The Four Stages of Ecommerce Operations

Almost all companies just getting started don't have the necessary capital expenditure (capex) to build out the physical assets necessary to power their operations. It's best to go from your garage to a 3PL who can take your inventory, fulfill orders, and ship to customers.

3PLs have been a vendor category since forever. There are thousands of options. Some are historically good, while new entrants do a good job of being tech-forward. Pick one that best fits your inventory mix along with customer and shipping profiles.

Be conscious of how 3PLs price their services. They are delivering value for you, but how they charge for that value can greatly differ. One 3PL might attempt to make a profit by upselling storage of your inventory then hook you on cheap fulfillment prices or shipping labels. Meanwhile, others might play the shell game and upsell shipping labels while making storage free. Consider how they charge and if it fits your business.

At small volumes it does not make sense to go direct to carriers to secure shipping contracts. It is much better to leverage someone else's contracts.

There are lots of options, from consultants to API aggregators to the store platforms like Shopify. Generally the best option is your 3PL because if you picked one that is optimized for your type of business, meaning SKU count and dimweights, then the 3PL's contracts with carriers shared amongst tenants will be best optimized for your shipping profile, too. 3PLs also pose a benefit in that they should have multiple carrier contracts available, often including regionals, too. This means you have the potential to shop for better rates per order, as well as have added resiliency. It's hard to achieve those things when using only one carrier.

Early on in your growth trajectory, returns will become a problem. Return rates are approaching 30%-40% of orders on average, depending on who you talk to. Both the cost of return shipping as well as the cost of reverse logistics processing can quickly add up.

It's best to partner with another company to manage returns as early as possible. Your 3PL can likely provide that service, too.

Young companies rightfully prioritize growth over everything. Hard to meaningfully optimize costs and performance if there isn't a lot of volume, isn't it?

Marketing has a key role in this primary objective, which is what every blog on the internet talks about. Brand building, social marketing, ad buying, and so on.

That's not what we are good at or what this guide is about. We focus on operations, which can play a role with major growth initiatives, too.

A new brand/product company probably won't expand the portfolio yet. But it does make sense to diversify channels. This is a contentious topic because some voices claim the importance of focus and owning your channel beats everything else. While that can sometimes be true, we have more consistently seen that young brands take too long to branch out to other channels. This means selling on Amazon/Walmart/Target marketplaces, securing omni channel distribution via big box stores, and so on.

But that's hard! Not only is there the margin hit, but ops now needs to harmonize fulfillment across multiple channels, deal with conflicting demand and availability, and many more constraints.

Because you are still using a 3PL at this stage, chances are your partner can help shoulder a lot of the burden. Consider leaning on them.

Customers are more likely to buy if they know and trust when they will get what they order. Studies show cart conversion goes up about 5% with a trustworthy promise, and can go as high as 12% if the promised date is competitive.

What do we mean by competitive?

Consumer preferences have changed. Over 80% now expect a free shipping option, while over 50% expect a 2-day or better shipping option. What's scary is almost 40% expect both!

Being upfront on product description pages or cart checkout flows about the fastest date a customer can get an order for free is how you start to be competitive and win consumer dollars that they might spend elsewhere. Increasingly decreasing the cost to meet that date until it's free to the customer is when you really start cooking.

But delivery promises are really hard. Did you know they are mostly just a guess? Even Amazon's down-to-the-minute delivery promise is based on a stochastic model with a threshold confidence variable.

So for you to do a competitive delivery promise for fast and free shipping, you need a strong data science foundation to project what is possible with your business operations.

Free shipping is without question one of, if not the, most important growth drivers that ops is responsible for providing. Ubiquitous industry studies point to over 80% of customers look for free shipping or will shop elsewhere.

During earlier stages of the business, if a company were to offer free shipping, providing it would be some sort of compromise: Either the company would only offer it on very long and cheap shipping methods, or bake the shipping price into the ASP without the customer knowing, or simply eat the costs.

Fortunately, approaching this size with these volumes, it becomes possible to design a free shipping program that is competitive and scalable. A lot hinges on the ability to play arbitrage across various carriers, methods, and locations, which means the program becomes more cost-effective and competitive as operations grows. While it's possible (and recommended) to get started with only one FC, it's generally best to secure two carriers and multiple methods before implementing the technology used.

