Issuu

In-Season Intelligence Protects Margin

Every season begins with a plan. You’ve analyzed last year’s data, set your margin targets, confirmed your buys, and mapped delivery timelines. The plan is sound. The numbers make sense.

Then reality arrives.

A trend peaks before your inventory arrives because a key delivery lands three weeks late. A competitor runs an aggressive promotion that shifts your wholesale accounts’ buying behavior mid-season. And your team—built to execute the plan, not to continuously monitor a thousand moving variables— doesn’t see the problem clearly until it has already cost you margin.

This is the central challenge in apparel today. It’s not that brands build bad plans—it’s that the season moves faster than the decision cycle, and thus plans become obsolete faster than most businesses can detect and react.

The brands that win in this environment won’t be the ones with the most precise forecasts. They’ll be the ones who see change earlier and act on it before their options narrow. Artificial intelligence (AI), embedded in and working alongside modern enterprise resource planning (ERP), will be the difference-maker—serving not as a replacement for human judgment, but as the early warning system that makes human judgment more timely and effective.

The Plan Was Never the Problem

Pre-season planning has genuine value. It establishes a shared frame of reference for the entire organization, acting as a baseline for buys, allocations, replenishments, and margin expectations. Without it, your team would operate without a compass.

But every pre-season plan is built on assumptions— about consumer demand, channel performance, supply chain reliability, competitive dynamics, and more. And it’s locked in months before a single unit hits the selling floor.

In a stable, slow-moving market, those assumptions might hold reasonably well. But in today’s apparel environment, they begin to erode almost immediately. Consumer preferences now shift faster and less predictably than historical data can capture.

Social media and the influence economy compress the window between a trend emerging and a trend peaking. That’s why the style that looked like a safe mid-season repeat buy can lose momentum before you’ve fulfilled production commitments. And the forecast you built in January isn’t wrong because your team made a mistake—it’s wrong because the market moved, and the forecast couldn’t.

So while pre-season planning still has an essential role, it must be treated as a starting assumption, not a locked-in commitment. And the question isn’t how to build a better pre-season forecast—it’s how to build an organization that can detect drift from the plan early enough to do something about it.

Why Seasons Drift, and Why It Happens

Quietly

Drift doesn’t announce itself. It accumulates across three predictable forces, each operating at the style level, often simultaneously.

Supply Variability

Late or uneven inventory receipts are a constant reality in apparel. When product arrives behind schedule, your selling window shrinks—a style that would have performed at full price for eight weeks might now only get five. ERP systems record these disruptions accurately, but you won’t see them until after the fact. By the time a delay shows up as a stock gap in your reports, it’s already cost you.

Demand Shifts

Weather deviations, viral social moments, and micro-trend cycles all move demand daily in ways that no pre-season model can fully anticipate. The speed of these shifts has accelerated significantly. Styles can go from on-trend to oversupplied in a matter of weeks, not months. Sell-through trajectories that looked healthy in the second week can turn within days.

Channel Divergence

Wholesale accounts, direct-to-consumer channels, and regional markets rarely move in lockstep. A style might be outperforming expectations online but sitting stagnant with wholesalers. These divergences create both hidden risk and hidden opportunity, but only if you’re looking at the data at the right level of granularity, in real time or close to it.

The Trouble With Detecting Drift

What connects all three forces is this: The drift they create is detectable early, but only if you have the right instruments. Your ERP contains the underlying data, but it’s designed to record transactions accurately, not to surface emerging patterns before they become visible problems. By the time drift appears clearly in a standard report, the season has moved on—and so have your best options.

The Real Cost of Waiting

Most apparel businesses understand, in principle, that late action is expensive. But the mechanics of exactly how waiting erodes margin are worth examining clearly.

Early in a season, when a style first begins to underperform, your options are numerous. You can make a targeted reallocation, moving inventory from accounts where the style is lagging to accounts where it’s selling through. You can test a modest promotional push to stimulate demand while the style still has momentum. Or you can adjust your forward buy before you’ve committed to more units.

These are low-disruption moves. They preserve pricing power. They protect the majority of the margin. Wait six or eight more weeks, and the picture changes entirely.

