Which Picking Model Fits Your Reality? In-Store, Backroom, Dark Store, or Hybrid

Choosing your retail picking model is not a theoretical exercise anymore. For food and grocery especially, it defines whether you can promise fast delivery, protect margins, and scale an omnichannel network without chaos.

So, the real question is not which model is trending.
It’s: which picking model fits the reality of my stores, my demand, and my promise to customers? How do I keep that flexible?

Below, we break down in-store, backroom, dark store and hybrid models and show how a solution like Lyzer lets you run all of them without rebuilding your tech stack every time.

‍

Why your picking model is now a strategic decision

Online grocery and quick-commerce growth turned stores into critical fulfillment nodes, not just sales floors. Leading retailers that integrate stores into an omnichannel network already see higher profitability and better customer satisfaction than those who keep channels in silos. 

Your picking model directly impacts:

• Cost per order (labor, inefficiencies, rework)
• Speed & SLA (1–2h, same-day, next-day)
• On-shelf availability & substitutions
• In-store experience (congestion, service levels)
• Scalability across banners, regions and formats

Recent research on e-grocery and in-store analytics reinforces the same message: layout, process and data-driven task allocation can materially reduce picking time and variability. 

‍

In-store picking is fast to start, but fragile at scale

Do in-store picking still make sense for supermarkets and convenience stores?

When you use in-store picking, you are treating each physical store as a small fulfillment hub. 

That is powerful for one reason: you move fast without building anything new. You tap into an existing network, existing inventory and existing teams. 

For many retailers, this is the fastest way to validate demand, offer same-day delivery in multiple locations and launch click & collect without heavy capex.

The trade-off appears as volumes grow. Pickers share the same space with shoppers. At low order volume, that coexistence is manageable. 

But when demand accelerates, you start to feel it in very operational ways: congested aisles at peak hours, staff constantly “in the way”, longer picking times because routes are improvised, and more pressure on on-shelf availability. 

Without real-time visibility and structured picking paths, the store team react. 

So, in-store picking works best as:

• A launch and learn model to get online off the ground quickly.
• A fit-for-purpose model for proximity and neighborhood formats with moderate volume.
• A signal system: when congestion, complaints, and picking time cross certain thresholds, the data tells you it is time to introduce backroom capacity or move part of the volume to a different model.

‍

Backroom picking: protecting the store without over-committing

Backroom picking means you pull the online operation slightly “backstage”. Instead of competing for space in the aisles, your team prepares orders in a dedicated area inside or attached to the store, with its own shelving, processes and flow. 

The customer still sees a normal, quiet store; the intensity of e-commerce lives out of sight. This setup starts to make sense when online volume is no longer marginal, store traffic is heavy, and in-aisle picking is visibly disturbing the experience or slowing the operation.

The real advantage is control. With a backroom, you can define clear zones, standard routes and cut-off rules. You can reserve inventory for online orders, reduce substitutions, and run picking in batches without blocking customers or staff on the shop floor. 

It becomes easier to measure productivity, train teams and enforce a consistent method across locations, instead of relying on improvised in-aisle routines that change by shift or by person.

But it is not a magic fix. You need physical space, and you need discipline: a reliable process to replenish the backroom from the store or DC, rules for what is kept where, and orchestration so that picking, replenishment and walk-in customers do not fight for the same stock. 

When done well, backroom picking is a natural evolution from pure in-store: a way to absorb higher online demand, protect the store experience, and prepare the ground for more advanced models such as dark stores or hybrid networks, without overbuilding too early.

‍

Dark stores: when online is no longer a side project

A dark store is a retail node with no shopfront and no browsers: it exists purely to serve online demand. Layout, processes and inventory are all designed for one thing: fast, accurate fulfillment. 

You cluster the highest-velocity items, design short routes, standardize workflows and give pickers a controlled environment where every step is measured. For operators who have already proven strong online demand in specific areas, a dark store stops being an experiment and becomes a strategic asset.

This model fits best when demand is high and concentrated (typically dense urban areas), SLAs are aggressive (same-day, under two hours, or express) and online are treated as a core channel rather than an add-on. 

In that context, the gains are tangible: higher pick rates per hour, fewer substitutions, more reliable availability, and a setup that connects naturally with automation or micro-fulfillment technology. 

You are no longer trying to “hide” an intense e-commerce operation inside a live supermarket; you build the right stage for it.

The trade-offs are very real. Dark stores ask for capital, robust systems and thoughtful network design. If demand is overestimated or volatile, you end up with underused capacity and rising unit costs. 

They also add complexity: you need strong orchestration between DCs, stores and dark nodes to decide where each order should be served from at any moment. 

That is why dark stores are no longer reserved for the biggest players, but they are also not the right first move for everyone. They work when they are the next logical step of a data-backed strategy, not just a reaction to a trend.

