De-risking complex retail tech stacks
The retail tech stack is growing – but so are your risks. Here's how to stop it becoming a mess. The recent Retail Technology Show in London had so...
7 min read
Darryl Kennedy May 9, 2025 7:00:00 AM
Cloud migration has become a badge of honour over the past decade. From ERP to eCommerce platforms and CRMs to CDPs, enterprise retailers have scrambled to “modernise” and “digitally transform” by moving off-prem and onto the cloud.
But here’s the uncomfortable truth: just because it’s cloud-based doesn’t mean it’s bulletproof.
In fact, for many retailers, the cloud has introduced more risk, and half of cloud transformations are abject failures.
We’ve worked with some of the UK's most recognisable retailers – Lovehoney, Currys, NOTHS, NBrown – to de-risk and performance-proof retail systems. And we’ve seen one dangerous assumption continually pop up: retailers assume the cloud is a safety net.
It’s not.
It’s just a different kind of infrastructure – with its own blind spots, bottlenecks, and breaking points.
Migrating to the cloud often solves one set of problems (legacy bloat, scalability limits, infrastructure costs) but introduces new challenges around integration, performance, testing, and visibility.
Traps, you could call them.
In fact, we will call them that. And in this article, we’ll break down a couple of those big traps, plus what you can do to avoid them.
The assumption: Amazon or Salesforce host it, so it must be reliable, right?
The reality: You’re still on the hook for how it performs, especially when your customers flood in during peak.
Scalability is one of cloud technology’s biggest selling points: the ability to handle surges of users or transactions by flexibly adding resources. But your own custom code, integrations, workflows, and your ERP, order management, or other back-end systems? They don’t just magically meet cloud’s yardstick. Nor do they automatically cope with increased volume.
With massive traffic spikes during sales or seasonal events always on the horizon, optimising for peak becomes a never-ending worry. So, you’d be forgiven for banking on cloud scalability to get through. However, cloud scalability is not automatic – misconfigurations or architecture bottlenecks can (and will) leave your site unable to cope with high demand.
The buck stops with you. If you haven’t correctly configured your cloud environment, auto-scaling features might not work as expected, leading to poor performance (or skyrocketing costs) during unexpected traffic surges. One forgotten setting or an unoptimised piece of code can throttle throughput when it matters most. The same rings true for all your back-end systems, too. We spend a decent portion of our time talking to new clients about this – cloud scalability is all for naught if your OMS falls over just as order volume starts to get exciting.
And no business is exempt. Even juggernaut retailers hit ceilings. We still think back to 2019 when Costco learned the hard way that infrastructure must be primed for peak load. They let the ball slip, trusted in cloud to auto-scale (we assume), and their websites struggled or went offline for a whole day when hit by holiday-level traffic – resulting in an estimated $11m loss. Not to mention the reputational damage. And annoyed customers.
These incidents underscore a vital point: scalability isn’t just about servers.
And so we come to the crux of the matter – testing scalability under real-world conditions. Annual load tests before the holiday season on-prem are a thing of the past (at least, we’re trying to make them that way).
Cloud demands continuous and more complex testing. But with a little forethought and good practice, like:
You will catch issues like your scaling rules being too conservative or a third-party service (e.g. payment gateway or identity provider) can’t handle the load.
What to remember:
Test for performance and scalability under real-world conditions.
Simulate Black Friday traffic, not just smoke tests.
Use observability tooling to spot weak links.
Don’t assume the platform alone will save you.
You must earn scalability in the cloud. It gives you the tools (elastic compute, global CDNs, database clustering, etc.), but retailers must architect their applications to use them effectively.
Worried about scaling for peak? Here’s how we ensured Not On The Highstreet easily handles peak trading, so their team can actually enjoy their holiday!
The assumption: Since the vendor tests the cloud platform, we don’t need to worry as much about quality assurance. Right?
The reality: Retail performance failures rarely come from the cloud provider. They come from everything you’ve built on top of it.
One of the first hurdles in cloud adoption is integrating new cloud systems with existing retail technology. In fact, retail technology leaders consistently cited integration as a major challenge – nearly 90% of organisations report connecting cloud services to existing systems is a “common” or “very common” problem.
Legacy databases, ERP systems, payment gateways, and other on-premises applications not designed to work in a cloud environment are constant hurdles for large retailers. It’s nearly impossible to escape technical and operational challenges during cloud projects, especially around integrating existing systems and managing data migration.
Plus, cloud platforms update regularly. Getting live is just the first step – many retailers overlook how cloud updates impact systems on an ongoing basis. We do a lot of regression testing for your existing clients to ensure that major Salesforce and NetSuite upgrades don’t impact integrations or any unique customisations they have applied (and that’s something no vendor can or will do).
Al in all, this means a list of issues that can disrupt your critical business processes.
Older applications may not be fully compatible with modern cloud platforms, leading to errors or data loss during migration. Without specialised middleware or refactoring to communicate with your cloud services, migrating your legacy inventory or order management system can result in corrupted data or broken functionality.
A business’s worst nightmare. Without robust integration, information may not flow correctly between systems – orders, inventory counts, or customer data can get out of sync. A cloud-based eCommerce front end, for instance, must reliably fetch stock levels from a legacy warehouse system; if that integration fails, the website might sell items that aren’t actually in stock or fail to process orders.