Excess inventory has built. The selling window has compressed. The accounts holding the product want relief. At that point, the primary lever is discounting, and the question becomes not whether to mark down, but how deep to go and how fast.

The financial impact is direct and significant, but there are longer-term costs as well. Repeated deep discounting trains your customers to wait for promotions. It puts pressure on full-price sellthrough across the rest of your assortment. And over time, it erodes the brand’s pricing authority and reputation.

The asymmetry here is important. The cost of acting early on a signal that turns out to be temporary is small—it’s a minor reallocation, a modest adjustment. The cost of acting late on a problem that has been compounding for weeks is large and often irreversible within the season.

Given that asymmetry, earlier action is almost always the right instinct. And the barrier isn’t willingness; it’s visibility.

The Road Ahead: AI’s Next Chapter in Fashion

Alain Tessier, Director of Product Management, Aptean

Right now, most AI tools in fashion are still primarily telling you what already happened—they can deliver a performance report from last month, an alert on a slow mover in the first quarter, a summary of last week. That’s useful, but it’s reactive.

Over the next two years, I expect the shift to be toward AI that tells you what to do next, with a clear rationale and projected outcome attached to the recommendation. So less after-the-fact reporting, more decision support.

I also expect AI to get much more deeply connected to the supply chain side of the business. Right now, demand signals often stay inside the brand and don’t travel upstream to suppliers fast enough. The brands that gain a real competitive edge will be the ones sharing those signals with their manufacturing partners in close to real time, enabling faster

production decisions, more flexible commitments, and ultimately shorter lead times. That kind of supplier collaboration, enabled by shared data, is still pretty underdeveloped in the industry.

And probably the most meaningful shift for the broader market: These capabilities are going to become increasingly accessible to mid-sized brands, not just the large enterprises that can afford big data teams and custom implementations. The technology is maturing quickly, and the barrier to entry is coming down.

That’s actually very much in line with what we’re focused on at Aptean—making sophisticated planning tools practical and accessible for companies that don’t have unlimited resources but still need to compete at a high level.

ERP and AI: Two Different Jobs, One Shared Purpose

There are two common misconceptions worth addressing directly: that AI is somehow in competition with your ERP system, and that investing in AI means reducing reliance on your core platform. Neither is true.

ERP has been and continues to be your system of record. It captures transactions accurately, enables management of inventory across your network, processes orders, tracks commitments, and provides the financial controls your business depends on. It tells you what happened, and that function is irreplaceable, no AI changes it.

What AI adds is a different capability entirely: the ability to detect what’s changing, before it fully shows up in your transactional record. AI observes the data within your ERP, continuously monitoring the signals; connecting performance patterns across styles, channels, and supply; and surfacing early indicators of drift before they become visible problems. It senses drift and flags it, and it can also model scenarios so you can evaluate trade-offs of potential decisions.

So in short: ERP tells you what happened. AI tells you what’s changing.

So this is not ERP versus AI. It’s ERP plus AI—the system of record and the intelligence working together and both connecting back to human judgment at the center. When a merchant or planner receives an early signal that a style is drifting from plan, they still decide what to do. AI doesn’t replace that decision—it gives the person making it better information, earlier, so they can have greater confidence acting on it.

The organizations that get this architecture right gain a meaningful operational advantage. Their teams aren’t spending time hunting for problems in reports. They’re responding to curated signals, with the context they need to act, at a point in the season when acting is still low-cost.

AI’s Role: Continuous Surveillance at Style Level

The scale of modern apparel operations makes continuous, manual performance monitoring impossible. A mid-sized fashion brand might carry thousands of active SKUs across multiple channels, regions, and account types. No planning team can watch all of that at once—and even if they could, the volume of data would make it difficult to distinguish signal from noise.

This is where AI earns its place. Not by making decisions on its own, but by conducting the surveillance that no human team could sustain at scale. Specifically, AI applied to in-season management can:

» Monitor the performance of every style, in every channel and region, continuously—not just in a weekly or monthly review cycle.

» Detect drift across supply, demand, and channel dimensions simultaneously, flagging styles where multiple signals are moving in the wrong direction at once.

