A tenant rep broker usually hears about a move when the tenant has already picked someone to run it. The company raised a round, hired 40 people and outgrew its floor, or it cut a team and wants to sublease half the space. By the time it calls a broker, the broker it calls is the one who reached out first.
Most of those moves leave a public trace months earlier. A layoff notice, a funding filing, a lease end date, a merger, a bankruptcy case. Each of these lives in a different database, under a slightly different company name, and almost no brokerage watches more than one or two of them.
This article covers six of those signals: what each one says about a tenant, where to get it, what it costs, and what it takes to turn six separate feeds into one alert a broker can act on.
TL;DR: Layoffs, funding rounds, lease expirations, new business registrations, M&A activity and bankruptcy filings each hint at a different kind of tenant move: growth, shrinking, relocation or exit. Each comes from a different source with its own gaps. To use them together you first match records that describe the same company, then score each signal by what it suggests and how close the move is likely to be, then test the score against moves you already know about in your market.
What each signal says about a tenant
A signal is only useful if you know which way it points. A funding round and a bankruptcy filing both mean "this company's space needs are about to change", but a broker calls those two companies with very different offers.
| Signal | What it usually suggests | Who the broker talks to |
|---|---|---|
| Funding round | Growth: more headcount, a bigger office or a new city | The tenant, about expansion |
| Layoffs | Shrinking: subleasing part of the space, or consolidating offices | The tenant, about sublease or downsizing |
| Lease expiration | A decision is due: renew, relocate or resize | The tenant, about options before the date |
| New business registration | A new entity or a new state: a first office or a satellite | The tenant, about a first lease |
| M&A activity | Consolidation: two offices merging into one, or one closing | The combined company |
| Bankruptcy filing | Exit or restructuring: leases may be rejected or kept | The tenant's advisers, and later the landlord with empty space |
No single signal proves a move. A funding round can go into product, not office space. A Chapter 11 tenant can keep every lease. The point of combining them is that two or three signals pointing the same way, about the same company, inside the same window, are much harder to explain away than one.
Where each of the six signals lives
Each signal has at least one real source today. None of them covers every tenant, and the gaps matter more than the prices.
| Signal | A source that exists today | What it costs | Where it stops |
|---|---|---|---|
| Layoffs | WARN Firehose, WARN Act notices from all 50 states in one feed | Free tier (25 calls a day), $49 a month paid API | The federal WARN Act applies to employers with 100 or more employees, and to layoffs of 50 or more workers at one site (between 50 and 499 only if they are at least a third of that site's staff). Several states, including California, New York and New Jersey, have their own WARN laws with lower thresholds. Small tenants still fall outside. |
| Funding rounds | SEC EDGAR Form D filings | Free, no API key | Form D is due within 15 days of the first sale, so it lands days or weeks after the round. Not every raise files one, and the amount is not always given. Crunchbase used to fill that gap for free and now needs a paid plan. |
| Lease expirations | Reonomy (ownership and lease data) or CompStak (lease comps, which include lease end dates) | Reonomy: $4,800 a year per user for the web app. CompStak: paid plans | Both cover what they have collected. Small leases and markets with few reported comps have thin data. |
| New business registrations | State Secretary of State registries, or OpenCorporates as an aggregator | Free tier: 200 requests a month, 50 a day | Coverage differs by state. OpenCorporates lists more jurisdictions than it has full company data for. The registered address is often a registered agent, not an office. |
| M&A activity | SEC EDGAR 8-K filings | Free | Covers deals where a public company is the buyer or the target. Most deals between private small companies never reach the SEC. |
| Bankruptcy filings | PACER, the federal court records system | $0.10 a page, capped at $3 a document. Searches are charged per page of results. You are not billed for a quarter in which you spend $30 or less | Built for looking up one case at a time. Watching many companies at once means repeated searches and paid pages. |
Several of these sources are run state by state, so a system built for one market will need changes before it works in the next.
The bigger gap is tenant size. The smaller the tenant, the fewer of these signals it leaves, and small and mid-size tenants are often the ones a single-market broker works with most.
Two kinds of tools come close and stop short. CRE data platforms such as Reonomy, CoStar, Cherre and CompStak cover properties, owners and leases, and none of them is built to watch layoffs, funding or bankruptcy. B2B sales data tools such as Bombora, 6sense and ZoomInfo track company events, including funding and layoffs, and cost $25,000 to over $300,000 a year, per Autobound's 2026 comparison. They score companies for software sales teams and do not tie a company to the lease and the building a broker cares about. We found no product that combines all six signals with lease data for CRE prospecting.
Turning six separate feeds into one alert a broker can act on takes three steps: match the records, score the signals, then test the score.
Step one: match records that describe the same company
Before you can score a tenant, you need to know which records belong to it. A WARN notice lists "Acme Logistics LLC." A state registration lists "Acme Logistics, Inc." A funding filing mentions only "Acme." Three records, one company. If the system keeps them apart, a broker gets three weak alerts and never sees the one strong one.
This problem has a name: entity resolution, or probabilistic record linkage. It compares records field by field and estimates how likely two of them describe the same company.
Splink, an open-source library from the UK Ministry of Justice's analytical team, does this at scale: about a million records in a minute or two on a laptop, and more than 100 million on larger infrastructure. It uses the Fellegi-Sunter model, which learns from the data itself and needs no hand-labeled training set. You still choose which fields to compare and where to set the match threshold.