Similarly, providing a fast shipping option is critical to ensure a market majority is addressed. Without fast shipping, a certain percentage of customers will shop elsewhere.

And, similar to free shipping, up until this point fast shipping was usually passed through to the customer. Want that jacket tomorrow? Pay $25 for FedEx Next Day Air.

Now that you have two carriers and multiple methods in place, it is increasingly probable that you can find fast shipping via a cheaper method that will probabilistically deliver on the date promised. For example, maybe the FC and the customer are located in the same state, in which case you could offer next-day shipping for free, then simply pick the cheaper UPS 5-Day Ground method during fulfillment knowing it will, in fact, get there tomorrow.

Chances are your company has been running promotional programs for a while. From the very beginning there were probably "Two-for-One" or bundling schemes. Cart threshold promotions based on an aggregate analysis—"Free shipping on orders over $X" where $X is a general price point that's good for your business—can happen much earlier, too. All are great early ideas to drive early growth.

At some point, greater growth is found through per-user frontend customizations.

For shipping, that means taking the same $X free shipping threshold, but making it dynamic to the individual user based on an understanding of where the user is. For Customer A, it might be $42 but Customer B might be $77. Granular customizations like that will drive greater conversion, but to do so, operations must have tighter backend integration with the frontend store.

Other ideas here fall under recommendations instead of promotions. Nothing too novel here, with some of the ideas likely already implemented by the marketing team and powered by your frontend store—"Customer who bought X also bought Y, Z" recommendation blocks. But think of the marketing team as exclusively a growth driver, which is a problem because profit and margins matter as the business gets into the $50M-$100M per year sales range, as greater savings will finance better cash flow and stronger promotional programs. If marketing is focused on the top line, who is focused on the bottom line?

When thinking this way, the area operations can help is understanding which recommendations provide superior margins over other recommendations. For example, maybe a customer is looking at a shirt, and there are 10 products other customers bought with that shirt. It turns out that 3 of those 10 products have dimweights that, when bundled with the shirt, come in at under 2 pounds and fit into a box size that uses a cheap shipping rate. If a customer buys one of those 3 SKUs with the shirt instead of the other 7, the cost of shipping will be much better. Tight integration between backend and frontend systems for a recommendation engine that also provides optimized operating costs is a major secret behind some of the biggest sites in the world. Saving 10% on margin over millions (and in Amazon's case, billions) of orders starts to add up!

This is the most controversial but the most important step in the framework. When does it make sense for you to bring fulfillment in house? The answer varies, but the general summary goes like this:

Up until this point, 3PLs have been an excellent option. But at about this stage, most companies start to see the volume-based pricing of 3PL business models negatively impact their own balance sheet. Important metrics like cost-per-unit start to become untenable and companies at this size start to get a bonafide COO with board mandates to improve cost structures.

Why is that?

The main reason is financial. 3PLs are financially modeled as operating expenses (opex). The more you use a 3PL (increased throughput), the more you pay (increased costs), because pricing scales linearly with your success. There is rarely a situation where the more you use a 3PL, the more you save. While 3PLs provide enormous initial ROI compared to the cost of building your own fulfillment, they provide poor ROI at scale.

So with enough volume, it becomes clear there are certain optimizations specific to your business that would have a meaningful impact on improving margin or customer experience that simply aren't being realized because the 3PL is optimizing for their economies of scale, not yours.

Therefore it is an inevitability that all companies eventually take fulfillment in house. The question becomes when.

To which the answer is... it depends. For a real-world example, a well known personal care brand sells around $1 billion of product online, but has a simple business profile of just a few SKUs all under 1 pound in weight and small in size. This company still uses a 3PL because the unit economics make sense.

Meanwhile, we have spoken to multiple apparel startups out of Los Angeles with similar complex SKU mixes, a concentrated customer base, and focused DTC strategy. For these companies, it made a lot of sense to stand up their own warehouse in Southern California as soon as possible, which meant before they were even to $20 million in annual sales. Why is that? Because with a very large SKU count (tens of thousands of unique items), storage is very complex and expensive, but with low order volumes, fulfillment is very easy. It's hard to find 3PLs who will cheaply charge for such large storage requirements when they can't make up the price through order volume or inventory turnover.

