5 Integration Mistakes That Will Haunt Your Black Friday

Black Friday and Cyber Monday aren't just a test of your website. They're a stress test for every system behind it — your ERP, your inventory management, your payment processing, your fulfilment workflows, and every integration that connects them.

We've supported businesses through enough peak trading seasons to know what breaks and why. These are the five mistakes that cause the most pain, and every single one of them is preventable.

1. No load testing on your integration flows

Your integration handles 200 orders a day without issues. Black Friday brings 2,000. What happens?

Most businesses don't know the answer until it's too late. Integration platforms have throughput limits — API rate limits on the source and destination systems, concurrency caps on the integration platform itself, and queue depths that can overflow under sustained load.

We've seen order syncs that work perfectly at normal volumes start silently dropping records when volume spikes. The integration doesn't error out — it just can't keep up, and orders sit in a queue for hours or get lost entirely.

The fix: load test your integration flows before peak season. Push through 5x your normal daily volume in a test environment and see what happens. Check processing times, error rates, and queue depths. If something breaks at 5x, it'll definitely break during a flash sale.

On Celigo, pay attention to your flow concurrency settings and the rate limits on your NetSuite or ERP connection. On Workato, check your task limits and recipe throughput caps.

2. Inventory sync isn't fast enough

If your inventory sync runs every 15 minutes, that's a 15-minute window where your website might be selling products you don't actually have. At normal volumes, the overselling risk is small. During a Black Friday sale, you can sell out of a popular SKU in minutes.

We worked with a retailer who ran a flash sale on a hero product. Their inventory sync ran every 30 minutes. In the gap between syncs, they oversold by 140 units. The cost of cancelling those orders, refunding customers, and dealing with the complaints far exceeded the revenue from the sale.

For peak trading, increase your inventory sync frequency. Ideally, inventory decrements should be near real-time — triggered by the order event rather than running on a schedule. If your platform supports webhooks for inventory changes, use them. If not, reduce your sync interval to the shortest period your systems can handle without hitting rate limits.

3. Error handling that relies on humans

During normal operations, a few integration errors a day is manageable. Someone checks the error dashboard in the morning, resolves the issues, and moves on. During Black Friday, error volumes can spike tenfold.

If your error resolution process requires a human to manually review and retry every failed record, you'll fall behind within hours. By Saturday morning, you'll have hundreds of unresolved errors, orders that haven't reached the warehouse, and customers who haven't received shipping confirmations.

Build automated error handling before peak season. Categorise errors by type. Transient failures — API timeouts, rate limit hits, temporary connection issues — should retry automatically with exponential backoff. Data validation errors should be routed to a queue with clear context so they can be resolved quickly.

Set up alerts for error volume thresholds. If your normal error rate is 5 per hour and it suddenly jumps to 50, someone needs to know immediately — not when they check the dashboard on Monday morning.

4. Payment and order data don't reconcile

This one often doesn't surface until after the peak period, when finance tries to reconcile. Orders in Shopify don't match orders in NetSuite. Payment amounts don't add up. Refunds are missing. Discount codes applied on the frontend aren't reflected in the ERP.

The root cause is usually one of two things: either the integration isn't capturing all the order data (partial payments, split payments, gift cards used as partial tender), or timing differences between systems create mismatches.

Before peak season, run a reconciliation on a sample of recent orders. Pick 50 orders at random and verify that the data in your ecommerce platform matches your ERP exactly — line items, quantities, prices, discounts, taxes, shipping charges, and payment amounts. If there are discrepancies now, they'll be amplified tenfold during peak.

Pay special attention to discount handling. BOGO offers, tiered discounts, and coupon stacking create complex pricing scenarios that integrations often struggle with. Test your most aggressive planned promotions through the full order-to-ERP flow before going live with them.

5. No rollback plan

What's your plan if an integration fails catastrophically during peak trading? If the answer is "we'll figure it out," you're not ready.

We've seen businesses lose hours during peak season because an integration broke and nobody knew how to fall back to a manual process. Meanwhile, orders are piling up, the warehouse is idle, and customers are getting worried.

Have a rollback plan for every critical integration flow. For the order sync: can someone export orders from your ecommerce platform as a CSV and import them into the ERP manually? Do they know how? Have they practised? For inventory: is there a way to manually update stock levels on the website if the automated sync fails?

You don't need to plan for every possible failure. Focus on the critical path — orders in, inventory out, fulfilment triggered. If those three things can keep moving even when an integration is down, you'll survive peak season.

Document the rollback procedures, make sure the relevant team members know where to find them, and ideally do a dry run before the peak period starts.

The prep window is shorter than you think

Most of these fixes aren't complicated, but they take time to implement and test properly. If you're reading this in October, you've got enough time. If it's mid-November, you're cutting it fine.

At minimum, the month before peak trading should be a code freeze on your integration flows. No new features, no changes to field mappings, no updates to transformation logic. Just monitoring, testing, and making sure everything is solid.

If your integration architecture isn't ready for peak season and you want help getting it there, get in touch. We've done this prep work with enough businesses to know what needs attention and what can wait.