The Operating Chief #1: How Jack Archer Cut $600K in 3PL Costs

We are delighted to announce “The Operating Chief,” a new CrossDock interview series featuring the people actually running the operation — not just commenting on it.

As part of this series, we will be interviewing the operators, logistics leads, and fulfillment architects who live inside the spreadsheets, on the warehouse floors, and in the carrier dashboards every single day. This series is built for them and for the people who want to understand how they think.

In our first edition, we have David Reifschneider, VP of Fulfillment at Jack Archer. With nearly two decades of scaling e-commerce and B2B fulfillment behind him, he recently launched Fulfilled By Jack — a fulfillment service built from the operator’s side of the table.

In this conversation, David talks about the real inflection points where fulfillment models break, what it actually took to bring operations in-house, and why the operation is never the building or the system, it's always the people, and more.

You have been inside DTC operations from $10M to $500M+. There is a specific inflection point where the fulfillment model that got a brand to $10M starts breaking. What does that break actually look like operationally? What starts failing first? And what in your experience are these inflection points where the fulfillment strategy needs to be evaluated?

The break is rarely dramatic. Three things fail in sequence: SKU proliferation outruns slotting logic and quietly drops UPH, channel mix shifts faster than the contract can differentiate DTC from wholesale economics, and peak season exposes everything that's been drifting all year.

Inflection points worth re-evaluating: ~$15M (you've earned a custom contract, not an over-charged simple rate card), $30–40M (in-house or dedicated capacity becomes credible), $75M+ (multi-node is real, not vanity), and any time channel mix shifts significantly.

At Jack Archer, you moved from outsourced to in-house fulfillment, cut costs by 38%, and delivered the project $1.7M under budget. Walk us through the decision. What were the specific signals that told you it was time to bring it inside, and what would have made you stay outsourced?

Four signals moved us off the fence: a constant cost-per-order curve we couldn't bend with the 3PL, storage fees they had no incentive to solve economically, a multi-channel strategy that couldn't be served by one generic contract, and the eventual FBJ as a service, which allows us to monetize owned capacity. Each on its own was insufficient; together they made the decision for us.

What would have kept us outsourced: lumpy volume that couldn't cover fixed labor in trough months, low SKU complexity, or a 95%+ DTC mix with no diversification. In-house is a fixed-cost bet on predictable volume and gives us full control of not only our own brand, but also the ability to partner with other brands and provide shipping and fulfillment services.

On coming in under budget: We scoped conservatively, ran a phased ramp, kept the 3PL as a runoff partner, and leaned on experience to prevent the surprises. Most in-house projects fail because teams try to switch the building, system, and labor model on the same weekend. Sequence it, phase it, and you’ll keep order…and prevent costly surprises.

You are building Fulfilled by Jack, a 3PL service that lets other external brands use Jack Archer's KC fulfillment infrastructure. A DTC brand becoming a 3PL is a real bet. What does unit economics look like? At what scale does it make sense, and what operational complexity did it add that you did not fully anticipate?

Unit economics work in three layers: direct services (pick/pack/store), where you compete on price, platform fees (onboarding, account management, SLA bonuses), and the margin of shipping rates, where most 3PLs live and die. Most brand-turned-3PLs underprice because they only see layer one. It’s also a problem if your shipping rates are already too high, how can you add margin to already bloated shipping costs, then try to pass those on to the smaller brands that need it most? It’s a no-win business strategy. We solve this via strong shipping partners and direct parcel rates and contracts. Diversification allows you to reduce parcel spend and pass those savings to your partners who need it. 

Scale: it works when you have genuine excess capacity, not when you're disguising a capacity problem as a 3PL business. If you're at 95% utilization with your own brand, you don't have a 3PL; you have a problem you'll regret in six months.

What I underestimated: contractual surface area (MSAs, carrier terms, IT segregation, customer-of-customer compliance) and management overhead, running multiple brands is roughly four times the meeting load of running one, because you've added a customer-success function on top of operations.

You renegotiated 3PL agreements at Jack Archer and secured $600K in annualized savings. What data did you bring to the table that the 3PL could not argue with? What was the actual leverage? Can you walk us through the various aspects in which the savings were delivered?