» Quantify trade-offs through scenario modeling in order to present leaders with a clear view of what different responses would mean for margin, sell-through, and forward inventory.

» Surface the signals that require a human decision and alert the right people at the right time, rather than burying them in dashboards that require a human to go looking.

How AI Works To Protect Margin, Step by Step

Given clean, well-organized data, the process through which you collaborate with AI to implement in-season intelligence is relatively simple.

1. AI monitors real-time data and surfaces trends

2. Leaders receive earlier performance signals

3. Decisions are made sooner and with more confidence

4. Actions are smaller and less disruptive, avoiding risk

5. Margin is protected

A human remains at the center of every decision—AI simply ensures that the right decision is made at the right moment in the season, not too late.

The practical effect is a compression of the timing gap. The window between when drift begins and when leaders see it shrinks from weeks to days, and the options available to your merchants and planners multiply significantly. Smaller corrections become possible, and pricing power is preserved. The season ends with better margin and less excess inventory than it would have otherwise.

Built for Fashion: Why Industry-Specific Technology and Clean Data Are Non-Negotiable

Aly Breeman, Senior Product Manager, Aptean

The main issue with generic software is that it often treats each variant of a garment as a completely separate item. So, for example, if you have one style in four colors and eight sizes, a generic system can quickly turn that into 32 different items (SKUs) to manage. But that’s not how fashion businesses think about their products, and it creates unnecessary complexity right from the start.

In fashion, one style is really the core product, and then you have the variants underneath it— the colorways, the sizes, and sometimes other dimensions as well. Industry-specific software understands that structure. It allows brands to manage one style with all its related variants in a much more intuitive and accurate way.

That has a big impact across the business. It improves product visibility, makes inventory management easier, supports better merchandising decisions, and gives teams a clearer view of what’s actually happening with the collection. It also reduces the risk of errors and duplication, which is especially important when you’re working across large, fast-moving collections.

So, in short, industry-specific technology doesn’t just digitize an existing process; it supports the way fashion businesses actually operate—and that leads to better efficiency, better control, and better decision-making, end to end.

What companies should be doing to prepare their systems and data management processes for AI really comes down to one key thing: getting the foundations right. AI is only as good as the data it works with, so if the underlying information is incomplete, inconsistent, or poorly structured, then the outputs will be too.

In terms of risk, one of the biggest issues is trying to layer AI on top of poor-quality data or fragmented systems. If different teams are working from different versions of the truth, or if product data is not maintained properly, then AI will only amplify those problems. So companies need to focus on data quality, governance, and consistency before expecting AI to deliver value.

In other words, the more aligned your systems are to the industry, and the more disciplined your data management is, the more effective AI will be. So it’s not just about adding AI; it’s about making sure the business is ready to use it well.

The Foundation That Makes It Work

While it can make a world of difference for your brand’s profitability season to season, it’s critical to keep in mind that AI is not a plug-and-play solution. Its ability to surface meaningful, actionable signals depends entirely on the quality and structure of the data it works with. In the apparel industry specifically, this is where many implementations fall short—not because the technology isn’t capable, but because the data foundation isn’t ready for it.

That’s in large part because apparel business data is inherently complex. A single style may exist in dozens of color/size combinations. Meanwhile, collections are organized by season, division, and channel, and performance needs to be evaluated across all of those dimensions simultaneously.

Generic software systems frequently flatten this complexity in ways that create confusion and data quality problems. Treating each SKU variant as a separate, unrelated item rather than as part of a coherent style hierarchy, for example, results in fragmented data that AI cannot interpret correctly.

Industry-specific technology is the solution for this challenge, because it mirrors the way apparel businesses actually operate. When your system of record organizes data the way your merchants and planners think about it—by style, with variants, seasons, and channel hierarchies intact—your AI has the context it needs to accurately identify patterns, generate reliable forecasts, and provide coherent recommendations.

The other key component of a solid foundation is more organizational in nature. Teams working from disconnected systems, or from their own local versions of the data, introduce inconsistencies that compound as AI tries to connect signals across the business. Thus, establishing and operating from an organization-wide single source of truth with clean, unified, accessible data across design, merchandising, supply chain, and finance is a prerequisite for AI to deliver the value you expect.

