E-Commerce On-Demand Delivery Expansion Reshapes Last-Mile Logistics Expectations

The Analysis

Ulta Beauty flipped the switch on nationwide same-day delivery through Uber Eats on June 1, stacking it on top of partnerships it already runs with Instacart and DoorDash. Read that as a signal, not a press release: on-demand delivery through gig platforms has become table stakes in consumer retail, and the bar just moved for everyone selling beauty, health, and high-repurchase goods.

For a brand, the scary part is not the headline — it is that you are now expected to be on three or more instant platforms at once, and the moment you are not, a competitor one tap away is already taking the order.

What actually happened

Two things changed on June 1, and only one is in the press kit. The visible move is Ulta adding Uber Eats to an existing Instacart and DoorDash roster, so a single national beauty retailer now sells instant delivery through three gig networks. The quieter move is the category line itself: rapid delivery, once a food-and-grocery habit, has spilled into beauty, health, and anything people reorder. That spill is the part that rewrites your operating assumptions, because it reaches SKUs you never built a single-unit fulfillment flow for.

Why this is happening now

The gig platforms are no longer experiment channels; they are mature delivery rails with spare capacity and a consumer app already open on the phone. Retailers chased food speed for years, trained shoppers to expect one-hour arrival, and those shoppers carried the expectation straight into the next category they buy. Ulta is not inventing demand, it is meeting a behavior that already exists, which is exactly why rivals cannot ignore it once one large player normalizes it.

Whom this hits, and by how much

The largest retailers absorb multi-platform selling as a line item — they have teams, systems, and volume to spread the cost across a huge base. Mid-size and emerging brands feel it as a forced entry fee: join or watch share leak to a competitor one tap away. The operators squeezed hardest are the fulfillment houses and brand warehouses behind the listing, because every one of these orders is a single-unit pick, not a case, and single-unit work scales labor faster than it scales revenue. Beauty and health SKUs also carry higher return and damage sensitivity, so the pick-pack error budget shrinks while the order count climbs, a nasty combination for any warehouse already at capacity.

When the pain reaches your cost and inventory

The cost shows up immediately on the first order, but the damage shows up in weeks, not days. The failure mode is inventory visibility: when the same stock is promised across three apps with different sync cycles, one platform oversells while another sits idle, and the brand eats the cancellation or the rush. Expect the break to surface within two to four weeks of going live if the integrations are not built for near-real-time. The lead time to fix it is the build time, so start before you launch, not after the refunds start arriving.

The point almost nobody is saying out loud

Everyone talks about speed, but the real tax is promise accuracy across platforms. A same-day badge is only as good as the inventory behind it, and three apps means three chances to lie to a customer in the same afternoon. The overlooked move is not 'get on Uber Eats' — it is building one inventory truth that all platforms read, and capping advertised speed to what your worst-synced channel can honor. Brands that promise one hour on a feed that updates every twenty minutes will refund more than they sell, and the refund hits the same single-unit margin that was already thin.

What would flip this call

If gig-platform capacity saturates during a peak window, promise times stretch and the table-stakes advantage thins, which would take pressure off late joiners. If consumer expectation snaps back toward a two-day norm, the forced entry fee loses its teeth. Neither is likely this year, so plan as if instant is permanent and treat any relaxation as a margin gift you did not bank on.

The practical read for a brand is that speed is now a cost center you must price explicitly, not a feature you give away to match a rival. The mistake most teams make is treating on-demand as a marketing line and letting fulfillment eat the margin in a different budget, so the true cost never surfaces until the refund column grows. The fix is to move single-unit economics onto the same page as the growth target: if a platform promises one-hour delivery, the inventory behind that promise has to be funded as part of the listing, not discovered after the oversell.

The brands that stay profitable are the ones that cap advertised speed to what the network can honor and let the slower channels carry the overflow, rather than promising the fastest badge on every app and paying for it in rush substitutions. Pick the platforms where the volume actually lives and treat the rest as spillover, because spreading thin across three apps multiplies the sync problem without multiplying the orders.

A quick number to ground it

Assume a beauty brand runs 500 active SKUs and joins three instant platforms, each promising same-day on 80% of the catalog. If real-time sync only covers 70% of stock and the three apps each pull from that pool, the overlap oversell risk lands on roughly 30% of 1,200 daily orders, or about 360 orders a day that need a fix, a cancel, or a rush substitution — at single-unit pick-pack cost on every one, before a single refund is paid.

↑ Back to the key points

— By Nadia Pryce