Four pieces of data, in order of impact: a unit-economics model built from their own billing data, normalized to cost-per-unit by channel, which usually exposes one channel subsidizing another. Live RFQs from three competitors with apples-to-apples scope, so they can't dismiss your benchmarks. An accessorial audit, where most of the actual savings live (accessorials are typically 18–25% of all-in cost and quietly mispriced for your order profile). AI has given us access to industry standards and baselines we simply didn’t have before. It’s like buying a car but knowing its numbers. AI is changing the game.

And fourth, a credible lever on negotiations. The 3PL needs to believe you'll leave. We had the in-house option costed and timeline-mapped, so the conversation wasn't 'give us a better rate,' it was 'give us a better rate or we're operating our own building in nine months.' The savings showed up the week after. We ultimately left anyway, but that runway gave us the savings needed to launch our own site. 

What are the three real symptoms you have personally seen that told you a brand had outgrown its 3PL? The actual things that start going wrong in the operation.

One: your ops team spends more time managing the 3PL than improving the operation. The relationship has stopped being a service and has become a project.

Two: Every operational change requires a six-week change order. The 3PL is moving at the speed of their largest customer's process, not yours. Small brands need to move fast to compete with the larger ones. Find those leverages and exploit them. 

Three: You can no longer identify where SLA failures originate. The operation has become a black box, and you've lost the ability to manage the customer experience. When you have no visibility into what is happening, you can’t control the customer experience. 

When a brand decides to leave its 3PL, what does the transition actually look like in practice? How long does it take, and what goes wrong that nobody warns you about?

The realistic timeline is six to nine months from board approval to full cutover. Compressing below six introduces risk; stretching past nine usually means weak project management and poor execution.

What nobody warns you about: Inventory variance on physical count will be worse than your system-of-record suggests. I’ve transferred inventory both in-house and out of a 3PL; the reconciliation always has a variance.

That's accumulated drift nobody pressed on while you weren't leaving. Carrier rates re-tender at the new facility, often 3–8% higher for the first six months until you rebuild volume credibility (model this, brands routinely forget to subtract the carrier delta from declared savings). And CX tickets spike during transition, even when execution goes well, because customers notice the change. Get in front of it and be proactive with customer messaging (even on the check-out or post-checkout page)

At Jack Archer, you are fulfilling for your own DTC site, marketplaces, and retail partners from the same infrastructure. When orders come from everywhere, what is the hardest trade-off to manage day to day? Where do the channels fight each other?

Channels fight over inventory, labor, and ship-by deadlines every single day. Inventory is the hardest because the right answer is genuinely contested. DTC velocity, a wholesale PO due Friday, a marketplace promo, a paid push next week, each has a legitimate claim to the same units, and the warehouse can only execute what allocation tells it.

The practical fix is a daily ten-minute allocation meeting that owns the inventory call and is empowered to override channel demands. Without it, channels resolve conflict through escalation, and the operation pays in chaos. Stay tight to your wholesale team and ensure demand and supply planners are aligned with the business strategies. You can’t foresee all issues, but staying connected can prevent most of them.

What kind of technology, software infrastructure, and readiness should one have to run an omnichannel fulfillment and distribution at this scale? What are the important considerations that are involved in choosing the right technology partner?

The stack, in priority order: An OMS that owns order source-of-truth and is genuinely channel-agnostic; a WMS that handles wave planning, slotting, and labor; an inventory engine with configurable available-to-promise rules; and a transportation layer for carrier selection. Most brands force one system to do the job of three, and that's where the operation could break. 

When choosing a partner, what actually matters beyond features: integration maturity with your existing stack, implementation methodology (talk to three customers who went live in the last 12 months), SOC 2 status (and roadmap) if you have an enterprise customer in your future, customization cost, and a clear integration scope.

And don't choose based on demo quality. Demos can be theater; you’re seeing what they want you to see. Reference calls tell you more in an hour than a six-week RFP. I’ve always created strong relationships with the leadership teams at WMS providers; that’s key and ensures proper scope, timelines, and delivery. 

You ran a sub-2-day national ground shipping footprint from a single node in KC. Walk us through until what scale a single-node works, and at what point, what revenue level, what geographic demand pattern, it stops working.

Kansas City is structurally the best single-node location in the US for ground, as you reach roughly 85% of the population in two ground days. A single node works up to roughly $75–100M in DTC parcel volume, with caveats: AOV needs to support ground shipping economics, customer concentration shouldn't skew heavily to one coast, and your delivery promise needs to be honest about what KC can deliver.

