The five new rules of markdown planning: from static calendars to self-learning systems

The static markdown calendar: familiar, but failing

The traditional markdown calendar served a purpose in its time. It allowed teams to plan inventory flows, schedule campaigns, coordinate marketing, and align on key trading dates. But in practice, it now does more harm than good.

It’s inflexible. If a product line tanks in week three, but markdown isn’t “scheduled” until week six, you’ve just lost three weeks of potential trading margin — and made the eventual discount deeper than it needed to be.

It’s blind to performance. The calendar doesn’t care if a line is flying off shelves or gathering dust. Everything moves through the same schedule — regardless of demand, inventory cover, or price sensitivity.

It treats all products the same. Not every product deserves the same treatment. A fast-moving essential with high elasticity needs a different markdown rhythm than a premium seasonal item or long-tail replen.

The result of all this? Markdowns become reactive, rather than proactive. Teams scramble to course-correct. Clearance piles up. And promotional performance becomes a guessing game: “that one worked, this one didn’t, let’s try again next season.”

Thankfully, there’s a better way.

Enter the era of self-learning markdown systems

Modern retail demands agility, not rigidity. And that’s where AI-powered markdown planning fundamentally changes the game.

At its heart, this shift is about moving from pre-defined calendars to data-driven, dynamic decisioning. From guessing to modeling, and from rules to learning.

The five rules of the new markdown planning model

1. Plan with probabilities, not predictions

Rather than saying “we’ll markdown these lines in week six”, retailers should think differently and more strategically. For example; “If sell-through is below X%, and demand is trailing forecast, initiate markdown simulation.”

AI models can continuously ingest new information (sales velocity, stock levels, demand curves, competitor pricing) and adjust the plan in real time — offering the most profitable set of options for action at the moment of decision.

2. Segment by strategy, not just category

Different products, regions, and channels have different roles in your commercial model:

• Hero SKUs: Maintain price, protect brand
• Volume drivers: Optimize for revenue
• Seasonal stock: Clear quickly
• Premium lines: Trade carefully, protect perception
  

AI tools allow you to assign different strategies and guardrails to each group — and apply pricing actions accordingly. This protects margin and ensures promotions are aligned with commercial intent.

3. Move from cyclical to continuous

Markdown planning used to happen in “rounds” — each season had three or four phases, and that was it. Now, planning is continuous.

Self-learning systems are constantly evaluating:

• Which products are underperforming?
• Where is demand softening?
• Which stores are accumulating slow-moving stock?
• Where could a shallow markdown unlock demand?

This turns markdowns from a quarterly exercise into a daily optimization loop.

4. Simulate before you act

AI models don’t just suggest markdowns; they simulate outcomes across multiple objectives. For example:

• “If we apply a 20% markdown to Line A, we expect to sell through 65% of remaining inventory in 10 days, recover £42,000 in gross profit, and reduce clearance risk by 40%.”

• “If we wait another week, we’ll likely need to discount at 30% to hit the same result.”
  

This allows commercial teams to choose the best option with full context—balancing risk, revenue, and margin intelligently.

It’s the difference between firefighting and forecasting.

5. Measure learning, not just performance

The most valuable part of self-learning markdown systems isn’t just the uplift they generate — it’s the feedback loop they create.

With every promotion or markdown:

• Elasticity curves are updated
• Seasonality profiles are refined
• Store or channel anomalies are flagged
• Discount thresholds are recalibrated
  

Over time, the system doesn’t just execute better markdowns, but gets smarter at planning them. It learns what works for your unique business, brand, customer, and trading rhythm.

And crucially, it frees your people to focus on why things work, not just what needs doing.

What this looks like in practice

Let’s bring this to life with an example from a project I led with a mid-market fashion retailer.

Old approach

• Seasonal calendar dictated markdown dates
  
• 30% markdown applied across all mid-season lines
  
• Performance reviewed two weeks later
  
• Significant over-discounting in some SKUs; sell-through still weak in others

New approach with AI-powered markdown engine

• Markdown triggers linked to real-time sell-through thresholds
• AI surfaced five options for each SKU, ordered by profit impact
• The team chose optimized markdowns (some as low as 10%, some deferred entirely)
• Gross margin improved by 410bps across the campaign
• Inventory cleared 12 days faster than previous season

The key? They didn’t “plan” the markdown months ahead. They let performance, elasticity, and stock position guide the plan, as it unfolded.

How to build a self-learning markdown capability

This kind of transformation isn’t just about tech. It’s about mindset and capability. Here are some simple steps to follow:

1. Audit your current calendar

Where are markdowns pre-set and rigid? Where do they ignore performance?

2. Define markdown strategies by product role

Assign commercial roles (protect, push, clear, test) to each line or category

3. Centralize data feeds

Ensure sell-through, pricing, inventory, and external signals are accessible to your pricing tools or teams in near real-time

4. Run controlled simulations

Use AI or analytics tools to test multiple markdown scenarios in key categories

5. Shift to weekly or rolling reviews

Ditch static planning meetings in favour of fast-paced, cross-functional trading huddles with live data and AI insight

Final thought: the calendar doesn’t know your customer

Static markdown calendars are comforting, but they’re blind to nuance, change, and reality. And, in a world where every percentage point of margin matters, comfort isn’t good enough.

Markdown planning should be a living process, not a fixed calendar. And the retailers who embrace adaptive, self-learning systems will find they no longer have to choose between profitability and agility. The best plan isn’t one you wrote three months ago, but the one that writes itself every day.