How Trace works
Trace is the record behind every Sirvo reason, and the ranking built on it. This page says what is in the record, how a space gets its list, how often the list is right, and what Trace does not do. Every number carries its date.
Record as of Oct 10, 2026 · build in use: Trace 45.9 (Astor line), Oct 20, 2026 · page updated Oct 21, 2026
The record
A store-level record of where brands actually open: built from public sources, cleaned and checked by code, and refreshed weekly for two years. Every reason Sirvo shows is a count from it.
What the words mean
- Brand with open US locations
- A brand with trading stores in the record. 2,789 today: 1,284 services, 588 shops, 508 food and beverage, 180 automotive, 229 other.
- In three or more states
- Trading in three or more states; DC counts. 1,441 of the 2,789. The ranking considers any brand with three or more stores.
- Location
- One trading store after duplicates are collapsed: one source's records at the same address within 15 m, and different sources' records within 50 m.
- Weekly run
- One pass over the sources. 98 since October 12, 2024; the last on October 10, 2026.
- Opened
- A store first seen with listing evidence. 35,242 since October 12, 2024.
- Closed
- A store missed on two runs at least 13 days apart.
- Reopened
- A closed store seen trading again.
- Lapsed
- A store unseen for 364 days.
How a space gets its list
Four steps, from the confirmed property to the Likely tenants list. The flyer is read before any of this; Trace starts from what you confirmed.
- 1
Candidates
Any brand with three or more trading stores is a candidate for every space.
- 2
Left out, with the reason
A brand is left out when it is already in the center, when it is not for this kind of space, when there is no size that fits, or when it is another use. Your answers can leave out more: brands that need a kitchen hood the space can't have, uses ruled out here, and categories under another tenant's exclusive. With the fallback formula, a brand already close by is left out too. The list says how many were left out and why.
- 3
Ranked by precedent
The rest are ranked by what they have done: how far their nearest stores are, who they open beside, what they opened nearby lately, and how far they reach. The ranking is trained on the record's own history of openings and measured again on every build.
- 4
Every reason is a count
Each reason on a row is a count from the record, dropped into a fixed sentence. A reason with a zero or missing count is left out, not written around. No sentence is written by a model.
Trace ranks by what brands have done, not what anyone has said. A high rank is strong precedent, not a signed lease.
The kinds of reason
- Nearby
- The distance to the brand's nearest store, and how many it has within 1 mile and within 10 miles. Where 150 or more stores sit within half a mile of the site, the rings are half a mile and 3 miles.
- Beside
- How many of an anchor's stores have this brand beside them, out of all the anchor's stores, and how many times chance that is. Anchors are brands with a store within 150 m of the site; with the fallback formula, the flyer's anchors and co-tenants come first. A Beside reason needs at least 5 pairs and a lift above 1.
- Openings
- Stores the brand opened within 25 miles in the past 26 weeks, with the most recent one. Where none opened that close, the stores it opened in the site's state.
- Size
- The brand's usual size range against the space, or its category's where the brand states none. A brand whose range fits none of the spaces is left out before ranking.
- Drive-through
- The share of the brand's stores with a drive-through, where at least half have one. A space's drive-through moves the list: brands with lanes rise where the space has one, and brands with lanes at nine in ten of their stores or more fall where it has none.
- Reach
- How many US stores the brand has, and in how many states.
- Coming soon
- How many of the brand's stores are marked coming soon within 25 miles.
- Mix
- The share of the stores within a mile of the site that are in the brand's group, such as food and beverage.
- Competition
- How many stores of other brands in the brand's category sit within a quarter mile.
- Department stores
- How many of the brand's stores sit beside a department store, out of all its stores.
- Density
- How many stores the brand's stores usually have within half a mile, against how many this site has.
- Growth
- How many stores the brand opened and how many it closed across the US in the past 26 weeks.
How often the list is right
Every build of the ranking is measured the same way, on real store openings, against two baselines. A copy trained without those openings is what gets measured; the build itself is then retrained on everything. The figures below are the build's own audit, read from the record on the date shown.
