Integrated Business Planning - Skincare
How AI-native connected planning helps CPG brands forecast earlier, commit inventory with confidence, optimize trade spend, and continuously rebalance once the season begins.

One SKU. Five Channels. A Buy Committed 100 Days Before the Event.
In CPG, the biggest planning decisions are often made months before actual demand shows up. Take one Vitamin-C serum SKU sold across five channels.
The manufacturer needs roughly 90–100 days of lead time. That means the business has to commit the November buy in July — before Black Friday demand is visible, before channel performance is clear, and before most of the season has even started.
That creates a simple but expensive question:
How do you make the right decision early, and still adapt when reality changes?
That is the problem connected planning is meant to solve.
The planning challenge
For this serum, assume:

The important point is not the forecast itself.
It is what the forecast triggers.
Demand Forecast → Purchase Order → Inventory → Allocation → Promotion → Revenue → Margin
If those decisions are planned separately, teams usually compensate by adding buffers. The result is often too much inventory overall, but not enough inventory in the right channel at the right time.
1. Forecast early enough to influence the buy
For a November event, the forecast has to become actionable in July.
TrueGradient forecasts demand at the level where the business needs to make decisions:
SKU × Channel × DC × Week
In this example, the November demand plan is approximately 152,000 units, including expected promotional uplift.
The key idea is straightforward:
A forecast is most valuable when it is accurate while the business can still act on it.
A perfect forecast in late October does not help much if the factory required the purchase order in July.
2. Turn the forecast into a supply commitment
The demand signal then feeds directly into supply planning.
Based on demand, lead times, inventory, safety stock and operating constraints, the business commits roughly 160,000 units to the manufacturer.
Instead of demand planning and supply planning working from different versions of the truth, the same number connects the two.

This is where the benefit of connected planning starts to become tangible.
The demand forecast is no longer just a planning output. It becomes the starting point for an operational and financial commitment.
3. Once inventory lands, the question changes
By late October, the product reaches the distribution network.
Now the question is no longer:
How much should we buy?
It becomes:
Where should the inventory sit?
The initial allocation may look something like:

But demand will not follow the original plan exactly.
Amazon may accelerate.
Walmart.com may outperform.
Some stores may slow down.
Another region may start selling out.
A static allocation cannot respond to that.
4. Rebalance weekly as demand becomes real
Once the season starts, the planning process should become a continuous loop:
Actual Sales → Reforecast → Inventory Check → Reallocation → Next Decision
In the example above, the system identifies excess inventory in slower locations and recommends moving approximately 9,200 units into Amazon FBA and Walmart.com.
That single decision can simultaneously:
- reduce lost sales,
- improve in-stock availability,
- avoid future markdowns,
- release working capital,
- and protect margin.
This is why in-season planning is not simply about forecasting again.
It is about turning new information into a better operational decision.
5. Trade promotion should use the same demand signal
The same SKU may participate in several events:

Promotion planning is often disconnected from demand and inventory planning.
That creates two common problems:
Marketing creates demand that supply cannot serve.
Or:
Supply builds inventory for demand that promotion never generates.
A connected system aligns promotion funding with expected demand, elasticity, available inventory and expected incremental return.
The goal is not simply to reduce trade spend.
It is to put each dollar behind the event where it creates the most profitable demand.
What does the connected loop change?
The business impact becomes clearer when we compare two approaches for the same SKU and event.

The point is not the exact numbers.
The point is how value compounds when forecasting, supply, inventory allocation and promotions work as one system.
And this is still only one SKU.
For a CPG company managing hundreds or thousands of SKUs across retailers, DCs and promotional events, the impact scales quickly.
Why this matters for business leaders
Most companies do not have a data problem.
They have a decision synchronization problem.
Demand planning has one number.
Supply has another.
Sales has its own assumptions.
Marketing has its promotional calendar.
Inventory teams react after the imbalance appears.
Connected planning brings those decisions together.
Instead of asking planners to spend their week reconciling spreadsheets, the system continuously asks:
Has demand materially changed?
Are we going to stock out?
Is inventory sitting in the wrong channel?
Should we increase or reduce a promotion?
Do we need to transfer stock?
What decision best protects service level and margin?
The planner then focuses on the exceptions that actually need human judgment.
The real goal is not a better forecast
Forecast accuracy matters.
But it is not the final business outcome.
What ultimately matters is:

That is how we think about TrueGradient.
Not as another forecasting tool, but as an AI-native connected planning system where demand, supply, inventory and trade promotion continuously inform one another.
One SKU. Five channels. A decision made 100 days in advance - and a planning system that keeps improving the decision as reality unfolds.

Ankur Verma
CEO, TrueGradient
Ankur Verma is CEO at TrueGradient, the AI-Native Planning OS for Consumer Brands and retail. He is passionate about solving the toughest business challenges through the application of Machine Learning, Deep Learning & Reinforcement Learning. In the past, he has worked at Amazon and Walmart, solving optimization on a massive scale and dealing with billions of time series (Product/Location combinations). His academic papers can be accessed here .



