The Stock That Won't Sell: RT062 Dead Stock Action AI Agent

Every kirana store, supermarket, and multi-store chain has it: stock that just sits there. Money frozen on a shelf. Capital you cannot reinvest. Shrink risk climbing every week. And nobody has time to decide — should we markdown, bundle with a fast mover, transfer to a branch that actually sells it, or return to the vendor before the window closes?

The decision isn't hard because the math is mysterious. It's hard because the data is scattered, the category manager is juggling forty priorities, and the default action is inertia. RT062 — the Dead Stock Action Assistant from DestinPQ — compresses that decision into a scored analysis, a single recommended lever, a recovery estimate in rupees, and a reference ID that survives the approval chain.

The use case: slow SKUs, fast decisions

RT062 serves retail operations teams, category managers, franchise inventory coordinators, and multi-store chains where dead stock accumulates silently until quarterly reviews. The trigger is a SKU that meets your dead stock criteria: typically ninety or more days on shelf, velocity below category threshold, and meaningful quantity tied up at a single branch or DC.

The outcome is structured, not automatic. RT062 scores the SKU, models margin impact for each action lever, recommends one primary path with a recovery estimate, and routes the suggestion to an approval queue. A regional ops manager approves or overrides. Nothing executes without human sign-off — but the analysis that used to take forty-five minutes in Excel happens in one conversation. Every submission logs Ref #RT062-XXXX for audit.

Who this is for

• Supermarkets and hypermarkets with thousands of SKUs and weekly freshness pressure
• Multi-store apparel and footwear where size/colour fragmentation creates branch-level dead stock
• Pharmacy and FMCG with expiry-driven return windows
• Franchise retail networks needing consistent dead-stock playbooks across territories
• Electronics and appliances where model-year transitions strand previous-gen inventory

What's broken today — spreadsheets, gut feel, and write-offs

Most retailers discover dead stock monthly — sometimes quarterly. A report lists SKUs sorted by days on hand. Someone highlights rows in yellow. A meeting happens. Actions get assigned. Three weeks later half the list is unchanged because daily urgencies won.

Markdown decisions often default to "30% off everything slow" because tiered logic lives in a category manager's head, not in a system. Bundle opportunities get missed because cross-SKU velocity data requires a separate pivot table. Transfers fail because nobody checks which branch sold twelve units last week while this one holds eighty. Returns miss vendor windows because expiry dates aren't tied to action urgency.

The cost isn't just the write-off line on the P&L. It's working capital trapped in the wrong place, shelf space that could carry winners, and staff time spent on manual analysis that RT062 delivers on demand through the same chat widget your team already uses for other DestinPQ agents.

How RT062 works: four levers, one recommendation

RT062 applies a consistent decision framework. Identify — dead stock criteria from your config (days on shelf, velocity percentile, minimum quantity). Analyse — margin, holding cost, expiry proximity, branch-level sell-through. Recommend — score each lever and pick the highest expected recovery. Action — queue for manager approval with full audit trail.

The four levers are not equal for every SKU. Markdown is fast but destroys margin. Bundle preserves perceived value by attaching slow movers to heroes. Transfer redeploys stock to demand without discounting. Return recovers vendor credit when windows and agreements allow. RT062 models each path and explains why one wins.

Action lever comparison (SKU AX-4412 example)

Lever scores for AX-4412 (120 units, 95 days)

Deep dive: each action lever

Demo walkthrough: Meridian Retail on rt062.html

The live demo at agent-demo.destinpq.com/rt062.html mirrors production behaviour on a sample retail site. Open the bottom-left chat widget. Send the analysis prompt with SKU, quantity, and days on shelf. RT062 validates inputs — SKU format, positive quantity, plausible days value — and returns the scored analysis with recommendation and quick replies.

Sample exchange:

• You: Analyse dead stock SKU AX-4412, 120 units, 95 days on shelf
• RT062: Velocity score 12/100 · Category: Home Care · Holding cost est. ₹2,840/month · Primary recommendation: Bundle with BX-2201 · Est. recovery ₹18,400 · Ref #RT062-20260705-094712
• Quick replies: Analyse another SKU · Show transfer options · Send to approval queue

The demo uses the same luxury-site plus widget pattern as other agents in the DestinPQ demo hub. WhatsApp handoff works on production deployments — franchise managers who live on mobile get identical analysis without opening the ops portal.

