Skip to content
Ashish Kumar AgnihotriOperating Partner

06Industry

E-commerce & D2C brands

An e-commerce operation does not scale the way its revenue does. Every new SKU, channel and delivery promise multiplies the ways an order can go wrong, and the model that coped at a few hundred orders a day quietly breaks at a few thousand. The fix is not more effort at the pack station — it is an operating model built for volume.

Growth in e-commerce is multiplicative, but most operations are built additively. Each new SKU adds picking complexity, each new channel adds its own rules and returns logic, each faster delivery promise tightens every deadline upstream — and the combinations multiply against each other. An operation that ran comfortably at a few hundred orders a day starts missing at a few thousand, not because anyone got worse at their job but because the number of ways an order can go wrong has multiplied past what informal coordination can hold. The founder feels it as a stream of unrelated fires: a mispick here, a late dispatch there, a marketplace penalty nobody saw coming. They are not unrelated. They are one operating model quietly exceeding its design limit.

The margin story is quieter and worse. Returns, refunds, failed deliveries and marketplace penalties rarely appear as one number anyone owns — they are scattered across finance, operations and channel dashboards, so the business sees revenue growing while the cash it actually keeps per order shrinks. Add the settlement lag on marketplaces, refunds paid out faster than they are recovered, and stock sitting in the wrong warehouse, and a profitable-looking brand can be permanently short of cash. Meanwhile fulfilment quality holds because a few people care enough to catch what the process misses. That works until volume rises, a peak season lands, or one of those people leaves — and then the error rate the process was always capable of shows up all at once.

I will be straight about what I am and am not. I have not run an e-commerce company, and I will not dress my background up as category experience. What I bring is the operating disciplines this sector runs on — throughput at volume, quality built into the flow, cash cycles compressed by design — proven at enterprise scale over 19 years: a billing cycle cut from roughly two months to fifteen days across 75 entities, quality lifted from 95% to 99% across more than 2,000 campaigns. The problems are structurally the same: high-volume flow, small errors that compound, cash trapped in a slow cycle. The disciplines transfer. The category knowledge — your products, your customers — stays yours.

What tends to break

  • Operational complexity compounds with every new SKU and channel — until the model quietly breaks.
  • Returns and refunds eat margin invisibly, scattered across dashboards nobody reconciles.
  • Fulfilment quality depends on a few heroic people, not on the process.
  • Marketplace, own-site and quick-commerce channels sprawl into separate operations.

How I help

  • Map the order-to-cash flow end to end and put numbers on where margin and days leak.
  • Build quality into fulfilment so accuracy stops depending on who is on shift.
  • Standardise the core operation across channels and adapt only the edges.
  • Install the weekly cadence and scorecard that keep throughput, returns and cash visible.

Sound familiar?

01

Revenue is growing and cash is somehow tighter every quarter.

02

Peak seasons are survived on adrenaline, not run on a plan.

03

You know your GMV to the decimal — and don’t quite trust your margin per order.

The fit

Process Transformation & Lean Six Sigma

Rebuilding how the work flows — measured, not just reorganised.

Book a consultationExplore the service

In depth

The operating detail for this sector.

Complexity compounds; capacity only adds

The trap in scaling an e-commerce operation is that its two curves grow at different rates. Capacity grows additively — another packer, another shift, another warehouse. Complexity grows multiplicatively — SKUs times channels times delivery promises times return paths, every new element combining with all the existing ones. For a while the gap is invisible, absorbed by good people working around the edges. Then it is not: mispicks rise, dispatch cutoffs slip, penalties arrive, and adding more people no longer moves the numbers because the constraint is coordination, not hands. The first thing I do is map the real flow of an order — every touch, queue and decision from click to doorstep — and put numbers on where time and accuracy actually leak. The redesign starts there, at the structure that has to carry the volume, not at the pack station.

Returns are an operations number wearing a finance disguise

Most brands treat returns as a cost of doing business — a percentage to be tolerated and provisioned for. Operationally, that is a surrender. A return has a cause: the wrong item was picked, the product arrived damaged, the description over-promised, delivery came too late to matter. Each cause is measurable, each is addressable, and each behaves differently by channel and by SKU. The work is to stop managing the blended rate and start segmenting the reasons — then design out the largest recurring causes one at a time, the same way any defect stream is attacked. Every return prevented is pure margin recovered: the product cost, the two-way logistics, the refund float, the customer relationship. A brand that measures returns by cause usually finds that a large share traces to a handful of fixable operational defects — not to fickle customers.

Fulfilment that survives its best people leaving

In most scaling brands, the honest answer to "why is our fulfilment accuracy good?" is a short list of names — the warehouse lead who double-checks the odd-looking orders, the ops manager who knows which SKUs get confused. That is quality by heroics, and it has a ceiling: heroes do not scale, they burn out, and they take the standard with them when they leave. The alternative is to make accuracy a property of the process — pick paths and packing checks designed so the easy mistake is hard to make, exceptions routed to defined owners rather than to whoever notices, and a small set of quality signals reviewed at a fixed rhythm. This is the discipline that lifted quality from 95% to 99% across more than 2,000 campaigns: not more inspection, but a flow that produces the right result when nobody special is watching.

Channel sprawl: one operation wearing many uniforms

Marketplaces, your own site and quick commerce do not just add volume — each arrives with its own SLAs, penalty regimes, packaging rules, returns logic and settlement rhythm. The common failure is to let each channel grow its own informal operation, until the same product ships three different ways and no number is comparable across them. The discipline is the same one that standardises any multi-entity process: separate the core that should be identical everywhere — inventory truth, pick-pack quality, dispatch cadence, exception handling — from the edges that genuinely must differ by channel, and standardise the first ruthlessly while respecting the second. One operating model with defined channel adaptations is measurable, improvable and scalable. Three accidental operations are none of those things, and every new channel added onto them multiplies the confusion rather than the reach.