‍

Hybrid networks: the reality for most retailers

A hybrid model is what happens when you stop asking “which single model is best?” and start designing your network location by location.

In practice, most mature retailers already operate a hybrid reality, even if they do not call it that. 

Some neighborhood stores prepare a few local orders directly from the aisles. High-traffic flagships move online volume into a structured backroom. Dense urban areas are served from dark stores or micro-fulfillment hubs. The same customer-facing promise runs across all of them: click & collect, same-day, next-day, express.

The power of a hybrid approach is that it lets you align the picking model with the role of each node: volume, footprint, catchment area, SLA, and economics. 

You do not overload a small proximity store with heavy e-grocery picking. You do not waste the potential of a dense urban cluster by treating it like a low-volume region. You shift volume over time: start with in-store, move recurring hotspots into backroom, and, once the data justifies it, transition part of the demand to a dark store. 

This continuous adjustment builds resilience. When one node is at capacity or disrupted, orders can be reassigned elsewhere without compromising the promise to the customer.

The catch: hybrid only works if it is orchestrated, not improvised. 

You need a unified control layer that sees every order, every stock pool, every SLA and every node’s capacity in real time, and can decide, order by order, where it should be picked and how. 

Without that, hybrid is just chaos with a nicer name. With it, it becomes a strategic advantage.

‍

How Lyzer enables any picking model (without a replatform every 18 months)

Do you need different systems for in-store, backroom and dark stores?
Short answer: No, if your orchestration layer is designed correctly.

Lyzer positions itself as that layer: a single platform from order to doorstep that lets retailers and e-commerce operators design, operate and scale multiple picking models in parallel. 

Here’s how that maps to your reality:

‍

1. Centralized order hub

• Aggregates orders from all channels (web, app, marketplaces, B2B).
• Applies rules to assign each order to the optimal node (nearest store, backroom, dark store, and external partner).
• Considers cut-offs, capacity, SLA, inventory constraints.

‍

2. Smart task allocation & workforce orchestration

• Breaks orders into picking tasks and routes them to the right picker at the right time.
• Balances loads between store staff, dedicated pickers and logistics partners.
• Helps avoid typical inefficiencies: idle capacity in one node, overload in another.

‍

3. Optimized picking on the ground

• Intelligent store mapping and smart cart logic to minimize walking distance and touchpoints.
• Supports multi-order picking safely, lowering time per order and error rates.
• Adapts to different layouts: classic supermarket, compact convenience store, structured dark store.

‍

4. Hybrid-ready by design

Because Lyzer is node-agnostic, you can:

• Start with in-store picking in your existing network.
• Transition selected stores to backroom without changing platform.
• Add a dark store in a key city and plug it into the same orchestration logic.
• Rebalance rules dynamically based on performance, demand and strategy.

No separate tech stack. No new playbook per model. Just configuration.

‍

5. Beyond software: Logistics as a Service

For operators who want speed without building everything themselves, Lyzer complements the platform with a vetted network of pickers and delivery partners, and an operational support to help you stabilize the model.

That combination is particularly relevant for banners testing new models or entering new geographies where agility matters.

‍

FAQ: direct answers for decision-makers 

Is in-store picking dead?

No. It is effective for low-to-medium volumes and as a fast go-to-market model. It fails only when unmanaged: no routing, no visibility, no prioritization.

‍

Do I need a dark store to offer 1–2h delivery?

Not always. You can use high-density stores as fulfillment nodes with the right orchestration and capacity planning. Dark stores become compelling once volume and SLA justify the dedicated setup.

‍

Can I run different models across my network?

Yes, and you should. The constraint is usually technology and governance, not the concept.

‍

When should I switch models?

Watch cost per order, pick times, NPS, in-store complaints, substitution rate and capacity utilization. When they start trending the wrong way, it is time to reconfigure nodes or upgrade the model, not just “hire more pickers”.

‍

How does Lyzer help in practical terms?

By centralizing orders, orchestrating tasks, optimizing picking routes and integrating last-mile, so you can evolve from in-store to hybrid to dark-store-supported networks without changing systems.

‍

What is the model you should follow?

There is no universal “best” picking model.

There is a model that matches your current demand, footprint and promise. And the technology layer that lets you do that without ripping and replacing every time.

If you are rethinking your picking strategy, start by mapping your reality. Then choose partners and platforms built to flex with it. Lyzer is built for exactly that.

‍

References: 

https://www.mckinsey.com/industries/retail/our-insights/reimagining-the-role-of-physical-stores-in-an-omnichannel-distribution-network

https://www.sngular.com/insights/169/insight-hybrid-retail-model

https://www.sciencedirect.com/science/article/pii/S0925527324003542 

https://frogmi.com/en/blog/technology-to-improve-product-availability-in-dark-stores/