As this simple comparison shows, the properties of your business determine the inevitable "when" with the key properties being ASP, AOV, and CPU. But, in general, we have seen what we call the "Fifty / One Hundred Rule" — A company either gets to $50 million in annual sales or 100,000 in monthly shipments. At that point, order volume is high enough to invest capex in improving opex. The most meaningful vehicle to do so is migrating from a 3PL to an owned FC.

It's still possible to leverage a third party carrier contract even while using your own FC. Aggregators, consultants, and resellers provide various benefits.

In addition to setting up your FC, or at minimum shortly thereafter, you want to start establishing your own contracts with a national carrier. Getting the first one done is the hardest part.

The reason is you will always achieve the best cost structures for contracts optimized to your SKU and shipping profiles. It is virtually impossible to do so otherwise.

A general rule of thumb: There are 4 "nationals"—USPS, UPS, FedEx, and DHL. USPS is best at very small parcel sizes; UPS is best at ground and commercial shipping; FedEx is best at fast methods and SLAs; and DHL is best at combining international shipping options. While all are a valid first carrier, pick the one that best fits your business at this stage.

Now that you control your own fulfillment processes and have direct ownership of shipping, you should combine the prior investments in a frontend delivery promise with the backend of shipping decisions.

In short, this is where it makes sense to tie shipping decisions with a promise given. "Promise what you can deliver, then deliver on what you promised." The latter is now possible.

That usually includes integrating pieces across your OMS, WMS, IMS, and TMS. (We jokingly call it the *MS group.)

When completed, you can now pass the customer promise at check out to the respective downstream systems to ensure an order is delivered on that date.

Either in conjunction with, or shortly after, you should secure a second contract with a national carrier. The biggest myth in ecommerce operations is the idea that the best shipping rates are found with a single carrier. In fact, carrier sales reps will go so far as to sell you on that specific point, and "customize" the contract to make you feel like the rate is as optimized for you as possible. Then they throw in volume breaks so that if you hit a certain limit, you get even bigger discounts.

Meanwhile, the reason the carrier sales rep needs your estimated shipping volumes is because they put the volume breaks near that threshold. Their job is to get as close to 100% of your volume as possible.

We have talked to countless companies who insist their rate with a specific carrier is the best possible in the world so it makes zero sense to add a second. They couldn't be more wrong.

In truth, carrier contracts are set up so that a customer "saves" money with some combination of properties, but the carrier "makes" the money back with other properties. For example, maybe it's a great rate for packages under 1 pound but a terrible rate over.

The condition people tend to under appreciate is the variability of orders and shipping, combined with the weekly volatility of rates and surcharges, combined with the hidden constraints of a carrier's network. It's almost always the case that engaging with two or more carriers using multiple methods allows for the business to arbitrage rates, SLAs, and reliability against each other to gain greater cost savings, faster speeds, more reliable delivery, and happier customers.

We typically see a business with two carriers pick USPS as one of the carriers and UPS, FedEx, or DHL as the other. This is because USPS is uniquely specialized at smaller parcel sizes, so many companies opt to prioritize USPS for small domestic shipments then fall back to another carrier for larger or international shipments.

With the prior programs in place, your outbound shipping process now has considerably more stress than it did before. You need to meet fast delivery dates usually priced as free.

Whenever operations is up against these types of constraints, we see an attempt to simplify process at the expense of costs. As a result, costs balloon and margins are destroyed. A simple example: A customer is expecting a package delivered in three days for free, so the shipping method picked goes by carrier SLA and you choose the expensive FedEx 2-Day Air option. That's an easy policy to scale.

It's at this point you need to prioritize cost-based decision making. We like to say "Constrain on speed, optimize for cost." It is a bad idea to progress to other growth initiatives, like opening a new FC, without prioritizing the latter half of that statement and get costs under control.

The best way to solve it is with data science. There are a lot of situations where costs can be cut, but it required sophisticated software and data modeling.

One FC poses a big scale problem: No matter the location, it is expensive to ship to nearly half the country. Certainly there are mitigation tactics, like putting the first FC close to the biggest concentration of customers. If 75% of purchases come from the Northeast, then it makes sense to put the FC in Pennsylvania, for example. Even so, long haul deliveries for parcel delivery to the opposite side of the country is a real margin killer.

That's why we advise companies to get to a second FC as soon as possible. We strongly believe that companies wait too long to open up a second FC. This can be a challenge, however. In fact, we joke that it's often harder to go from one-to-two FCs than it is to go from zero-to-one!

