. It does not establish that a mover accepted the work, that a pickup window was agreed or that delivery occurred on that date. Defaults in the booking experience may also influence the selection.

Still, the concentration is useful. Much of the recorded demand arrived with little time between the request and the desired service date.

That creates a coordination problem. Vehicle availability, lifting help, item dimensions and building access may need to be resolved within hours.

The data do not tell us why each customer chose that date or whether they would have accepted a later one. They also do not establish what customers would pay for faster service.

A short drive does not describe the work at either address

The median straight-line distance between pickup and drop-off in the named-item sample was 8.0 miles.

More than half of requests, 251 of 459, had endpoints less than 10 miles apart. Approximately three quarters, 346 of 459, had endpoints less than 25 miles apart.

These are distances calculated from recorded coordinates. They are not road mileage, driving time or a carrier’s billed distance. The sample also includes longer-distance requests and requests with identical endpoint coordinates.

Distance describes the space between two addresses. It leaves out the work inside them.

Among 458 named-item requests with descriptions or additional details:

• 89 contained stairs-related terms.
• 71 mentioned an elevator.
• 136 mentioned either stairs or elevators.

Some requests mentioned both.

That means 29.7% of text-bearing named-item requests discussed access using those terms. It does not mean 29.7% required stair carrying or experienced an access problem.

The count includes statements such as having no stairs or no elevator. Some descriptions may also be generated. It measures what was recorded, rather than a verified building condition.

The distinction is practical. Two addresses can be close together while the job involves disassembly, a narrow landing, elevator access or carrying furniture upstairs.

Recorded local quote medians were $165

Pricing was analyzed using the recorded customer-facing quote total.

To prevent a request receiving many offers from dominating the result, the analysis retained the latest eligible quote per mover and request before the reporting cutoff, calculated a median for each request, and then summarized those request-level medians.

The local sample below contains requests with endpoints less than 25 straight-line miles apart. That is this report’s analytical definition of a local route, rather than a road-mile service boundary.

The couch sample is a subset of the first row. The middle 50% runs from the 25th to the 75th percentile.

These figures describe recorded offers from January through September. They include requests that later expired or were cancelled.

They are not paid-invoice averages, current bookable prices or a claim that every couch delivery costs $165.

Only 199 of the 346 local named-item requests had eligible quote records. The pricing sample therefore describes requests that attracted recorded offers, rather than everyone seeking moving help.

The work was also not standardized. Crew size, vehicle, stairs, service date and carrying requirements varied between requests.

Records flagged as customer proposals or AI-generated quotes were excluded. Newer system-origin quotes were analyzed separately because those requests used generic item buckets. Combining them with older named-item records would obscure a change in how the service was described and priced.

Conversations reveal what the item list leaves out

Structured records can tell us that a customer selected a couch and entered two addresses. The conversation may reveal that nobody has agreed who will carry it.

Laborhutt reviewed 24 GetStream threads linked to a deterministic sample from the broader 988-request cohort. They contained 60 nonempty text messages within the reporting period, including system notifications and templated introductions.

The examples below are paraphrases. They illustrate service details rather than measuring how frequently customers raise them.

The customer needed to know who was coming

In one furniture-moving thread, a customer asked whether the mover had an adequate vehicle and whether anyone else would be coming.

The customer then asked about couch disassembly and explained that the apartment was on the third floor without an elevator.

The item list alone did not resolve those questions. The customer needed agreement about vehicle suitability, crew size and the carrying work.

Another customer already had lifting help

In a washer/dryer thread, the customer said people were available to help move the items between garages and asked about a vehicle-only arrangement.

That is a different service from supplying a crew that handles all the lifting.

Both requests involve bulky items. They require different divisions of work.

A date still needed a pickup window

In another thread, participants discussed an evening pickup versus the following day.

That conversation added detail a requested service date could not convey. Before the move could be arranged, the participants still needed to agree when it could happen.

These examples suggest what a booking process should capture: who lifts, who supplies the vehicle, what must be disassembled, where the item must end up and which pickup window both parties can meet.

They do not establish how often these issues occur or whether the jobs were completed.

Online item submissions showed a similar small-load pattern

PostHog provides another view of expressed demand before a database job is necessarily created.

The analysis combined two item-submission events and used each tracked person’s first positive item count during September 11 through September 30, 2026.

After filtering for the production app, U.S. delivery-market tags, positive counts and known bot/internal activity, the sample contained:

The similar first- and last-submission results suggest the one-unit finding was not especially sensitive to which submission was selected in this sample.

The similarity to the database results is descriptive corroboration within Laborhutt’s product. It is not independent national validation.

Browser identifiers can represent anonymous users, the records may overlap with the database sample, and counts include generic size buckets.