It stops working when customer expectations move past your delivery promise faster than the math supports. The signal is conversion softness on coastal customers, and 'where's my order' tickets are concentrated by zone. Most brands at $30–60M that add a second node are solving a marketing problem with an operations solution. Stay single-node as long as the data supports it; once you move to bi-nodal or more, complexity increases significantly. 

At Stitch Fix, you took over the lowest-performing fulfillment center in the network. Within 12 months, it was the top-ranked facility. What did you do in the first 30 days? And what was the hardest decision during that turnaround?

The first 30 days were diagnosis, not action. I was told prior to hire, this particular facility had become misaligned with the company culture. Not only was this accurate, but I discovered the leadership team and many of the hourly employees needed turnover.

Having the right leadership team, aligned to facility and company goals was key. Once that was complete, the real work began of diagnosing where inefficiencies lived. We attacked them one by one, improved layouts, reduced wasteful movement, and optimized our pick path with density. We took a fresh look from door to door. Everything and everyone was on the table. Once we turned the corner, the team could feel it, the energy was positive, buy-in happens organically, and performance increases. 

The hardest decision was personnel. A long-tenured manager who was knowledgeable, well-liked, and actively undermining the changes, not maliciously, but because they implicated his own past decisions.

The lesson: When you've identified the issue, move faster than your instinct tells you to. The team is watching to see whether you'll do the hard thing; they know when things are broken, and it’s your job to find it, fix it, and restore a safe, high-quality, and efficient operation. 

You have managed a startup crew, and you have managed large teams in an operation such as Blue Apron. What works with 50 people that completely breaks at 500? How does the job actually change?

Direct communication breaks. At 50, the leader knows everyone, senses the floor mood by walking it, and resolves conflict informally. At 500, you can't know everyone, your strategy gets game-of-telephoned through three layers, and conflict needs structure, not relationship.

The job changes from doing the work to designing the system that does the work. The failure mode is operators who get promoted into a 500-person job and try to keep doing the 50-person job, staying in every operational decision, assuming information flows because they sent it, and not noticing when culture shifts from 'we' to 'they.''

Having a strong leadership team is critical. You’re doomed without a strong, aligned, and effective leadership unit. It doesn’t mean you’re not still visible or trying to get to know your entire team…it means the message and work you’re doing is also being done by every other leader on your team, and you’re aligned and informed up and down the chain. 

You mention deploying agentic AI to replace coordination overhead in your operations. Give us one concrete example. What was the manual process, what does the AI workflow look like now, and what did it actually change in headcount, speed, or accuracy?

Weekly KPI reporting. Manual process: An analyst pulled data from WMS, OMS, the carrier portal, and finance every Monday, built thirty KPIs in a master sheet, produced a deck, and spent two days answering follow-up questions. Roughly twelve analyst hours a week, plus six hours of leadership reading a deck that arrived two days after the data was actionable.

Agentic workflow: Data pulls run overnight, calculations run on a validated backend, and an agent generates the deck on Monday at 7 am with a written narrative that flags any KPI moving past threshold and surfaces a probable cause. The analyst now spends two hours reviewing edge cases and one hour on follow-ups.

What actually changed: Nine analyst hours recovered, three days of latency removed, and the leadership conversation shifted from 'what happened' to 'what should we do.' That's a different conversation, and it produces different decisions. Accuracy improved, but the failure mode changed; agents can be confidently wrong in ways humans aren't, so the human review step is non-negotiable.

You drove $400K in incremental shipping savings at Jack Archer through carrier mix optimization. For a DTC brand doing ~$30M, how do you decide the right split between UPS, FedEx, USPS, and regionals? What is the framework?

Build the framework backward from the customer experience, not forward from the rate card. Start with: What's our delivery promise, what zones are the customers in, what AOV supports what service level, and what failure mode does each carrier expose us to.

For a typical $30M apparel brand: 40–50% UPS or FedEx Ground (always have a primary and a secondary, never one), 25–30% USPS for lightweight and remote zones, 15–25% regionals where they have density advantages, and 5–10% premium services. Regionals are the lever brands underuse; they can deliver one-day faster transit at 10–20% lower cost, but the integration friction scares people off. Solve the friction; the savings are real. Again, leverage AI where you can to save labor and increase the integration timeline.

Most DTC brands accept their rates and run one annual negotiation. What are they getting wrong? What data should they be bringing to the table that they typically do not?

They're treating it as a procurement event when it's continuous performance management. Carriers adjust networks, fuel methodology, accessorials, and dim factors multiple times a year; annual negotiators are responding a year late.

