Unified Retail Automation: Tying In-Store and Online Together

Most retail technology treats in-store and online as separate worlds. Separate POS for the storefront, separate e-commerce backend, separate inventory systems, separate customer databases, separate loyalty programs that don't talk. Every gap between them is somewhere a customer falls through.

This is the operational picture for most Canadian retailers I work with: a Lightspeed POS in-store, a Shopify store online, a fragmented view of customers, manual reconciliation between the two systems, and a constant low-grade frustration that "the systems don't talk to each other." Most of this is solvable with the right automation layer. Here's what I build.

The four integration points that matter

Don't try to "fully unify" retail systems in one project — most retailers who try end up with a half-built integration that breaks more than it fixes. Instead, build these four integrations in order:

1. Unified inventory. Real-time sync between Lightspeed (in-store) and Shopify (online). A sale in either channel decrements stock everywhere within 60 seconds. New inventory entered in one system shows up in both.

2. Unified customer records. A customer who shops in-store and online has one record with one purchase history, regardless of which channel they used. Loyalty points accumulate across both.

3. Unified order routing. When an online order comes in for a product that's in stock at a physical store, the system routes it there for ship-from-store or pickup, rather than to the central warehouse.

4. Unified reporting. A single dashboard showing daily revenue, by channel, by store, by category, with the margins reconciled.

In that order. Trying to do them out of order — especially attempting unified reporting before unified customer records — creates dashboards full of duplicate counts and missing data, and the team loses trust in the system.

What the integration layer looks like

For a Canadian retailer with one Shopify store and 1-5 physical Lightspeed locations, the standard build:

• n8n as the orchestration layer, running on a self-hosted VPS or n8n Cloud.
• Lightspeed Retail API for in-store events (sales, inventory adjustments, customer creates).
• Shopify Admin API for online events.
• PostgreSQL as the unified customer database. Each customer has a single record keyed by email or phone, with sources from both systems linked.
• Webhook listeners on both systems trigger the sync flows in real time.
• Reconciliation jobs run hourly to catch any events that webhooks missed.

The trickiest part isn't any single integration — it's the customer-matching logic. Same person, in-store, gives "Sarah" as a first name, no last name, no email. Same person, online, has "Sarah Wilson" with email and phone. The system needs to identify these as the same customer without creating duplicates or wrongly merging unrelated people. We use a confidence-scored matching algorithm with manual review for edge cases.

What changes operationally

Retailers I've deployed this for typically see:

• Inventory accuracy goes from 92-95% (manual reconciliation, weekly) to 98-99% (real-time sync). Stockouts in one channel while sitting on inventory in another largely disappear.
• Online-to-store conversion (customers who browse online and buy in-store, or vice versa) becomes visible and trackable. Most retailers underestimate this by 3-5x.
• Customer lifetime value rises by 15-30% because the loyalty program actually rewards cross-channel behavior. Cross-channel customers spend more than single-channel customers — when they finally get credit for both, retention goes up.
• Ship-from-store becomes operational. Orders ship faster, shipping cost drops on local deliveries, and slow-moving in-store inventory clears.

The Canadian retail-specific challenges

A few things that matter more in Canada than in US retail playbooks:

• GST/PST/HST handling differs by province, and the sync needs to handle each correctly. A sale processed in one province with shipping to another needs the right tax math applied at the right point.
• Bilingual customer data (French/English) is a real concern for retailers selling across Quebec and rest-of-Canada. Customer language preference must be in the unified record.
• Carrier integration. Canada Post is the default for most Canadian retailers, with its own quirks. Some retailers use additional carriers (Purolator, FedEx, regional couriers) for specific routes. The integration layer needs to handle multi-carrier rate-shopping.

What it costs

A typical unified retail automation build for a 1-3 location Canadian retailer:

• Build cost: $6,000-$15,000 depending on integration complexity (the customer-matching layer is the biggest variable)
• Lightspeed and Shopify subscriptions: these you already pay, no change
• n8n hosting + ongoing variable costs: $50-$200/month
• Ongoing support: $300-$600/month for monitoring, edge case fixes, occasional new flows

Larger retailers (5+ stores, multi-channel beyond Shopify) get into more complex builds but the unit economics improve at scale.

What to skip

Two pieces of "retail AI" that I see oversold and rarely deliver:

• AI demand forecasting for inventory replenishment. Statistical models from your POS already do this well. Adding LLMs to it doesn't meaningfully improve forecasts for most retailers — the bottleneck is product-level data quality, not model sophistication.
• AI personalization on the in-store experience. Beacon-based targeting, AI mirror try-ons, dynamic price displays — interesting tech, almost never produces ROI for SMB retail. Save the money.

What does work: unifying the systems so the team can actually serve customers without checking three tabs to see if a product is in stock.

If you're running a Canadian retail operation where the in-store and online systems don't fully talk, book a discovery call. Or read more about how we approach workflow automation including retail integrations.