The operational pain and capital expenditure is more than worth it. The savings of eliminating Zone 4-8 shipments has an enormous impact on costs, while opening up the possibility of fast or free shipping to 60%-80% of the population.

If you have been outsourcing returns, you might have kept the outsourced partner handling this aspect of the business while you open up your first FCs.

We recommend considering treating returns management separately from order fulfillment, and thus bringing returns in house only after much of the underlying infrastructure is in place to support it.

Reverse logistics is its own category of problems to solve. The key considerations, however, fall along the same themes throughout this guide: Focus on maturing the integration of frontend and backend systems to optimize operating expenses and improve customer experience.

For example, tie together website account management with the returns process so customers can file a return ticket, print labels, and track return statuses. Dynamically pick the cheapest shipping option for the return shipment the same way you decide a method for outbound delivery. Or make the delivery address the closest FC in order to reduce shipping costs. Ideas like that.

For years, regional carriers were relegated to afterthought by ecommerce companies. Efficiency and scale provided by the four national carriers meant that most ecommerce companies wanted to secure a single contract with a national, wipe their hands, and move on.

At a certain size, adding regional carriers becomes one of the biggest driver of improved costs and customer experience. Fortunately, many more companies are realizing this and regionals are rightly no longer being considered an afterthought.

Costs are a main reason. Regional carriers offer tighter local networks that can lead to better rates for the areas they cover. Speed of delivery often competes or beats national carriers, too.

But an increasingly number of companies are adding regional carriers for resiliency above all else. The industry is shifting away from the single-point-of-failure afforded by consolidating on a single national carrier. The majority of companies we spoke to experienced major issues with USPS during Q4 peak. Who cares if you have the (perceived) best rate in the world with USPS if it takes three weeks to deliver a package and miss Christmas? Customers loyalty diminishes, stalling growth.

Once you get to two national carriers, it is usually smarter to begin including regional carriers before adding another national carrier. We recommend getting to three or four, effectively one for each corner of the country.

The challenge is onboarding and managing them. Complexity explodes as you now need to add these carriers to your internal policies and technology. Weekly surcharges have to managed. API integrations maintained. As a result, operators need to secure the direct business contracts with a regional carrier, but also ensure their technology stack can accommodate the scale.

Expanding geographical markets is an inevitable inflection point. While the marketing half of the business might have expanded at earlier stages, we argue it is usually around this stage of the business where it makes sense for operations to optimize that expansion. For most companies, leveraging the international capabilities for the four main national carriers is likely sufficient for quite awhile, but there will come a time where adding international carriers for concentrated markets becomes critical to performance and costs.

Let's reset: You now have a complex network consisting of many FCs, carriers, methods, and packaging types. At this stage, complexity is further increased with an expanding SKU portfolio and diversified channel distribution. Maybe you have physical stores now, too.

It is at this point where the order-level decision making needs to be improved beyond static rulesets and policies that has likely been unchanged in your OMS for years. In the past, the OMS might have handled orders with a basic rule like "If shipping to this state, use that carrier" applied to all orders. It's time to evolve that.

The answer is dynamic order routing across the complex network on an order-by-order basis. "Shipment Generation" is another term.

The key is to move decisions down to the order level as opposed to the system level. For example, for a given order, don't use a static rule from the OMS that applies to all orders, but instead look at real-time properties of your network to make decisions: What are inventory levels at each FC? Where is the buyer located? What are the dimweights of all SKUs? What does future inventory probably look like? Does it make sense to send as split shipments or as one? From which FC? And so on.

Shipment generation is a hard problem to solve. There is no existing commercial system that does it properly for ecommerce. Technology teams constantly have to build customized software on top of their OMS, WMS, TMS, and IMS, but even then many of the decisions aren't modeled properly.

Perhaps no greater ecommerce trend was accelerated during the COVID-19 pandemic than the urgent need to harmonize digital experiences with physical stores. A small subset of online-only stores, like chewy.com, thrived the past few years because of the explosion of customers forced to shop online. But the more consistent trend was consumers sticking with familiar brick-and-mortar brands or stores as dollars flooded to online purchases and those incumbent companies trying to figure out how to merge their digital storefronts with existing physical assets.