Earlier item events exist from July 30, but no events before September 11 meet the complete U.S. market-tag filter used here. This filtered sample should not be presented as nine months of analytics coverage.

The wider market measures a different part of delivery

Armstrong & Associates, working with the National Home Delivery Association, estimated the U.S. third-party logistics big-and-bulky last-mile delivery market at $10.63 billion for 2025 and projected $11.42 billion for 2026.

Its definition covers bulky shipments from a final warehouse or terminal to the consignee. Laborhutt’s consumer requests also include person-to-person moves, in-home work and other routes.

Those populations differ. The figures cannot be combined to calculate Laborhutt market share. Source: Armstrong & Associates, July 20, 2026.

Earlier consumer research also distinguishes speed from a reliable appointment.

DispatchTrack’s survey of 1,606 American adults, conducted March 1 and 2, 2022, reported that 44% of bulky deliveries arrived outside the scheduled window and 31% of respondents considered early delivery inconvenient.

Those are historical survey findings from 2022, rather than 2026 rates or Laborhutt delivery outcomes. Source: DispatchTrack survey and methodology.

For researchers, the distinction matters. Asking for service tomorrow, securing a pickup window and receiving an item on time are separate events.

This report measures the first and uses conversations to illustrate the second. It does not measure the third.

How the report was built

Database selection

The October 14 export contained 2,078 Job records and 1,757 Quote records.

Selection used each record’s original creation timestamp, rather than the database timestamp associated with migration. There were 1,072 requests created from January 1 through September 30, 2026, UTC.

Requiring both endpoints to identify a U.S. state or the District of Columbia, while excluding explicitly non-U.S. countries, left 1,016 requests. All had positive listed quantities.

Known admin/company customer accounts and titles explicitly marked test, testing or dummy were checked. None of the 1,016 requests were removed by those checks. Anonymous or unmarked test activity may remain.

Repeated requests

Twenty-eight repeated records were removed, leaving 988 requests from 844 customer accounts.

A repeat was defined as the same customer account, exact endpoint coordinates, requested arrival timestamp and normalized item-type/quantity pairs. The earliest original request was retained. Quotes linked to removed repeats were mapped to that request.

This rule can merge genuinely repeated work with identical inputs. Conversely, changes in time, spelling or location can leave repeat attempts in the data.

Of the 988 retained requests, 459 named at least one of the furniture or appliance categories used in the core analysis.

Generic-only loads were excluded from that core sample. The named-item cohort therefore does not capture all bulky demand, particularly after the interface introduced size buckets.

Prices and distances

Eligible quotes had a positive finite customer-facing total and an original creation date inside the reporting period.

Customer-proposed and AI-flagged records were excluded. Records without a system-origin flag were treated as the legacy/mover pool, without claiming that every historical origin was independently verified.

Revised offers were reduced to the most recent eligible record per mover and retained request. System offers were reduced to the most recent per request and analyzed separately.

No completed-job restriction or arbitrary price trimming was applied. Medians and interquartile ranges limited the influence of extreme offers. Percentiles used linear interpolation.

Distances were calculated between recorded coordinates using a great-circle calculation. They are straight-line distances, not routed driving distances.

Conversations and product analytics

For GetStream, eligible channels were grouped to one channel per request using a deterministic hash ordering. Twenty-four requests were then selected using a second fixed ordering.

GetStream returned 61 messages. Sixty nonempty text messages fell within January through September. Up to the latest 100 messages per selected channel were requested.

Templates and system messages were included in the reviewed material. No customer names, contact information, exact addresses or verbatim customer messages are published.

PostHog used production-environment, app-hostname, U.S. market-tag and known-bot filters, along with event-time person properties to exclude known admins and Laborhutt email addresses. These checks cannot remove all unidentified staff, duplicated people or test activity.

The calculations are exploratory report definitions. Records were not joined across the database, PostHog and GetStream to estimate a conversion rate.

The source populations overlap and must not be added together. No statistical margin of error is provided because the data are not a probability sample.

The next measurement belongs at the pickup address

The strongest pattern in this sample is the combination of a small listed load and a short requested lead time.

The conversations explain why that combination still requires coordination. Someone must establish whether the vehicle fits, whether a second person is coming, whether furniture needs disassembly and whether the agreed service includes carrying it upstairs.

A stronger next study would record those conditions alongside accepted pickup windows and completed delivery times. That would allow researchers to examine how service scope relates to price, assignment and reliability.

For now, the evidence supports a narrower finding: many Laborhutt customers asked to move a small load soon, while the details of doing the work still needed to be settled.

Suggested citation: Laborhutt, 2026 Last Mile Bulky Item Report, January through September request data, prepared October 14, 2026.