Help

What exactly is a read?

One address and one concept, scored against Chicago's restaurant record back to 2002. You get the score, the drivers behind it, and what the address itself has already proven: who traded there, for how long, and how it ended.

What if nothing has ever traded at my address?

The read still works. When the address has no licence history of its own it falls back to the block, and the report says which it used rather than presenting a block read as an address read. An address with no history is a real finding, not a failure. It usually means the space was not previously food.

Can I send the report to a landlord, a lender or a partner?

Yes, that is what it is built for. Reports export to PDF and can be shared by link, and the document carries its own context: the address, the concept, the data vintage, the model version that produced it, and a glossary of every number in it. A forwarded report should not need you in the room to be read correctly.

What is the 2ndSight score?

It is an estimated probability that a restaurant concept at a given Chicago address is still open 4 years after opening. It is not a grade, a rating, or a rank out of 100: a score of 62 means roughly 62 in 100 comparable concepts were still trading after 4 years.

Is a good score a high number?

An average concept in an average location scores 57. That is the line to read yours against. Above it the record is in your favour, below it against.

How is the score calculated?

Every score starts at 57, the model's prediction for an average concept in an average location. Five drivers then move it up or down: the concept's own track record citywide, the operator's prior openings, how often restaurants in the zip turn over, how crowded the zip is with the same cuisine, and how that cuisine specifically has fared in the zip. The adjustments sum to the score exactly.

Does the score change from one address to the next?

Not within the same zip code. The location half of the score is measured at zip level, so the same concept and the same operator score identically at two addresses in one zip, and the ranking of concepts is the same for both. That is deliberate: finer location data (income, foot traffic, walkability, transit distance, competitor counts by radius) was tested and did not improve accuracy on restaurants the model had never seen, so claiming block-level precision would be claiming accuracy we measured and did not find. What the address does change is the rest of the read: that property's own licence and closure history, the comparable restaurants shown, and the write-up. Your own track record moves the number more than the block does.

Which market does this cover?

Chicago today. The model is trained on Chicago's own licence and inspection record, and a market is only opened once there is enough history to train it honestly. If you enter an address outside coverage, we say so and take your city instead of guessing.

How current is the data?

Licence and inspection data refreshes daily. Google verification and rent bands refresh monthly. Both timestamps are shown in the product, so neither is ever mistaken for the other.

Why don't you show how many restaurants closed recently?

Because we cannot yet know. A closure becomes visible in the licence record roughly three years after it happens: a licence has to lapse past its renewal window before the city's data reflects it. Any count of recent closures would therefore be a large undercount presented as a fact, so we publish none. It is also why survival figures stop short of the present rather than running to today.

Why isn't the pharmacy on my block counted as a competitor?

Chicago's Retail Food Establishment licence covers far more than restaurants. Groceries, corner shops, pharmacies, hotels and school cafeterias all hold it, and they are roughly 45% of all holders. Counting them would inflate every competitor and density figure in the product, so they are filtered out everywhere a business is counted. If a figure elsewhere counts licence holders, it is counting a different and much larger population.

What does the Site Report add?

The read tells you whether the concept fits. The Site Report is what you negotiate or underwrite with: revenue and rent bands, an operating model with a payback estimate, the full ranked concept slate for that zip, the competitive field, and a diligence checklist. It is free at the moment, like everything else here.

Do I have to pay for any of this?

No. Everything is free right now: every read, every Site Report, every Chicago dashboard, and the watch digest. We are still working out which part of this is worth paying for, and we will say so on the pricing page before that changes.

What if the model changes after I take a report?

A report is a snapshot, stamped with the exact model version that produced it, and it never changes underneath you. When the model retrains, existing reports stand as issued and re-running the address produces a new report under the new version.

Do you offer team or API access?

There are no plans or seats at the moment, so a whole desk can use this today. API access into an underwriting or site-selection workflow is a conversation rather than a product: tell us what you need it to do.

Can I get a refund?

There is nothing to pay for at the moment, so there is nothing to refund. The refunds page sets out the terms that would apply if that changes.

Can I use a score to decline a loan or a tenant?

No. A score is a statistical estimate of how comparable sites have performed, not a decision about any person or business. It is an input among others, and using it as a basis for an adverse decision about an applicant is outside the terms.

What can the score not see?

Your menu, your operating skill, your lease terms, your capital, and your team. It describes how comparable concepts at comparable locations have historically fared, which is a statement about the record and not a prediction about your business. Most of what decides whether a restaurant works is not in any public dataset.

Are your terms and privacy policy reviewed by a lawyer?

No, and we say so rather than implying otherwise. They were written in-house and are published unreviewed by decision, with the open questions recorded rather than hidden. That is a real limitation and you should weigh it accordingly.

How do I get help, and how fast will you reply?

Email hello@use2ndsight.com. We answer within one business day. It is a person, not a queue: for a one-person product sold to operators and brokers, response speed is the substitute for company size.

How do I report a mistake in the data?

Tell us the address and what is wrong with it. Public records carry errors, our derivation of them can carry more, and a correction from someone who knows the block is worth more than another pass over the same file. Corrections are tracked, not just acknowledged.

How do I stop receiving email?

Every email we send carries an unsubscribe link and honours the native unsubscribe control in Gmail and Outlook. Report emails you asked for are transactional and separate from the lifecycle sequence, so opting out of one does not silently cancel the other.

Email hello@use2ndsight.com. We answer within one business day. Tell us the address you were looking at if the question is about a specific read — it is almost always faster than describing it.