It’s early autumn, and retailer software teams know what that means: being feverishly heads down on performance, load, and scalability testing in anticipation of Black Friday and Cyber Monday (BFCM) and the people hunting down deals online.
Or, at least, that’s how it was before AI took over the online shopping world.
Many shoppers this season won’t be humans; they’ll be AI agents. Shopify’s 2025 holiday report found that 64% of shoppers planned to use AI tools that season. This shakes up performance and load testing because traffic sources are unpredictable and retailers increasingly rely on AI within their sites and testing workflows.
Another wrinkle in BFCM 2026 is that, while BFCM is still very relevant, the event is now a longer holiday demand curve than a one-weekend traffic spike. Adobe reported that Cyber Week represented 16.3% of holiday spend in 2022, 16.9% in 2024, and 17.2% in 2025. That is an enormous amount, but the data shows the season is broader than the event itself.
Many software teams load test once a quarter, and some once a year. But with longer shopping cycles full of unpredictable AI shoppers, BFCM readiness now depends on continuous performance validation. This validation must occur across the full revenue path, including AI-driven experiences, APIs, third-party dependencies, and business-critical user journeys.
The changing shopping landscape
BFCM remains a major cash cow for retailers, but the old “one giant weekend” frame of BFCM is only half true now. Consumers are now shopping more throughout the year than over a few days. The National Retail Federation (NRF) found that 46% of consumers had started holiday shopping before November in 2022, which was up from 39% in 2019.
Why is this? It’s partly due to ongoing discount sales: Adobe noted in a report that major shopping days like BFCM were “losing prominence” as discounts spread across more of the season.
Part of the reason also has to do with shoppers’ increasing use of AI. Adobe found that traffic from generative AI sources to U.S. retail sites rose 1,300% during the 2024 holiday season and was up 1,950% year over year on Cyber Monday.
Retailers, too, are increasingly relying on AI. Shopify’s report found that nearly nine in ten businesses were investing in AI-powered discovery.
That changes performance and load testing because traffic sources are less predictable, as more users arrive through AI-assisted discovery rather than just paid search, email, or direct traffic.
But it also changes performance and scalability testing from the other side. The application surface is wider because retailers increasingly depend on conversational support, AI recommendations, search relevance layers, and model-backed experiences. Demand for AI goes even deeper for retailers: teams now expect AI help inside the testing workflow itself.
While BFCM still matters disproportionately for retailers, winning now requires readiness for a longer, noisier, more distributed peak.
Why BFCM readiness means year-round performance validation
AI may have changed the shape of retail traffic and the complexity of the retailer’s tech stack, but it does not change the fundamentals of performance engineering. Teams still need to prove that the most revenue-critical journeys meet acceptable latency, error, and stability thresholds under realistic demand.
But most teams test performance once a year with a specialist and hope it holds at the next peak event or go-live. Sometimes, it doesn’t hold, and teams find this out at the worst possible time. When this happens, the result is very expensive and very public.
With longer shopping cycles, faster release cycles, and unpredictable performance due to AI-referred traffic (and sometimes AI-powered features), this annual cadence is no longer acceptable for many retailers.
Instead, BFCM readiness has shifted from peak-load and scalability testing to continuous performance validation.
Continuous performance validation means teams now need to continuously validate that every revenue-critical journey, including AI-assisted discovery and service flows, still performs under real demand before peak season arrives. They certainly still need the old disciplines of load, stress, spike, and monitoring. But these ideas now need to be embedded into release workflows and tied to business-critical journeys rather than treated as a quarterly or annual infrastructure exercise.
More specifically, great BFCM teams should be continuously validating the following, long before peak season arrives:
- Search
- Cart
- Checkout
- APIs
- Mobile flows
- Global regions
- Performance evidence that engineering and business stakeholders alike can read
The buying journey is now longer, more API-dependent, and increasingly AI-shaped. Teams need to adjust their performance testing accordingly, by transforming it into continuous performance validation.
Leapwork Go: The continuous performance validation engine
Leapwork Go helps teams prove critical customer journeys will perform when demand is highest. Instead of waiting for a pre-holiday performance test, teams can continuously validate search, cart, checkout, APIs, mobile experiences, and other revenue-critical workflows year-round.
By making load and performance testing efficient enough to run every sprint, Go helps teams catch slowdowns, capacity issues, and regressions before they become customer-facing problems. Teams can recreate real-world traffic, test at scale across more than 20 global regions, and bring performance validation directly into CI/CD so every release is evaluated against the same performance expectations.
Getting started does not require rebuilding everything from scratch. Teams can record real-world user traffic or reuse existing JMeter and HAR files, then scale those scenarios without scripting or managing performance-testing infrastructure.
The outcome is more than a successful load test. It proves the site and infrastructure your revenue depends on can withstand real demand. So when Black Friday arrives, and both human shoppers and AI-driven commerce put pressure on your systems, your team isn’t hoping the site will hold up. You have been proving it all year.