Everywhere you look, you’ll find interdependencies. Between POS systems, e-commerce platforms, CRMs, and supply chain systems – everywhere. Migrating one component to the cloud without mapping all its dependencies can trigger unexpected failures. One of the most common hiccups comes from when retailers overlook how many downstream systems (the Retail Technology Iceberg, as we call it) rely on a piece of data or service. One overlooked API or data feed that store kiosks, mobile apps, or analytics platforms depend on can cause a domino effect of issues.
The bottom line is this: integration requires thorough planning and testing. It’s not enough to migrate applications; you must re-architect data pipelines and interfaces so that cloud and on-prem systems work seamlessly together.
What to remember:
Treat cloud platforms like any other system:
Own the full test strategy (or find a partner that can help you do so)
Validate end-to-end journeys across POS, ERP, eComm, CRM
Don’t outsource responsibility to your SI or platform provider
Automate regression cycles to keep up with release velocity and regular cloud updates
Custom logic. Personalisation layers. ERP syncs. Third-party add-ons. Payment workflows. It’s your responsibility to ensure these things don’t break – because the vendor’s SLA won’t cover your checkout crash.
Skipping this work can leave a retailer with one foot on each side – and a giant crack in between.
Want to eliminate bottlenecks, blind spots and last-minute chaos? Here’s how our comprehensive test service ensured that Currys were confident with change after every sprint.
Migrating to the cloud is a significant leap forward for retail technology, but it doesn’t make your tech stack bulletproof. Yes, you’ll gain flexibility and scalability, but you also inherit new responsibilities and risk areas.
“Bulletproof” retail tech isn’t made in the cloud console. It’s forged through testing – before the pressure hits. That’s a gargantuan topic to cover, and we don’t want to be reductive. But we want to be helpful, so here’s a checklist of everything you should be testing.
Cloud infrastructure doesn’t fail because it’s not powerful enough. If it fails, it’s because you didn’t model reality.
Test like your peak depends on it…because it does:
Simulate peak trading conditions (e.g., Black Friday, influencer spike, January returns rush). Realistic traffic. Real concurrency. Real stress.
Test scaling mechanisms: Auto-scaling only helps if you validate the triggers. Don’t assume your cloud setup will respond fast enough.
Evaluate latency across regions: Many cloud platforms throttle or degrade under load. Measure API response times at scale across your key markets.
Retail success depends on orchestration across multiple systems. A single API timeout or field mismatch can break everything.
Don’t just test systems. Test flows:
Verify full journey continuity: From online checkout to ERP to warehouse dispatch to returns. Validate the actual chain, not just individual links.
Inject failures: Turn off integrations mid-flow. Kill services. Throttle APIs. Then test how your system recovers.
Model chaos: What happens when the CRM sends corrupted customer data? When OMS misses a shipment status update? If finance receives duplicate records?
Cloud platforms change constantly. So do your dependencies. If you're not testing continuously, you're testing too late.
Operational stability is a moving target, so hit it before it moves:
Automated regression testing: For every release, integration, or patch from a third-party SaaS platform.
Service change detection: Test what happens when your eCommerce platform or payment provider silently changes behaviour.
The cloud can undoubtedly strengthen a retailer’s capabilities – if used wisely. But without due diligence, a cloud-based system can fail just as spectacularly as any other.
And enterprise retail doesn’t forgive failure (and your customers don’t forget). So, for large online retailers with complex ecosystems: remain clear-eyed and proactive after migrating. Thorough integration planning, retraining of teams, retooling of processes, and robust testing at every level are mandatory.
This checklist is just a starter: a taste, a morsel. Want to see how ready you actually are?
Book a FLOW session with Spike! You’ll get a full Clarity of Vision Blueprint to assess your change pipeline, tech risk posture, and go-live certainty in just 4 weeks.
From there, you have options. But if you have even a niggle of doubt, it’s worth understanding where you are and where you need to get to. That will put that niggle to rest.
We are Change Flow Maximisation Experts for Retail Technology.
We help retailers to deploy technology change with certainty. We’re all about proving quality and performance and creating visibility and confidence so you can increase the pace of change, eliminate bottlenecks and blind spots, and remove last minute go-live chaos.
More features, slick platform integration and increased performance drive higher customer retention and increased order value. And long-term gains through a simpler, scalable tech stack and lower tech debt.
We work exclusively with retailers bringing you a Clarity of Vision Blueprint within 30 days and Go-Live Certainty at Speed within 6 months using our unique FLOW mechanism, our proven set of tools and techniques.
The retail tech stack is growing – but so are your risks. Here's how to stop it becoming a mess. The recent Retail Technology Show in London had so...
1 min read
2024: another year of pressure for retailers. As if running an online retail business wasn't already a circus, retailers had to brace and...
Here is the final blog in our ‘Quickly starting with Test Automation’ series, part 3 – Advanced Automation tips. We advise you read parts 1 and 2...
1 min read
That is the question on everyone’s minds. The coronavirus has shaken up the world and left many of us feeling dizzy at the prospect of not only how...
The COVID-19 outbreak has been unique in the way we’ve seen connectivity and data not only influence the global response, but help to coordinate it...