enzwa lab · bearing · weighted trade-off model

Where should we put it? And what would change that answer?

This page runs the six sector briefs Bearing ships with, and asks the six questions a recommendation has to survive six months later. Every ranking on it is computed by re-running the model itself.

A tool for choosing where to put something, and for keeping the reasoning.

The problem

Choosing a city for a new office, plant or team means weighing a dozen things that pull against each other: wages, talent, rent, tax, flights, risk. Somebody decides how much each one counts, usually in their head, and six months later nobody can reconstruct it.

What it does

Bearing scores every candidate city on all of those measures at once, lets you set how much each matters for this particular job, and shows the shortlist reordering as you change it. The weights stay on the page with the answer.

An architectural massing model: plain basswood and grey-card blocks of varying heights on a neutral board, lit from a low angle so each throws a long shadow. — candidate places, reduced to comparable objects

question 01

Where should we build?

Run the corporate and shared-services brief and the model returns Bangalore first, at 70.29 out of 100, with Toronto second at 69.29. These are the top eight of 33.

#	city	country	composite
1	Bangalore	India	70.29
2	Toronto	Canada	69.29
3	Hyderabad	India	68.55
4	Austin	United States	68.32
5	Warsaw	Poland	67.64
6	Dubai	UAE	67.59
7	Krakow	Poland	67.36
8	Bucharest	Romania	67.15

question 02

Is that the answer, or the answer to this brief?

The same 33 cities, the same indicators, the same day. Only the weights move. Six shipped briefs return four different winners, and no city takes more than two of them. The four numbers in the second column are the weights on talent, operating environment, risk and cost.

brief	weights	first	score	second
Corporate / GBS	40 / 15 / 13 / 32	Bangalore	70.29	Toronto
Semiconductor / Adv. Mfg	24 / 30 / 20 / 26	Austin	72.80	Toronto
Tech / R&D / AI	44 / 32 / 12 / 12	Austin	73.05	Singapore
Regional HQ	26 / 34 / 26 / 14	Singapore	74.81	Toronto
General manufacturing	16 / 28 / 18 / 38	Dubai	71.93	Toronto
Logistics / distribution	12 / 40 / 13 / 35	Dubai	73.40	Toronto

Several separate massing studies lined up along a workbench, each a different arrangement of blocks on its own base, receding into soft focus. — six briefs, laid out side by side

question 03

How far ahead is it?

On the semiconductor brief Austin leads Toronto by 0.40 points out of 100. Two of the six briefs are decided by less than a single point, which is well inside the tolerance of the underlying benchmark data.

gap between first and second, in composite points · swipe for the full scale

question 04

How much would the weights have to move?

Hold the corporate brief and move one weight at a time, from 0 to 80, leaving the other three where they are. Teal is the range where Bangalore still comes first. The vertical mark is where the brief actually sets that weight.

one macro weight swept at a time, corporate brief held · the number on the right is the distance to the nearest edge

The recommendation survives a 21-point error in the talent weight and a 3-point error in the risk weight. That asymmetry is the useful part: it says the argument worth having is about risk and cost, and that the talent weight can be wrong by a wide margin without changing what the model recommends.

question 05

Which city is never wrong?

Toronto comes second under five of the six briefs and third under the sixth. It wins none of them. A weighted model rewards the sharpest fit to the weights it was given, so the most balanced candidate is everybody’s second answer and nobody’s first.

rank under each brief · teal is first place

That is worth seeing before a shortlist is cut to three, because the city that survives every brief is the one a ranking sorted by first place will quietly drop.

question 06

So what is the deliverable?

The name at the top is the smallest part of it. What a decision needs six months later is the tree the score came from, the weights somebody chose and signed, and the distance to the nearest point where the answer changes. Bearing keeps all three on the screen, so the shortlist can be questioned rather than defended.

Bearing describes its own benchmark values as representative composites calibrated to published 2024 to 2026 sources, and says to validate them against live, brief-specific data before committing capital. That caveat travels with every number on this page.

question 07

Move the weights yourself.

Bearing runs in the browser. Answer five questions about a brief, and the engine scores all 33 cities down to the site-level factors and shows its working at every level of the tree.

Open the model →

A single massing model alone on a large bare board, one long shadow reaching across the empty surface. — one answer, and the room it was chosen from