P
AI for Retail Buyer

Custom GPT: Build a Category Intelligence Assistant

ChatGPT

For Retail Buyers

Tools: ChatGPT {{tool:ChatGPT.plan}} | Time to build: 1-2 hours | Difficulty: Intermediate-Advanced Prerequisites: Comfortable using ChatGPT for negotiation briefs and sales commentary. see Level 3 guides

What This Builds

A Custom GPT that already knows your category. your vendors, price architecture, key metrics, competitive set, and buying philosophy. Every conversation starts fully briefed. Instead of re-explaining your category each time you open ChatGPT, you open your Category Intelligence Assistant and immediately get relevant, specific outputs for negotiation briefs, trend analysis, vendor evaluations, and business reviews.

Prerequisites

  • {{tool:ChatGPT.plan}} subscription ({{tool:ChatGPT.price}}). Custom GPTs require a paid plan
  • Comfortable using ChatGPT for writing and research tasks (Level 3)
  • 30-60 minutes to document your category context (the most valuable investment)
  • Optional: vendor list, OTB summary, and seasonal plan as uploadable documents

The Concept

A Custom GPT is like having a junior buyer assistant who has read all your category files, knows your vendor relationships, and has been briefed on your strategy. except it's available 24/7, never forgets, and can produce polished output in seconds. The key is writing a detailed system prompt (your "briefing document") that gives it genuine category context.

Build It Step by Step

Part 1: Document your category context (do this first)

Before building the GPT, write out your category context in a text document. Include:

Category basics:

  • Category name and subcategories
  • Retailer type and positioning
  • Annual sales volume and budget range
  • Your key performance metrics (sell-through target, margin target, turn target)

Vendor landscape:

  • Top 5-10 vendors by volume
  • Their key characteristics (price points, performance, relationship quality)
  • Which vendors you're growing vs. managing vs. exiting

Competitive set:

  • 3-5 primary competitors you track
  • How you position against them (price, style, quality, exclusivity)

Buying philosophy:

  • What makes a product right for your category
  • Your customer profile
  • Trend approach (trend-forward vs. commercial vs. basics-heavy)

Save this as a document. you'll paste it into the system instructions and/or upload it.

Part 2: Open the Custom GPT builder

  1. Log into ChatGPT at {{tool:ChatGPT.url}} with your {{tool:ChatGPT.plan}} account
  2. Click Explore GPTs in the left sidebar → Create (top right)
  3. Click the Configure tab

Part 3: Write your system instructions

In the Instructions field, paste a comprehensive briefing using this structure:

Copy and paste this
You are a Category Intelligence Assistant for a retail buyer. Here is your complete category briefing:

CATEGORY CONTEXT:
[Paste your category basics from Part 1]

VENDOR LANDSCAPE:
[Paste your vendor summary]

COMPETITIVE SET:
[Paste your competitive positioning]

BUYING PHILOSOPHY:
[Paste your approach]

WHAT YOU DO:
When asked to draft negotiation briefs: Use our actual vendor performance data and competitive alternatives to build leverage-based briefs. Always include specific proposed terms, not vague ranges.

When asked to write sales commentary: Frame performance in terms of trend impact, inventory position, and forward action. Be concise and confident.

When asked to evaluate new vendors: Assess against our category standards (margin, sell-through potential, customer alignment, fill rate reliability).

When asked about trends: Connect trend observations directly to assortment implications — what to buy more of, what to reduce, what gaps to fill.

ALWAYS:
- Reference our specific category context when generating outputs
- Give specific recommendations, not generic frameworks
- Ask for additional data if it would significantly improve the output

NEVER:
- Give generic retail advice that could apply to any category
- Assume data you haven't been given — ask for it

Part 4: Upload category documents (optional but powerful)

In the Knowledge section, upload:

  • Your current vendor list with key metrics
  • Your current seasonal OTB plan (de-identify if needed)
  • Any category strategy documents you've written

The GPT will reference these when generating outputs.

Part 5: Set conversation starters

Add these so anyone opening the GPT knows how to use it:

  • "Build a negotiation brief for vendor [name]..."
  • "Write weekly sales commentary for [week]..."
  • "Evaluate this new vendor pitch: [describe]..."
  • "I need a trend narrative for [season/category]..."

Part 6: Test with real scenarios

Test 3-4 real scenarios before sharing with colleagues:

  • Paste a vendor performance summary and ask for a negotiation brief
  • Give weekly sales data and ask for commentary
  • Describe a new vendor and ask for an evaluation framework
  • Ask for a trend narrative based on observations you paste

Refine the instructions based on what's missing or generic in the outputs.

Part 7: Share with your team

Click Save → set visibility to Only people with a link → share with your assistant buyer or manager. They'll need a {{tool:ChatGPT.plan}} account.

Real Example: Negotiation Brief in 90 Seconds

Setup: Category Intelligence Assistant knows the full Women's Accessories category context, including that Vendor X does $620K annually with 66% sell-through (vs. 72% category benchmark).

Input: "Build a negotiation brief for my Vendor X meeting Thursday. I want 8% cost reduction and net 60 payment terms. I have quotes from two competitors at better cost."

Output: A 500-word negotiation brief that references Vendor X's specific underperformance against category benchmark, frames the competitive quotes as genuine leverage (not a bluff), proposes opening at 12% reduction, accepts at 8%, and includes specific language: "We've received competitive proposals from two vendors at [X]% better cost. To justify maintaining your current OTB allocation, we need to see cost alignment..."

Time saved: 2 hours of research and writing → 90 seconds + 10 minutes of editing.

What to Do When It Breaks

  • Outputs feel generic → Your system instructions lack specificity. add more concrete category details (actual vendor names, real metric targets)
  • GPT invents vendor details you didn't provide → Add to instructions: "Never state specific vendor metrics unless I provide them in the conversation"
  • Knowledge files aren't being referenced → Ask explicitly: "Based on the vendor list I uploaded, which vendors are underperforming on sell-through?"
  • Colleagues can't access it → Confirm they have paid ChatGPT accounts; Custom GPTs don't work on free tier

Variations

  • Simpler version: Skip the knowledge file upload. the system instructions alone give 80% of the value
  • Extended version: Create separate Custom GPTs for different functions (one for negotiation prep, one for trend research, one for assistant buyer coaching)

What to Do Next

  • This week: Build the GPT and run 3 real tasks through it, comparing to your usual workflow
  • This month: Add your vendor list and OTB plan as knowledge files; refine instructions based on output quality
  • Advanced: Create a shared version for your buying team. it becomes institutional knowledge everyone can access

Advanced guide for retail buyer professionals. Requires {{tool:ChatGPT.plan}} subscription at {{tool:ChatGPT.price}}.