Criteria-First Buying Agent
Build a personalized market comparison around the buyer's own criteria.
- Difficulty
- Moderate
- Time to result
- ~days to results
- Steps
- 5
- Confidence
- 97%
The Criteria-First Buying Agent reverses seller-led marketing by building a temporary recommendation system for one buyer. The buyer first defines the capabilities, constraints, price sensitivity, and evidence that matter. An agent gathers candidate products and creates a sortable comparison or lightweight application organized around those criteria rather than each vendor's preferred messaging. Published prices, reviews, documentation, and other sources become structured evidence. For shortlisted products, the agent can perform trials, execute defined test cases, and feed the observed results back into the comparison. The output is a personalized buying surface that lets the user rank options using their own priorities. Its mechanism is criteria definition, evidence aggregation, direct testing, and evidence-based reranking.
Origin
Extracted from Marketing Against The Grain
Core principles
- 01Start with the buyer's criteria, not vendor positioning.
- 02Aggregate options into a buyer-controlled comparison surface.
- 03Rank products using evidence relevant to the intended use.
- 04Test products when published information is insufficient.
- 05Keep the final choice tied to explicit priorities.
How to run it
- 1
Define purchase criteria
List the required capabilities, budget constraints, preferences, and deal breakers before examining vendors.
Pro tip Separate mandatory requirements from preferences.
Watch out Criteria chosen after seeing products are more vulnerable to vendor framing.
- 2
Collect candidate options
Gather relevant vendors or suppliers and collect comparable information about each one.
Pro tip Record source URLs and retrieval dates for important claims.
Watch out Do not assume a missing website claim proves a missing capability.
- 3
Build the comparison surface
Structure the products in a sortable table or application based on the buyer's criteria.
Pro tip Allow filtering by mandatory requirements before applying softer rankings.
Watch out A polished interface does not compensate for weak or inconsistent evidence.
- 4
Run product trials
Where feasible, have the agent sign in and test whether shortlisted products complete defined tasks.
Pro tip Use the same test script and sample data for every vendor.
Watch out Do not authorize purchases, sensitive data entry, or irreversible actions without human approval.
- 5
Ingest results and decide
Add trial outcomes to the comparison, rerank the candidates, and document why the leading option best fits the criteria.
Pro tip Preserve failures and uncertainty instead of reducing every result to a single score.
Watch out Weights should not conceal a failed mandatory requirement.
In the wild
A buyer who prioritizes price asks an agent to gather rubber mat suppliers and build a custom aggregator that can be sorted around that criterion. The same workflow could incorporate reviews, dimensions, delivery terms, and sample-test results before ranking suppliers.
→ The buyer gets a purpose-built comparison rather than manually browsing each supplier's website.
Common mistakes
Starting with vendor categories
Adopting the seller's comparison structure can hide the factors that matter most to the buyer.
Treating claims as tests
Marketing pages and reviews are useful evidence, but they do not prove that a product completes the buyer's real workflow.
Is it for you?
Best for
Buyers comparing software, suppliers, or products with several explicit and verifiable decision criteria.
Not ideal for
Purchases driven primarily by subjective experience, confidential negotiations, or criteria the agent cannot observe or test.
From the transcript
“I care about A, B, C D and E. Now build me an aggregator app that aggregates them all together and allow me to like…”
“Like sign in, use the software, tell me does it do A B and C and you can actually ingest that back into the app.”
“This app now allows me to sort through the suppliers that I'm trying to find in a very customized way based upon what my core…”
From the episode
This FREE AI Agent Does a Team’s Work in 35 Min (Manus AI)