Thought Leadership

Choosing the right mix of products for productivity

Sneha Radhakrishnan

June 26, 2026

 

I started my career on fixed systems. Conveyors, sorters, the unglamorous infrastructure that moves product through a building reliably, thousands of units an hour, year after year. Over time my work grew to cover the other side as well: robotics development, and then the integration work of getting both to behave as one system rather than two. So, when the industry started announcing that robotics was king and the conveyor and the sorter were finished, I had a front-row seat. I have since watched that same pendulum swing its way back.

That experience is the reason I want to push on something. If you say "it isn't robotics or fixed automation, it's the right mix, and the winner is whoever integrates them best," you will get a room full of nods. Integrators say it. Operations leaders say it. The serious analysts say it. As a principle, "hybrid is the future" is about as bold as "measure twice, cut once." It is closer to consensus than to a fresh idea.

So I am not going to pretend I am revealing anything. The people who run warehouses already know the duality is false. The story about robotics sweeping fixed automation into history does not really live on the floor. It lives one layer out, in the press, the funding decks, and the feeds, where a clean narrative about the new beating the old reads better than the messy truth.

The question I find worth asking is not whether to mix the two. Everyone agrees on that. It is the question almost nobody answers well. Where does the line actually fall, why, and why do so many operations that believe in hybrid still get it wrong? Agreement turns out to be cheap. Doing it well is rare. That gap is what I want to spend this on.

A correction the consensus usually skips: this is a phase, not a trend

Even people who accept the hybrid view tend to narrate the present moment as a one-way ratchet, with fixed automation in permanent decline and robotics rising forever. The growth numbers seem to back it up. The fastest growth right now sits at the flexible end of the spectrum: mobile automation, goods-to-person, ultra-high-density storage.

Look at why, though, and the permanent-decline reading falls apart. A large share of recent projects are brownfield, meaning companies are upgrading and expanding sites they already run rather than building new mega facilities from scratch. Brownfield is the natural home of flexible, modular technology. When you are working inside an existing footprint, with existing constraints and a tighter capital appetite, the ability to fit the space you have, work in confined areas, and scale up in small increments is worth more than peak throughput.

That is a construction cycle, not a verdict. High-throughput fixed automation did not get worse. The shape of demand changed. Shuttles still dominate high-intensity, high-volume operations because in a linear, high-velocity environment their speed and efficiency are very hard to beat. When greenfield construction picks back up, and it does in cycles, demand for high-performance fixed systems comes back with it, because new large facilities get built to maximize capacity. Even inside the categories supposedly being displaced, the nuance holds. Conveyor as a category keeps adding revenue, much of it from parcel, where the economics of moving standardized items along a fixed path are very difficult to beat with anything else.

So the real picture is not one line falling and another rising. It is a portfolio rebalancing in response to where, what, and how customers are building. The mix moves. The need for both does not. Anyone who tells you the current mix is the permanent mix is reading a phase as a destination. I have lived through enough of these phases over my career in different industries to be wary of that.

The reason it's hard: every customer is at a different point on the same map

There is no single right answer here, because no two operators are solving the same problem at the same moment. Plot any decision against two things, how much throughput it demands and how much flexibility it needs, and the whole market spreads out across the map:

  • A central hub or regional distribution centre moving massive, stable volumes sits in the high-throughput corner. This is where fixed automation belongs: high-rate sortation, dense shuttle storage, automated palletizing. Flexibility is secondary, because the flow is predictable enough that committing the concrete pays for itself.
  • A last-mile or micro-fulfillment site sits at the other end. Lower absolute throughput, but a premium on sort density, compact layouts, and the flexibility to handle a churning assortment in a small building. Mobile automation and compact flexible systems earn their place here, not because they are newer, but because they fit the constraint that matters.
  • A brownfield retrofit is defined by the building it already has. The real question is not the theoretical optimum, but what you can install around the columns, docks, and ceiling height you are stuck with while keeping the site running. Modular robotics often wins by default, and then a run of conveyor ties it back into the existing pick and pack line.
  • A greenfield facility can design from a blank slate and will lean on high-throughput fixed systems for the core flow, then layer robotics into the variable, hard-to-predict edges such as returns, value-added services, and the slow-moving long tail.

In none of these cases is it only robotics or only fixed. The greenfield giant uses robots at the edges. The brownfield retrofit uses conveyor to stitch its robots together. The hub automating its core still needs flexible capacity for the unpredictable. The duality falls apart the moment you stand inside a real building, which is exactly why a generic "we believe in hybrid" gets you nowhere. The belief is free. Knowing which corner this customer is in today is the work.