Cash is a cycle, and cycles can be compressed

E-commerce cash disappears into the gaps between events: the days between dispatch and marketplace settlement, between a return landing and the refund being recovered or restocked, between buying inventory and selling through it. Each gap looks small; multiplied by volume they decide whether growth feeds the business or starves it. The discipline that compresses them is the one that took a billing approval cycle from roughly two months to fifteen days across 75 entities: map the end-to-end cycle, interrogate every step for whether it adds control or only delay, parallelise what was needlessly serial, and reconcile the numbers so finance and operations are looking at the same truth. The aim is a cash cycle that is designed, measured weekly and owned — not an accident of whatever each channel and warehouse happens to do.

What I bring here — and what I do not

I am not the person for your merchandising, your growth marketing or your brand. I have not run a D2C company, and if what you need is category instinct — which products, which price points, which creative — that is not the seat I take, and I will say so in the first conversation. Where I earn my keep is the operating machine underneath: fulfilment that holds its accuracy at volume, returns attacked by cause, channels run as one measurable operation, a cash cycle that stops leaking days. Those disciplines are proven — from 95% to 99% quality across more than 2,000 campaigns, a two-month cycle cut to fifteen days — and they transfer, because the structure of the problem is the same even when the product is different. You keep the category judgement. I build the system that lets it scale.

Questions

Common questions.

Because complexity grows multiplicatively while capacity is added linearly. Every new SKU, channel and delivery promise combines with all the existing ones, multiplying the ways an order can go wrong — and at some volume that passes what informal coordination and good intentions can hold. The breakage shows up as seemingly unrelated fires: mispicks, missed cutoffs, marketplace penalties. They are usually one problem — an operating model past its design limit. The fix is structural: map the real flow of an order, find where time and accuracy leak, and redesign the model to carry the volume it now faces.

Stop managing the blended return rate and start segmenting the causes. A return happens for a reason — wrong item picked, damage in transit, an over-promising listing, late delivery — and each cause is measurable and fixable. Attack the largest recurring causes one at a time, the way any defect stream is reduced, and measure the movement. Every prevented return recovers the product cost, the two-way logistics, the refund float and often the customer. Most brands that do this find a handful of operational defects driving a large share of returns — which is far better news than blaming customer behaviour.

Almost never first. The instinct to switch providers or platforms usually arrives before anyone has mapped the process the tools are meant to serve — and a new system layered on an unexamined flow tends to automate the same queues and defects at higher cost. Fix the operating model first: the pick-pack flow, the exception handling, the channel rules, the reconciliation. Then judge the 3PL and the software against a process that is actually defined. Sometimes the provider does need to change; more often the relationship improves because you can finally specify, measure and hold what you need from it.

The mechanics are the same even though the setting differs. Compressing a billing approval cycle from roughly two months to fifteen days across 75 entities came from mapping the end-to-end flow, deleting steps that added delay without adding control, parallelising what was needlessly serial and reconciling the numbers into one trusted view. An order-to-cash cycle responds to exactly that treatment: settlement lags, refund recovery, inventory sitting in the wrong place and unreconciled channel reports are all delay that can be designed out. The product is different; the discipline — find the days, remove the ones that protect nothing — is identical.

No, and I will not pretend otherwise. My 19 years are in operations, quality systems and process transformation at enterprise scale — most recently leading Business Excellence at Publicis Groupe, across 500+ clients and USD 750M+ in media spend. What that built is deep command of the disciplines e-commerce lives or dies by: throughput at volume, quality that holds without heroics, cash cycles compressed by design. Those transfer directly. What I do not bring is category judgement — merchandising, pricing, growth creative — and I will not take a seat that needs it. The honest offer is the operating system, not the category instinct.

Yes — if you separate the core from the edges. Inventory truth, picking accuracy, packing quality, dispatch discipline and exception handling should be identical regardless of where the order came from; SLAs, packaging rules and returns logic genuinely differ by channel and should be handled as defined adaptations, not separate operations. The failure mode is letting each channel grow its own informal process until nothing is comparable and every improvement has to be made three times. One standardised core with explicit channel edges is measurable and scalable — and it makes adding the next channel an extension, not another rebuild.

Fewer numbers than you currently look at, and better ones. A weekly operating review needs the handful of signals that show whether the machine is holding: orders shipped complete and on time, error and return rates by cause, the age of unresolved exceptions, and where cash currently sits in the cycle — inventory, receivables from marketplaces, refunds owed. The point is not dashboards; it is a cadence where those numbers are owned, trends are caught while they are cheap to correct, and decisions get made. Most founders drown in channel analytics while the four numbers that govern the operation go unwatched.

When the constraint is demand rather than operations. If orders are modest, fulfilment is comfortably managed and the real question is how to grow revenue, you need marketing and product answers I do not sell — and installing operating structure too early adds cost without adding speed. The signal that it is time is operational strain: growth that arrives faster than the operation can absorb, margin leaking through returns and penalties, peak seasons survived rather than run. If you are not there yet, I will say so and tell you what to watch for.

Keep exploring

What is a fractional COO?The plain-language guide.Fractional COO servicesThe embedded operating engagement.Fractional COO in IndiaGurgaon-based, India and global.Pricing & engagementHow engagements are priced.The diagnosticThe fixed-fee first step.Case studies & resultsMeasured outcomes by sector.

Other sectors

Media & newsroomsAdvertising & agenciesBPO, GBS & shared servicesScaling & founder-led companiesMulti-entity & enterprise operationsSaaS & software companiesPE & VC portfolio companiesGlobal Capability Centres (GCCs)