A usable first version needs three rules:
- Clean names and addresses the same way every time (drop "LLC" and "Inc.", standardize street names).
- Look for a shared website domain, phone number or address across records.
- Merge two records automatically only above a set match score. Send anything below it to a person for a quick check. A missed merge costs a few minutes; a wrong merge sends a broker after the wrong company.
There is a second match most guides skip: company to building. A WARN notice names a specific site. A Form D names a headquarters. A state registration often shows a registered agent's address. To know which lease is in play, each record has to be tied to an actual property, using the address on the signal and the ownership and lease data from the table above.
Step two: score each signal by direction and timing
Once records are matched, each company has a set of signals. A simple score that adds them up treats a three-year-old bankruptcy the same as last week's layoff notice, and a funding round the same as a layoff. That score is mostly noise.
A better score answers two questions per company.
Which way is it moving? Keep growth signals (funding, new registrations) and shrinking signals (layoffs, bankruptcy) apart, or score each move type on its own. A company with both a funding round and a layoff notice is probably restructuring, and a broker needs to read the news before picking up the phone.
How soon? Signals age differently, so give each its own timing rule:
- Event signals (layoffs, funding, M&A, bankruptcy) lose weight as they age. Sales intelligence vendors weight software buying signals from the last 7 to 14 days most and fade anything past 30 days, per Autobound. Office decisions move slower than software purchases, so treat that window as a floor and widen it, for example to 3 to 6 months, until your own data says otherwise.
- Date signals (lease expirations) work the other way. They gain weight as the date gets closer. Tenants with larger spaces often start looking 12 to 24 months before the lease ends, so the score should rise across that window.
Weights between signals are a starting guess until you test them. Two or three signals pointing the same way inside the same window should count for more than any one signal alone.
Step three: test the score before you trust it
A score nobody has checked against real moves is a guess with a number on it. The check is simple. Take moves you already know about in your market from the last two or three years: tenants who relocated, expanded or left. Pull the signals that existed for those companies 6 to 24 months before the move, and see how many your score would have flagged, and how early.
Then run it the other way. Take a sample of companies the score flags today and see how many actually move. That number tells you how many calls a broker will make for each real lead, and it is the number that decides whether the system is worth building further.
What already exists to build on
None of the parts has to be invented. Record matching works today on 100 million records and more. Time-based weighting is standard in sales intelligence tools. The work is assembling these parts for one market's data and tuning them against that market's moves.
That assembly is the kind of work we in Brocoders have done in real estate. For AreaButler, our engineers joined the client's team and built features that combine CRM data, a real estate data platform and OpenAI into one property analysis for brokers. For Lake, we rebuilt a vacation rental platform's backend in 3 months. Lake then grew from 500 to 40,000 connected properties through 6 property management system integrations. Both sit within our wider proptech and real estate work. We have not yet shipped a tenant signal score for commercial real estate.
What a missed or false signal costs
A missed signal costs a commission. Leasing commissions commonly run 3% to 6% of the total lease value. On a 10,000 square foot, five-year lease at $40 per square foot a year, the lease is worth $2 million, so the commission is $60,000 to $120,000. One deal won because a broker called a month earlier can pay for a lot of data.
A false signal costs trust. A broker who calls a tenant about expansion right after it announced layoffs looks like they did not read the news. In a business built on relationships, a few calls like that cost more than the data saved.
Data alone does not solve either problem. A Reonomy seat is $4,800 a year, and B2B sales data tools run $25,000 to over $300,000 a year. Neither matches records across sources, ties companies to leases, or scores signals by direction.
Which parts to rent and which to own
Rent the data. WARN Firehose, SEC EDGAR, PACER, OpenCorporates and the lease data vendors already collect it, and copying them would take time for no gain. Check each one's access first: CoStar, for one, sells no API and forbids automated extraction, which we cover in build, buy or hybrid: AI tenant sourcing for CRE.
Own the logic on top: record matching, the company to building link, the scoring rules and the test results. These depend on how your market behaves, and they get better every time you check a flagged company against what it actually did. If a data vendor raises its price or changes its terms, you can swap the feed and keep your matching rules, your scoring and your history of what worked. That layer is what our AI development work builds, tuned to one market's data.
A starter framework you can run by hand
Most brokerages that do this today check each source by hand, one at a time, when someone has a free afternoon. Here is a version you can run on a spreadsheet before hiring anyone.
- Match first. For each new signal, clean the company name and address, and check whether the company is already on your list (same domain, phone or address). If you are not sure, check by hand before adding it.
- Tag the direction. Mark each signal as growth (funding, new registration), shrinking (layoffs, bankruptcy), consolidation (M&A) or decision due (lease expiration within 24 months).
- Apply timing. Let event signals fade over a few months. Let lease expirations rise as the date approaches.
- Flag patterns. Call when two or more signals point the same way for the same company. Read, do not call, when they point in opposite directions.
- Keep score. Record what each flagged company actually did. After six months you will know which signals work in your market.
This will miss things a built system would catch. It is enough to learn whether these six signals predict anything worth acting on in your market before you spend real budget on a build. If the answer looks promising, a short paid feasibility study is the lower-risk next step, and the same approach works for any proptech or real estate workflow built on scattered public data.
Sources: WARN Firehose; US Department of Labor, WARN Act; SEC EDGAR APIs; OpenCorporates API; PACER pricing; NextAutomation, Cherre vs. CompStak; Splink, Robin Linacre; Autobound, 2026 intent data provider comparison; Leasing commission structures.