Hiring
How many people does your ecommerce operation actually need?
Ecommerce teams are sized by anecdote — one overworked generalist or a headcount copied from a bigger competitor. The four scaling drivers that actually set team size, the sequence roles should be added in, and the honest signs you are one person short.
Ecommerce operations teams are almost never sized; they accrete. A founder does everything, then one 'ecommerce person' inherits everything, then a second hire arrives two years after the first one stopped coping. Ask how the headcount was chosen and the honest answer is usually 'it's what we could afford' or 'it's what we've always had' — neither of which has anything to do with the actual driver: how much operational surface your catalog and channels generate every week.
Sizing this properly is literally our business — structuring teams is what we sell — so treat this as our sizing model shown in public. It is the same logic behind the interactive team-builder on our homepage, and the same logic that produced a 14-person team for one 16,000-SKU brand: not ambition, arithmetic.
Key takeaways
- —Four drivers set the workload: catalog size and type, channel count and strictness, order and enquiry volume, and the content standard you sell at
- —One-of-a-kind inventory changes the math completely — per-piece research time means catalog size drives headcount linearly, not logarithmically
- —Channels multiply rather than add: each marketplace brings its own weekly rhythm, and five channels is five rhythms even at identical sales
- —Roles should be added in sequence — generalist, then catalog specialist, then channel owners, then content, then a lead — not by cloning the overloaded generalist
- —The understaffing signals are behavioural, not numerical: maintenance postponed, channels unopened, one person unfireable and unhoidayable
The four drivers
1. Catalog: size × type
A thousand programmatic SKUs from clean supplier data is a feed problem — one person maintains it comfortably. A thousand one-of-a-kind pieces is a research operation: at an hour-plus per properly catalogued piece, intake alone is a full-time function before anything is maintained. Mixed catalogs need the split stated: the programmatic side scales with tooling, the singular side scales with people, and averaging them produces a team that is wrong for both.
2. Channels: count × strictness
Every live channel has a floor cost in attention regardless of its sales: Wayfair's supplier rhythm, the design marketplaces' enquiry and offer culture, Etsy's listing cadence. Strict platforms carry higher floors — a Wayfair or Amazon presence generates tickets, suppressions and data demands weekly whether you sold or not. The honest budgeting unit is channel-rhythms per week, and the common failure is opening a sixth channel with a team already rationing attention across four.
3. Volume: orders, returns, enquiries
Transaction-driven work scales with buyers, not catalog: order issues, freight exceptions on oversized goods, returns processing, and — in the design tier — pre-sale enquiries and offer negotiation that are effectively a sales function. This is the driver founders undercount, because at low volume it hides inside everyone's margins; the day it stops hiding, it takes catalog maintenance hostage first.
4. Standard: the quality bar you sell at
The same SKU costs different labour at different standards — adequate-for-Amazon imagery versus luxury-banner editorial, minimum-viable listings versus the condition-reporting standard curated venues expect. The standard is a legitimate choice; costing it honestly is not optional. Premium positioning bought without premium content labour is how brands end up visibly cheap on their most expensive shelf.
The sequence roles should be added in
- —One generalist: viable while catalog intake is light and channels are one or two — the configuration most companies stay in two years too long
- —Catalog specialist second: data quality is the foundation every other function stands on, and it is the first thing a drowning generalist silently drops
- —Channel ownership third: once a strict marketplace matters commercially, it needs a named owner who knows its rhythm — shared ownership is how suppressions sit unnoticed
- —Content next where the standard demands it: photography coordination and listing copy as a function, not a favour
- —A working lead at four to five people: review, prioritisation and cover stop being ambient and become a job
- —Named backups throughout: any account only one person can run is a resignation letter away from being nobody's
The understaffing signals
Teams rarely announce they are too small; the work announces it. Maintenance becomes episodic — audits happen after incidents rather than on calendars. Growth work stalls — the new channel stays 'next quarter' for four quarters, because the team's whole capacity is absorbed keeping today running. Quality quietly floats down to whatever the busiest week permits. And one person becomes load-bearing: unhoidayable, unpromotable, unfireable, the single point through which three channels' knowledge flows. Each signal is also the cheap moment to act — the expensive moment is the incident that finally makes the case, on the platform's timeline instead of yours. The fully-loaded arithmetic of the next hire almost always beats the cost of the incident.
Size the team from catalog type, channel rhythms, transaction volume and the standard you sell at — then add roles in sequence with named ownership and backups, before the overloaded generalist becomes the incident that sizes it for you.
Frequently asked questions
It follows four drivers, not revenue: catalog size and type (one-of-a-kind pieces cost research time per unit), number and strictness of channels (each has a weekly floor cost in attention), order and enquiry volume, and the content standard you sell at. One generalist holds one or two channels and light catalog intake; five channels of one-of-a-kind inventory at a luxury standard is a specialist team by arithmetic, not ambition.
After the founding generalist: a catalog specialist. Data quality is the foundation every channel stands on and the first duty an overloaded generalist silently drops. Channel ownership comes next once a strict marketplace matters commercially, then content as a function, then a working lead around four to five people.
Behavioural before numerical: maintenance happens after incidents instead of on calendars, the new channel stays 'next quarter' indefinitely, quality floats down to what the busiest week permits, and one unhoidayable person carries knowledge nobody else has. Each is a cheaper intervention point than the incident that eventually forces the hire.
Yes, structurally: every piece needs attribution research, condition documentation and photography direction — an hour or more of skilled work per listing — so catalog headcount scales with intake volume rather than flattening the way programmatic catalogs do. It is why a 16,000-piece one-of-a-kind operation can honestly need a 14-person team.