The part most people skip: an actual decision rule

Most versions of this argument stop at "it depends, integrate them" and call that the insight. It is not insight; it is a shrug. The operator does not need to be told the answer is "it depends." They need a way to make the call. Roughly, this is where I draw the line:

  • Commit to fixed automation where the flow is predictable, the volume is high, and it will stay that way long enough to pay back the capital. Fixed automation is a bet that the future resembles the present. When that bet is safe, with high-velocity SKUs, stable throughput, and a core flow nobody plans to re-architect next year, the cost per unit and the reliability are hard to argue with. The concrete is a feature, not a burden.
  • Reach for flexibility where uncertainty is the bigger cost. Volatile volumes, a shifting SKU mix, an awkward existing footprint, or a business that has to scale in steps rather than one heavy investment. Here you are paying a premium for each unit of throughput in exchange for optionality, the ability to redeploy, add a few units at a time, and adapt without tearing out the floor. That optionality has real value, and the common mistake is refusing to pay for it when your own demand really is uncertain.
  • Tier inside the building, not just across buildings. The most frequent execution failure I see is treating a facility as one uniform problem when it is not. Fast movers and slow movers want different storage. The core flow and the variable edges want different handling. A good design routes by velocity and predictability, with fixed and flexible systems each taking the slice they were built for.

You can see the tiering in the applications:

  • Storage and buffering: high-rate shuttles carry the fast, high-velocity SKUs, while flexible high-density and goods-to-person systems handle the long tail and the space-constrained sites.
  • Sortation: large central facilities can justify high-throughput fixed sorters because volume pays back the capital, while smaller and last-mile sites favour compact flexible sortation and mobile solutions built for sort density and layout flexibility rather than peak rate.
  • Palletizing and de-palletizing: robotic arms cover mixed-case and variable patterns while fixed systems deliver rate and consistency on standardized flows, and plenty of lines now run a robotic cell next to conventional palletizing, routed by product profile.
  • Picking: robotic systems are advancing fast and changing the labour equation, but they sit inside a flow that conveyor and sortation still carry. The robot picks. The fixed infrastructure transports and consolidates. Neither stands alone.

The two failures are symmetric, which is the uncomfortable part

I will say the thing that lands awkwardly in a room betting heavily on robotics. There are two ways to get this wrong, and they are the same mistake in different clothes.

The first is the legacy reflex, automating everything with conveyor because that is what we have always done, defaulting to fixed infrastructure for a flow that is too volatile or too uncertain to justify the commitment. Everyone in the industry is happy to criticize this one. It is the safe target.

The second is the hype reflex, and it is the mirror image. Throwing robots at every problem because that is what the market rewards and what the board wants to hear, deploying premium flexible solutions against a high-volume, stable, predictable flow that a fixed system would have run faster and cheaper for a decade. Far fewer people with a stake in robotics will say this out loud. A robot pointed at the wrong constraint is not innovation. It is the old mistake with better PR.

Both reflexes fail the same discipline, which is matching the tool to the constraint, building by building and flow by flow. This is why believing in hybrid is necessary but nowhere near sufficient. I have watched teams that sincerely believed in the mix still fall into one reflex or the other, because believing in balance and drawing the line in the right place are two different skills.

What integration really means

If the thesis is consensus, the advantage must sit somewhere else, and it does. It has moved away from owning the best individual machine toward knowing how to compose machines into a system, and increasingly toward building a system that can be recomposed as the business changes.

That is why the software conversation is now inseparable from the hardware one. A modern, modular, API-first control layer is what lets a high-throughput shuttle line, a mobile robot fleet, and a robotic pick cell behave as one coordinated operation instead of three isolated islands. The operators pulling ahead treat the warehouse as a hybrid architecture from the start. Fixed where throughput and predictability reward commitment, flexible where uncertainty and constraint reward adaptability, and an intelligent software spine holding the two together so the mix can shift as the cycle turns without ripping out the floor.

That last point matters most. If the mix is a phase rather than a destination, the asset that pays off over time is not any single machine. It is the recomposability of the system as the phase changes. The operation that wins is not the one with the most robots or the most conveyor. It is the one that can rebalance without starting over.

Where this leaves us

The story about robotics killing fixed automation survives because it is simple, exciting, and easy to put on a slide. Operations do not run on slides. They run on cost per unit, on throughput against real demand, on the floor space and ceiling height and capital you have, and on a long list of small, specific decisions about where each technology fits.

So no, the hybrid thesis is not a revelation. Most of the field already accepts it. The hard part, and the part I think is worth your attention, is that accepting it changes very little on its own. The win belongs to the people who can tell a phase from a trend, draw the fixed-versus-flexible line in the right place for this building, resist both reflexes with equal discipline, and build something that can be rebuilt when the cycle turns. Everyone agrees the two should coexist. Very few make them coexist well. After years of working on both sides of that line, I am convinced the difference between the two is where all the value lives.