SKU analysis: what RT062 evaluates

When you submit a dead stock query, RT062 pulls from your connected SKU master and store-level inventory — API sync or scheduled CSV for pilots. The analysis layer computes:

• Velocity score — units sold per week vs category median over trailing 90 days
• Days on shelf — time since last meaningful sale or receipt, with ageing buckets
• Quantity at risk — units held × cost price = capital trapped
• Margin structure — current GP%, impact of each lever on net margin
• Expiry proximity — for dated goods, days until unsellable
• Branch dispersion — same SKU performance at other locations in the network
• Bundle affinity — co-purchase patterns for attach recommendations
• Vendor return eligibility — agreement terms, restocking fees, window status

Junk input is rejected. Empty SKU, zero quantity, or negative days do not produce a fake analysis — the agent asks for valid data, consistent with DestinPQ production validation across the agent catalogue.

Approval queue: recommendations, not auto-execution

RT062 deliberately stops at recommendation plus approval routing. Category managers and regional ops leads retain control. When an analysis completes, the agent can push to the approval queue with the full scorecard: recommended lever, alternatives ranked, recovery estimates, and reference ID.

Approvers see override options — accept bundle, switch to transfer, force markdown for speed. Each decision logs against the original reference so you can measure recommendation acceptance rate and recovery realised vs estimated. After ninety days, that data tells you whether your markdown tiers are too aggressive or whether transfer logistics cost is eating projected gains.

Real scenario: apparel chain with size fragmentation

A regional apparel chain holds SKU DX-9920 — size XL, colour navy — at Store #7: 84 units, 112 days on shelf, zero sales in thirty days. Store #3 sold nine XL navy units in the same period. RT062 scores Transfer at 91, Bundle at 54 (no strong affinity hero in category), Markdown at 38, Return at 0 (no vendor return on fashion).

Recommendation: transfer 60 units to Store #3, hold 24 for local markdown event. Est. recovery ₹1,12,000 vs ₹31,000 on 40% local markdown. Regional manager approves via queue. Ref #RT062-20260702-141033. Logistics books transfer. Shelf space at Store #7 freed for new season intake. Without RT062, the SKU sits another quarter until write-off.

How to implement RT062 on DestinPQ

Subscribe to RT062 on agents.destinpq.com, connect inventory data, configure category rules, and embed on your ops portal using the rt062.html template as reference.

1. Subscribe to RT062 — provision API token from the agent catalogue.
2. Connect SKU master + store inventory — REST API preferred; CSV sync for pilot.
3. Define dead stock criteria — default 90+ days and velocity below 20th percentile; adjust by category.
4. Configure markdown tiers and transfer rules — category-specific bands and inter-branch cost tables.
5. Embed widget on inventory ops portal; optional WhatsApp for field managers.
6. Route approval queue — regional ops manager or category lead; SLA for pending actions.
7. Test with known dead SKU — validate score, recommendation, ref ID, approval flow.

<script src="https://cdn.destinpq.com/agent.js"
  data-agent-id="RT062"
  data-token="YOUR_API_TOKEN"
  defer></script>

Start with one category pilot — home care or personal care works well because bundle affinity is strong and expiry pressure creates urgency. Expand to apparel and electronics once transfer rules and branch sell-through feeds are validated.

Recovery economics: why one lever beats the others

Retail finance teams often measure dead stock by write-off value alone. RT062 reframes the question: what is the expected recovery for each action, net of execution cost? A 30% markdown on 120 units might clear shelf fast but leave ₹18,000 on the table compared to a bundle that moves the same units at 12% effective discount while lifting hero SKU basket size.

Transfer looks free but carries logistics cost and receiving labour. Return yields vendor credit but may incur restocking fees and requires documentation. RT062 surfaces these trade-offs in rupees so approvers decide with numbers, not habit. Over a fiscal year, shifting even twenty percent of dead stock decisions from default markdown to optimal lever can recover lakhs in margin — the reference log proves it.

Metrics to track after deployment

Join RT062 reference IDs to your inventory and finance systems and track: dead SKUs analysed per week, recommendation acceptance rate, estimated vs realised recovery, time from identify to action, write-off ₹ trend by category, and transfer success rate (units sold at receiving branch within 30 days). Category managers who review a weekly RT062 summary report close the loop faster than those waiting for monthly ERP extracts.

Frequently asked questions

Part of the DestinPQ 7 Days · 7 Agents series — practical AI agents for restaurant, retail, hospitality, and field operations. Agent ID: RT062 · Dead Stock Action Assistant.