BOPIS, curbside pickup, and ship from store all became a prioritized focus point since 2019. We noticed companies didn't focus on this until roughly this size, mostly because smaller companies don't have the available capital to both buildout ecommerce-centric fulfillment centers and physical stores. With that said, we certainly spoke with several DTC companies with only one FC and less than $100M in annual online sales who had already invested in opening physical stores pre-pandemic. Therefore the range is large for when this is a focus, and we pick this stage only because it was the average amongst a large sample.

Regardless, it's a hard problem to solve. The most important component is the integration between the frontend store with backend inventory. Fortunately inventory management systems have gotten much better at providing real-time inventory visibility at warehouses. Unfortunately, store-based fulfillment, whether via inbound pickup or outbound shipping, depends on the same visibility on inventory within stores. This is a notoriously hard problem and continues to be the key bottleneck to providing this option to customers.

Adding a third FC is important because the ability to offer 2-day shipping (or profitable free shipping) goes from roughly 60%-80% of the population to 80%-90%. Adding that third FC to the middle part of the country is the last step before marginal returns kicks in with regards to population coverage. A fourth, fifth, etc., FC has diminishing returns where the benefit is only found with true scale (think millions of shipments per month).

The good news is going from two to three FCs is easier than prior expansion. Many of the "in the building" processes and technologies that needed to be pluralized across two FCs can be applied to the third.

The difficulty comes with adding a third node to the network. Updating decisions like "Which FC should we ship from?", "Do we do a split shipment?", and "Where should we allocate this new inventory?" become exponentially harder.

On average, ecommerce companies with multiple fulfillment nodes will see between 2%-5% of their orders be split shipments. Sometimes splits are good for customer experience and margins. Usually they are bad.

At a smaller scale, it's not a big deal. Furthermore, with only one FC, you will never have splits! And even with two FCs, the need for splits will be rare. But the jump to three FCs and beyond means the decisions around splits becomes too complex to manage. Coupled with the volume of orders at this stage, and the costs could be too much to bear as well.

It's at this point where a concentrated effort to minimize and manage split shipments becomes a true needle mover.

The goal of the program is simple: Reduce the occurrence of split shipments by a few percentage points. The way to get there is extremely complex, however. The solution typically includes stochastic modeling of inventory placement to optimize first allocation efforts, future inventory level forecasting (done in real-time at the moment of order fulfillment!), and an analysis of shipping costs for the given order in question.

Hot on the heals of shipment generation and split management is optimizing inventory placement. When acquiring inventory, the initial placement (or "first allocation") of inventory can have the biggest impact on outbound shipping costs compared to any other decision in the supply chain. Don't put all the winter coats in the Miami FC, for example.

At earlier stages in the business, placement is straight forward. Very young companies rely on the 3PL to make those decisions. Middle stage companies with one or two FCs either ignore the decision or make it simply by mirroring inventory across both FCs.

As complexity grows through the continual expansion of fulfillment nodes, carriers and methods, and SKU counts, "first allocation" becomes a critical decision.

The solution is yet again rooted in data science. A sophisticated data model will inform which SKUs should be sent to which nodes based on probabilistic forecasting of the location of which customers buying what products.

Initial inventory placement ("first allocation") is really important. But it will never be perfect because of the dynamic nature and volatility of ecommerce. At this size, regular inventory rebalancing will be required to ensure that inventory levels are as optimized as they can be for future orders.

There are a few ways to do it, but the primary concept is to blend probabilistic purchase forecasting with current inventory levels. If there is a high probability that a certain SKU will go out of stock in the East Coast FC but many customers will likely purchase it, then it might make sense to send some of that SKU as a bulk shipment between FCs. This will prevent long-haul and split shipments in the future.

While it's always good to have optimized inventory levels, the volume likely isn't high enough until around this stage of the business to make it an economically sound project to prioritize.

The final stage of operations evolution, and where this guide comes to an end, is continual expansion of your physical footprint. With enough scale, adding the fourth or more FC will prompt meaningful cost savings and improved customer experience on the margins, which will become the key areas remaining for continual margin improvement or revenue growth.

With each new node added into your network, most of the prior projects must be recalibrated. The way in which you route orders, manage splits, promote delivery promises, place inventory, and so on will change with each new FC.

Therefore one way to think about adding new physical nodes is to separate projects between "in the building" efforts and "out of the building" efforts. Yes, things like labor and robots and pick-and-pack lines are critical to the success of the FC. But ensuring technology that manages your network can easily assimilate a new physical node is just as critical.