Data brands typically don't bring: Dim weight analysis by SKU (most are paying for air after carriers raised dim factors), accessorial spend as a percentage of base spend (often 22% and renegotiable), zone-level audit of actual SLA performance vs. carrier commitment, and a credible alternative, even a low-volume regional relationship tells the national carrier you can move volume if you need to.

Be proactive and assertive. You can bet the parcel companies are! This is one of the most effective areas to use AI- build your tools and analyze your spend weekly. Don’t wait for a big annual RFP to address cost inefficiencies. 

At Blue Apron, you scaled a facility from 300 to 1,000+ employees while the business was growing faster than anyone planned for. What was the closest the operation came to breaking, and what saved it?

A multiple-week stretch of extreme growth, where hiring exceeded our ability to onboard and train. Trainers themselves had been on the floor for under 60 days. Quality was visibly degrading in customer complaint data within 48 hours of new cohorts going live, and we were a couple of decisions away from a public food safety problem that would have been existential. We needed more and more space, more ingredients, more labor…it was chaotic.

What saved it was a leadership team willing to slow growth to protect the operation. We took a throttle to orders, lower than what marketing could deliver. That requires a leader who'll choose operational integrity over a quarterly number; not every leadership team makes that call, and the ones who don't are the ones who end up with public failure. It takes fortitude to slow orders, which goes against all we’ve ever learned in business, to allow the team to build the structure and processes to accommodate the explosive growth.

Without it, every day is chaos. Managing upward is key; if you can’t convince the C-team what needs to happen, and how it will happen, you’ll never get in front of it. Don’t get me wrong, you’re just buying time – eventually the CMO will resume the order pipeline, so you'd better have a strong, quick plan and execute flawlessly. 

When you walk into a fulfillment operation that is failing, what do you assess in the first 72 hours? What is the first thing you fix?

Hours 1–12: Get on the floor, every shift, watching. You can tell within an hour whether the problem is process, people, equipment, or system. Hours 12–48: Dig into the underlying transactional data, not the management dashboards. Hours 48–72: Have structured conversations with leads, supervisors, and trusted floor associates.

The first thing I fix is almost always the daily management cadence. Failing operations universally have broken stand-ups, either none, or status reports instead of problem-solving, or the wrong people in the room. Fixing the cadence costs nothing, takes a week to install, and creates the operating system through which every other fix happens. Trying to fix anything else first is fixing the engine on a car with no tires.

More than two decades in this. From Amazon in 2000 to Jack Archer today. What is the one piece of operational wisdom you wish someone had told you at the start that you had to learn the hard way?

The operation is not the building, the system, or the process. The operation is the people, and everything else is in service to them. I spent the first ten years optimizing systems and assuming people would adapt; they did, but slowly and unevenly, and the gap between what the system was capable of and what the operation actually delivered was always a people gap.

The corollary, which took longer to learn: You don't fix a people problem with a system, but you don't fix a system problem with people either. Knowing which you're looking at is most of the job.

Matching up to Amazon's and Walmart's delivery promise is no longer a nice-to-have for a brand, but a survival risk. What do the Brands have to constantly do to match up to that promise? Where do smaller and leaner brands have an advantage over those large retailers, and conversely, where are they handicapped?

Three things, continuously: Monitor the gap between delivery promise and delivery performance by zone, weekly, and watch the variance, not the average (the customer who had a bad experience doesn't care your average is good); Communicate transparently, because Amazon's real edge is partly logistics and at least equally expectations management; and ruthlessly examine AOV-to-shipping ratios, because free shipping is a unit economics decision, not a marketing one.

Where smaller brands win: Unboxing experience and brand affinity (an Amazon box looks like every other Amazon box), operational speed (a small brand changes pack-out in a week, a retailer takes six months), and being excellent in a niche where retailers have to be adequate across everything.

Where they're handicapped: cost structure, network density, capital depth, and technology. Amazon can lose money on shipping for a decade as a moat; a small brand cannot. The strategic answer isn't to out-Amazon Amazon. It's to deliver an experience that's differently good, fast enough, transparent enough, and connected to a brand identity Amazon can't replicate. Customers understand that, and you get a ‘non-Amazon’ pass in most cases, but that only goes so far. Be confident in who you are, what you can achieve/promise, and protect your margins. 

If you enjoyed this conversation, please recommend us to your colleagues and friends. More exclusive content and interviews from across the supply chain world are on the way.