On 3,498 real US store openings (Apr 11 – Oct 10, 2026), the brand that actually opened was in Trace's top 25 for 45.9% of them. Ranking by store count alone gets 12.9%.
Trace 45.9, the build in use since Oct 20, 2026.
| Was the brand that opened in the top… | Trace | Store count alone | Fallback formula |
|---|---|---|---|
| Top 10 | 30.2% | 4.9% | 14.3% |
| Top 25 | 45.9% | 12.9% | 27.6% |
| Top 50 | 61.3% | 25.6% | 44.5% |
Same 3,498 openings, same window. The fallback formula is the rule-based ranking Sirvo uses when no trained build is ready.
How it is measured
- The question
- Of real US store openings in the window, how often was the brand that opened in Trace's top 25 for that site?
- The sample
- 3,498 openings by brands that take in-line and end-cap space, chosen by rule: observable for at least 40% of the window, not noisy, at most 40 per brand, duplicates removed; April 11 – October 10, 2026, all 50 states and DC.
- Held out
- Openings in the window were held out of the copy that was measured, so it never saw them while training.
- The baselines
- The same question answered by store count alone, and by the fallback formula.
- The build
- Trace 45.9 (Astor line), trained October 20, 2026 on the record as of October 10, 2026. Four gates, all passed.
- The name
- Every build is named by its own top-25 hit rate, inside a line named for a New York street — Astor first, then Bowery, Canal, and so on down the alphabet as the ranking is rebuilt from the ground up. A build never changes after it is named; a worse build gets a lower name.
- What it is not
- A prediction for any one space. It measures precedent on past openings.
- The method
- Questions: admin@sirvo.io.
The same test, by state
| State | Openings scored | Trace top 25 | Store count alone |
|---|---|---|---|
| California | 298 | 39.6% | 7.7% |
| Florida | 263 | 47.1% | 13.7% |
| New York | 166 | 39.8% | 16.3% |
| North Carolina | 155 | 58.1% | 12.9% |
| Texas | 423 | 43.7% | 15.1% |
New York State is where Sirvo starts.
The five states reported in the build's audit, as slices of the same 3,498 openings. Smaller samples are less certain: New York's figure rests on 166 openings.
Every build, measured
| Build | Line | Trained | Record as of | Openings scored | Top 10 | Top 25 | Top 50 | Store count, top 25 | Fallback formula, top 25 | Status |
|---|---|---|---|---|---|---|---|---|---|---|
| Trace 45.9 | Astor | Oct 20, 2026 | Oct 10, 2026 | 3,498 | 30.2% | 45.9% | 61.3% | 12.9% | 27.6% | In use |
| Trace 44.9 | Astor | Oct 19, 2026 | Oct 10, 2026 | 3,748 | 29.1% | 44.9% | 60.1% | 12.0% | 27.5% | Superseded |
Every build the ranking ships is measured the same way and added here, including the ones that do worse. The table starts with the first build that carried its own audit. Why the sample changed between builds is for the method memo.
What Trace does not do
It does not read the flyer.
Sirvo reads the flyer and asks you to confirm the draft. Trace starts from the confirmed property.
It does not promise a signature.
A high rank is strong precedent, not a signed lease. Likely tenants are candidates to approach, not commitments.
It does not cover everything yet.
In the record today: 2,789 brands with open US locations across all 50 states and DC, 1,441 of them in three or more states; the ranking considers any brand with three or more stores. Not yet: local tenants, industrial and office.
It does not write around a gap.
When the record has no count for a reason, the reason is left out. Check every list against what you know.
It is measured on the past.
The hit rate measures precedent on past openings. It is not a prediction for any one space, and a new build gets a new figure; the table above keeps every one.
It does not name its sources here.
The record is built from public sources, cleaned and checked by code. What matters on a row is the count, its denominator and its date, and those are on every row.