# How to Track YC Startups Before Demo Day: A Practical Guide

> Learn how to track YC startups before Demo Day: select a batch, record profiles, verify claims, compare dated signals, and schedule updates without treating missing data as zero.

![How to Track YC Startups Before Demo Day: A Practical Guide](https://yc-dossier.nicheflash.com/media/articles/2731-zkoollqtfrp5-16x9.jpg)

- Source: https://yc-dossier.nicheflash.com/blogs/how-to-track-yc-startups-before-demo-day
- Author: [YCDossier Editorial Team](https://yc-dossier.nicheflash.com/authors/ycdossier-editorial-team)
- Publisher: YCDossier
- Published: 2026-10-15
- Updated: 2026-10-15

To track YC startups before Demo Day, choose a batch, create a baseline profile for each company, then collect dated, source-backed signals using the same method. Keep company claims separate from independent observations, preserve earlier records, and mark missing information as unknown rather than zero. This process works for any YC batch; Demo Day is a useful milestone for timing updates, not a requirement for the tracker.

 ## What should you track about YC startups?

 Use two linked records. A company profile holds relatively stable details, such as company name, batch, product description, founders, status, location, and the source and date you checked. A signal record holds a dated observation about a claim or visibility measure. Keeping them separate lets you update a profile without erasing what you previously observed.

 Use consistent evidence types: **company-disclosed** for information published by the company or a founder, and **independently observed** for information you collect from another source or through a repeatable search. Record status separately: **observed**, **not found in checked source**, **unavailable** when a previously accessible source cannot be checked, **excluded** when a result fails your rules, or **unknown** when you have not verified a value. Unknown and not found do not mean zero.

 ## How to build and maintain a YC startup tracker

 YC’s company directory can help identify companies and batches, but treat it as a starting point rather than a fixed count of a cohort.[1] Follow the same steps for each batch you track.

 ### 1. Select a batch and set the scope

 Choose the batch and decide which companies belong in the tracker: for example, companies listed in the directory when you start. Record the date you assembled the list and your inclusion rule. Add new companies as separate entries if the directory changes; do not silently change the original scope.

 ### 2. Record a baseline profile

 For each company, record its name, batch, product description, founders, status, location, profile source, and profile checked date. Copy claims carefully rather than rewriting them as verified facts. Note when a field is absent, and leave it unknown until you can check a source.

 ### 3. Choose signals and collect evidence

 Select a small set of signals you can define and repeat. For company-disclosed claims, capture the exact claim, unit, time period, and wording; record where it appeared and when you checked it. If the claim is numeric, note what the number counts and any stated denominator or measurement window. Do not convert a vague statement into a precise metric.

 For independently observed visibility, such as web mentions, log the exact search terms, search tool or source, collection date, and time window. Count only results that clearly refer to the company. Exclude namesakes, duplicate pages, and results outside your stated scope. Keep a note of what you included and excluded so another person can repeat the check.

 ### 4. Validate each claim and decide whether it is comparable

 Use this review checklist for every new signal:

 - Does the source clearly refer to the company, and can you identify who published it?
- Does the record preserve the claim’s wording, unit, definition, and measurement period?
- Is the observation dated, and can you tell when the underlying activity occurred?
- For a comparison, are the signal definition, source type, collection method, and time window sufficiently similar?

 Compare values only when the definitions, units, coverage, and time windows match closely enough to support the comparison. If they do not, keep the observations but label them incomparable or exclude them from that comparison; explain why. Mark a value unknown when you have not checked it. Use “not found in checked source” only when you have searched a named source and scope. Do not treat either status as evidence that activity did not occur.

 ### 5. Add dated updates and schedule the next check

 Each review creates a new signal record; it never overwrites an earlier one. If a company changes a claim, retain the previous wording and source, then add the new claim with its own date. If a source disappears, record it as unavailable and note what you could no longer access. Set a fixed weekly or monthly review interval, depending on how closely you need to monitor changes. Add a separate check around Demo Day if that milestone matters to your work.

 **Copyable tracker field schema**

| Record | Fields |
| --- | --- |
| Company profile | Company; batch; product description; founders; status; location; profile source; profile checked date |
| Dated signal | Company; signal; value and unit; evidence type; source; collection method; observation date; measurement window; definition; status; notes |

 **Fictional completed signal entry:** Company: Example Co. Signal: monthly active users. Value: unknown. Evidence type: independently observed. Source: Example Co. public website, homepage and About page. Collection method: manually reviewed both pages for an explicit monthly active users claim. Observation date: 2026-03-17. Measurement window: not stated. Definition: monthly active users as stated by the company. Status: not found in checked source. Notes: this entry covers only the two named pages, not the company’s other materials.

 ## How should you interpret web mentions?

 Web mentions measure visibility in the sources and time window you checked; they do not establish customer value, revenue, or traction. A company can receive attention without demonstrating those outcomes, and a low count can reflect limited coverage or a search method that misses relevant results. Use mentions to decide what to investigate next, not to declare a winner. Keep company-disclosed claims and independently observed visibility separate.

 ## What does YC Bench measure?

 Treat any score or visibility count as a prompt for further checks, not as a prediction that a startup will succeed.[2]

 ## Frequently asked questions

 ### How often should I refresh a public YC startup tracker?

 Use a fixed weekly interval for close monitoring or a monthly interval for a lighter overview. Record a new observation date each time and preserve previous entries.

 ### Can I compare startups from different YC batches?

 Only when the signal definition, source, collection method, and time window are comparable. Disclosure practices and available evidence can differ by batch, so state what you measured and when. If the conditions do not match, keep the records separate rather than presenting the comparison as like-for-like.

 ### What should I do when a company updates or removes a public claim?

 Keep the earlier observation with its date and source, then add the updated claim or unavailable status as a new record. Do not overwrite the old entry or assume why the claim changed.

 ## Sources

 - [1] Y Combinator company directory, used to identify listed companies and batches.

## References

1. [ycombinator.com](https://www.ycombinator.com/companies/?batch=S26)
2. [arxiv.org](https://arxiv.org/abs/2604.02378)