The Loci System — Genius Loci
The Loci System · The trust page behind The Index
How the machine works, in plain terms.
The Index is the short list. This page is how that list gets made — and why you can check it. Demand first, sourcing second, scoring third, gates that can stop anything, and a human Review Desk before anything public ships. No step is skipped, and no step is hidden behind a slogan.
Where this stands today: the method below is defined and the scoring is being calibrated in shadow — checked by hand before anything is published. The Index lists no entries yet, and this page claims no products, scores, or services beyond what is live today. Start from Start if you arrived here first.
One pipeline. Five places it can stop.
Pick a stage to see what actually happens there — what goes in, what gets checked, and what ends the run. The detail matters more than the diagram.
Demand — start with what people want
It starts with what people are actually looking for — not with inventory that needs moving. Demand is read from searches people make, questions they ask, and what they buy, gathered into a brief with the evidence attached. A product with no real demand behind it does not advance, no matter how good the margin looks.
What goes in
- Search and question signals — what people type and ask
- Sales and traffic signals from the store itself
- Market and scout signals, checked against each other
What gets checked
- Is the demand real, or a product in search of a problem?
- Is there evidence, or only a hunch?
- Does the signal hold across more than one source?
Honest limit: early on, many candidates read as “demand: unproven.” That label stays until there is evidence — it is a valid state, not a number rounded up to a yes.
The six signals behind the score
The same six questions The Index publishes. The score is a summary of them — not a substitute for them — and every published entry will show its working.
Real demand
Are people actively searching for, asking about, or buying this — or is it a product in search of a problem?
Price against the alternatives
What does the same job cost elsewhere? The price has to hold up against the best comparable option, not just a weak one.
Quality you can point to
Materials, construction, specs, reviews with substance, return and failure patterns — evidence, not adjectives from a supplier listing.
Competitive position
If a better-known option does the same thing better or cheaper, that gets said plainly — including when it means dropping the product.
Value for both sides
It has to be a fair deal for the buyer and viable for the store. A product that only looks good because the margin is good does not pass.
Knowable, not guessed
The score is built only from signals that can actually be checked. Where something cannot be known yet, the entry says so instead of filling the gap.
Three bands. One honest caveat.
Index Scores run 0–100. The bands are the working shape of the system while it is calibrated against hand-checked spot checks of real catalog products — so the numbers mean something before they are ever shown next to a product.
Worth a look. Clears the basic bar — real demand, defensible price, no quality red flags.
Strong. Beats the obvious alternatives on the signals that matter, not just one of them.
Best of the best. The rare tier the full package is reserved for. Expect very few products here.
The caveat: these bands are in calibration, not set in stone. If the spot checks show a band is too easy or too strict, the band moves — and each entry, when entries publish, will show the score it earned under the version in force at the time. Deeper scoring detail lives on The Index.
Gates, not vibes: a fail anywhere drops it
Scoring summarises. Gates decide. Each gate below is a pass/fail check with a drop branch — a strong score on one signal does not rescue a failure on another.
No demand, no run
No evidence of demand means the candidate stops here. “Demand: unproven” is recorded honestly and the product does not get sourced on hope.
On fail: dropped from contention. It can re-enter only when a real signal appears.
The right product, actually sourceable
Sourcing checks the Genius Loci catalog first, then Shopify Collective. Supplier reliability and a defensible price are part of the check — a near-match is not a pass.
On fail: dropped. If the right product cannot be sourced well, the slot stays empty.
Price, quality, competition, value — all of them
Price against the best comparable, quality from evidence you can point to, competitive position stated plainly, and value that works for buyer and store. All four must hold.
On fail: dropped, with the reason recorded. If a better or cheaper option exists, the honest move is to say so.
Nothing public without a human yes
Even a candidate that clears every automated check stops at the Review Desk. Agents propose; a human disposes. See the next section for how that gate works.
On hold or reject: nothing publishes. A hold is a decision, not a delay tactic.
The human gate at the end of the machine
The system is built so that speed never outruns judgement. Agents do the searching, matching, checking, and drafting. They do not publish.
Approve · Hold · Reject
Every public output — a post, a page, a product package — arrives at the Review Desk as a proposal with its evidence attached. The operator reviews it there, in one place, before anything is armed to publish.
-
Approve
It ships. Only an approved item moves to publishing — social through the Review Desk approval itself, commerce through a release decision and the publishing step that follows it.
-
Hold
It waits. Held items stay held. A switch turned off holds future work instead of posting it, and an unreadable state posts nothing.
-
Reject
It stops. Rejected work does not go out, and the reason feeds back so the same mistake is not proposed twice.
-
—Agents never publish directlyNot to social, not to the store. Every lane ends at a human decision first.
-
—One queue in, one gate outWork enters through a single dispatch queue as structured tickets — objective, context, constraints, acceptance criteria — and leaves only through the Desk.
-
—Feedback closes the loopSales, traffic, search, and returns data flow back into scoring. A listed product that later fails a check comes off — with the change visible on The Index.
-
—The person behind it is namedThe operator who holds the Desk is the founder — see the Founder Bio. The agents who do the work are documented at the Agent Hub.
Transparent about the method. Careful about the rest.
A trust page earns trust by saying where its own edges are.
The method, in full
- The pipeline: demand → sourcing → scoring → gates → Review Desk
- The six signals, the three bands, and the pass/fail logic
- The honest states — “demand: unproven,” “calibrating,” “not yet listed”
- Where each franchise hub fits, and what job each page does
What stays behind the Desk
- Internal thresholds still being calibrated — a number published before it is settled would be a guess dressed as a fact
- Supplier, cost, and routing internals that are working details, not methodology
- Any product, score, or service that has not actually shipped — this page describes the system as it stands today, nothing further
The six hubs are the machine, running in public
Each franchise is one part of this system, performed daily. Start at Start, or go straight to the hub that interests you.
Real systems, built in public, one flagship piece a day — the system you just read about, being built.
Its job: the services door
Visit The Build → Hub · The agents Agents at WorkThe agent roster doing real work inside the system — the demand hunters, sourcers, and scorers in action.
Its job: Stack Sheet capture
Visit Agents at Work → Hub · The judgement The BreakdownProducts and claims taken apart in detail — the test stage, done in the open.
Its job: Scorecard capture
Visit The Breakdown → Hub · The evidence The ProofEvidence, shown daily — and the future home of Index Drop announcements when entries publish.
Its job: Index Drop signup
Visit The Proof → Hub · The comparisons The StackHead-to-head comparisons, decided on evidence — the price and competition signals, side by side.
Its job: Scorecard capture
Visit The Stack → Hub · The winners The SelectCurated picks from the catalog — the closest thing to The Index that exists today, without the full scoring package.
Its job: the shelf — the Index door
Visit The Select →