Putting It Into Practice: Aptean Fashion & Apparel

The architecture we describe here—ERP as system of record, AI as intelligence layer, humans as decision-makers—is how Aptean Fashion & Apparel is built to function.

Built on Microsoft Dynamics 365 Business Central, it’s a purpose-built ERP for fashion and apparel brands and wholesalers. It manages the full complexity of the industry—from style-based product hierarchies and seasonal planning to pre-season bookings, allocation, and markdown management—in a single connected platform. And because the system of record is developed industry-specific from the ground up, you can be confident that the data foundation your AI depends on is structured correctly before the intelligence dives in.

Speaking of, our vertical AI—Aptean Intelligence is not bolted on as a separate tool. It’s embedded in both our AppCentral platform and our apparel ERP,

giving teams access to AI capabilities in the context of the workflows they are already executing. This is an important distinction, as AI that sits alongside your ERP requires your people to move between systems and translate between contexts. AI that is woven into your ERP instead surfaces insight where the decision is actually being made.

The AI agents already available for Aptean Fashion & Apparel reflect exactly the kinds of in-season signals we’ve discussed. For example, the style performance analyzer agent monitors sell-through and inventory timing to identify styles at risk of missing their selling window, enabling proactive intervention before excess stock accumulates. The style substitution agent matches and scores similar products by material, color, season, price and more, perfect for those instances when stock runs low on a popular item mid-season—where the sale might otherwise be lost, now you can quickly find and propose a replacement for the customer.

Beyond agents, AppCentral’s GenAI Query gives every team member the ability to ask natural language questions of their business data, gleaning the insight that previously required an analyst or custom report. And Role-Based AI Workspaces bring together the signals, actions, and tools most relevant to each role, reducing the time your people spend navigating systems and increasing the time they have to spend on decisions that matter.

The broader AppCentral platform also connects Aptean Fashion & Apparel to a curated set of industry-specific applications, including product lifecycle management (PLM) for product development, electronic data interchange (EDI) for trading partner communication, seamless shipping tools and integrated payment solutions. This ensures your brand has an end-to-end foundation without the complexity of assembling and maintaining disconnected point solutions.

Now you know that the competitive advantage in apparel is no longer built exclusively in the preseason plan. It’s built in the weeks between plan and performance—in the ability to see what’s changing before it costs you margin, and to act with the confidence that comes from having a clear, connected picture of your business and the market.

Aptean Fashion & Apparel is designed to deliver on those fronts. If your business is ready to move from reactive to proactive—to see earlier, act sooner, and protect the margins you’ve worked to build—let’s have a conversation.

Making the Shift: AI Adoption and Change Management

Ken Weygand, Solutions Architect, Aptean

When it comes to change management with AI, first remember that you’re dealing with humans. They may think they’re being replaced by the technology, but that’s not the case. AI is meant to improve the efficiency of tasks being performed and to enhance the decision-making process, not replace it. You have to keep in mind that you’re transforming your employees.

Then, as you’re looking to deploy AI within your company, go for the quick wins. Quick wins build internal trust and momentum. So start with using AI to address specific business problems, like inventory planning or marketing content generation, and then expand use cases according to your goals.

Lastly, I’ll mention data readiness. AI is great at working through massive amounts of data, but that data needs to be clean, unified, and accessible. Disconnected systems and data residing in silos are going to slow the rollout and limit the impact of AI. Put the effort in on the front side to get your data ready.

Brands should take the plunge with AI because demand is more unpredictable than ever before. You have trends that can change in a heartbeat—a couple of TikTok or Instagram posts, and sales could be through the roof. Of course, that’s if you planned accordingly and have the inventory to fulfill the sudden spike in demand.

AI can help you analyze large datasets—including but not limited to social signals, sell-through rates, even things like weather and regional preferences— to help forecast demand more accurately than using traditional methods.

As for first steps, I’d recommend starting with a focused use case that can have a positive impact on the brand, such as having it help with size and assortment planning for a set of styles. You can call this a pilot program. Compare the results of the AI-driven decisions versus the expected results of decisions made without AI to measure the downstream improvements.