Methodology

A decision file should show how it knows.

Tensor Advisory makes the research chain visible: the question, the sources, the assumptions, the recommendation and the conditions still open.

The chain

Five steps from question to next test.

01

Frame the decision

Set the product, corridor, market, decision owner and the condition that would change the answer.

02

Assemble evidence

Use the most relevant official datasets, regulatory portals, company material and industry sources for the question.

03

Cross-check

Record assumptions, compare signals, label proxies and separate what a source establishes from what it suggests.

04

Write the route

Turn the evidence into a recommendation, conditions, open questions and a next test the team can own.

05

Review and carry forward

Challenge the reading in a debrief, then define what a custom scope or operating cadence should answer next.

The evidence standard

A source is useful only when its boundary is understood.

Official trade and regulatory sources can answer specific questions about indicators, tariffs, rules, formalities or registration. They do not, by themselves, establish demand, partner suitability or a commercial outcome.

Where a figure is a proxy, estimate or forecast, the report labels it. Where the evidence is thin, the recommendation narrows the wedge or states the next check instead of filling the gap with false precision.

Open the primary source directory

A real example

A trade benefit still has to survive the paperwork.

“A zero tariff with no origin proof is just the old tariff with extra steps.”

EU-India FTA impact report, page 13. The report pairs this conclusion with a product-level origin checklist and a dated status update.

See the report context

Method in practice

Ask for the evidence before you ask for the answer.

Start with an existing report or send the question that needs a custom evidence boundary.

Bob, your AI guide