A lease abstract is a structured summary of the commercially and legally significant terms of a lease, captured as discrete fields so that billing, accounting, reporting and critical-date tracking can run from data rather than from the document. A good abstract for a commercial lease has around forty fields, grouped into parties, premises, term, rent, escalations, recoveries, options, obligations, insurance and critical dates. Every reconciliation error, missed option and mis-billed escalation we've seen in a lease migration traces back to a field nobody abstracted.
This is the checklist we use when onboarding a portfolio into a new system, the fields that get missed most, the QA rule that catches them, and where AI abstraction helps and where it doesn't.
What a lease abstract is for
A lease is fifty to a hundred pages written by two sets of lawyers. The people who administer it, the property accountant billing rent, the analyst running the CAM reconciliation, the asset manager watching for option dates, will read it once, if at all. The abstract is the version they actually use. It has to be complete enough that nobody needs the PDF for routine work, and accurate enough that when they do open the PDF, it agrees.
Abstracts get made at three moments: when a lease is signed, when a portfolio is acquired (the due-diligence abstract), and when leases are migrated into a new system. The third is where this checklist earns its keep. Two hundred leases at forty fields is eight thousand data points, and the system you're moving into will bill from every one of them.
The 40 fields
Field numbers are for reference in the QA section. "Source" is where in a typical lease the field lives, because the abstractor's first job is knowing where to look.
Parties and premises
| # | Field | What to capture | Typical source |
|---|---|---|---|
| 1 | Landlord entity | Exact legal name and entity type of the lessor; the owning entity, not the management company | Preamble; signature page |
| 2 | Tenant entity | Exact legal name, entity type, state of formation; trade name separately | Preamble; signature page |
| 3 | Guarantor | Name, guaranty type (full, limited, rolling), cap and burn-off conditions | Guaranty exhibit |
| 4 | Premises | Suite/unit identifiers, floor, building address | Section 1; site plan exhibit |
| 5 | Rentable and usable area | Both figures, the measurement standard (BOMA 2017 Office, etc.) and whether area is subject to remeasurement | Section 1; definitions |
| 6 | Pro-rata share | Tenant's percentage of the building/project and the denominator used | Definitions; operating expense clause |
| 7 | Permitted use and exclusives | Permitted use; any exclusive granted to this tenant; any exclusives of other tenants that bind this one | Use clause; exclusives exhibit |
Term and dates
| # | Field | What to capture | Typical source |
|---|---|---|---|
| 8 | Lease execution date | Date signed by the last party | Signature page |
| 9 | Commencement date | The defined term, and whether it's a fixed date or triggered (delivery, substantial completion, opening) | Term clause; work letter |
| 10 | Rent commencement date | Often later than commencement; capture the trigger and the number of free-rent days | Rent clause |
| 11 | Expiration date | Fixed date or computed from commencement | Term clause |
| 12 | Lease year definition | Calendar year, or 12 months from commencement, or partial first year | Definitions |
| 13 | Early access / beneficial occupancy | Dates and conditions of access before commencement | Work letter; delivery clause |
Rent
| # | Field | What to capture | Typical source |
|---|---|---|---|
| 14 | Base rent schedule | Every step with start date, end date, annual and monthly amount, per-sf rate | Rent schedule / exhibit |
| 15 | Escalation method | Fixed steps, fixed %, CPI (which index, floor, cap), or market reset with the mechanism | Rent clause |
| 16 | Free rent / abatement | Months abated, which charges are abated (base only, or opex too), and any clawback on default | Rent clause; concessions rider |
| 17 | Percentage rent | Rate, breakpoint type (natural or fixed), tiers, reporting frequency, gross-sales definition flags | Percentage rent clause |
| 18 | Security deposit | Amount, form (cash or letter of credit), reduction schedule, conditions for return | Deposit clause |
| 19 | Late charge and default interest | Grace days, late fee amount or %, interest rate on arrears | Default clause |
Operating expense recoveries
| # | Field | What to capture | Typical source |
|---|---|---|---|
| 20 | Lease structure | Full-service gross, modified gross, NNN, absolute net | Definitions; opex clause |
| 21 | Base year or expense stop | Year or $/sf figure; whether grossed up; separate for taxes vs opex if applicable | Opex clause |
| 22 | Recoverable expense categories | CAM, taxes, insurance, utilities, management fee (and its cap %); which are billed separately | Opex clause |
| 23 | Gross-up provision | Target occupancy %, which expenses it applies to | Opex clause |
| 24 | Expense cap | Cap %, type (cumulative, non-cumulative, compounding), which categories are controllable | Opex clause |
| 25 | Exclusions | Any tenant-specific exclusions beyond the standard list (capital, leasing costs, etc.) | Opex clause; exclusions rider |
| 26 | Reconciliation terms | Deadline for landlord's statement, tenant's audit window, audit cost-shift threshold | Opex clause |
| 27 | Utilities | Who pays, how metered (direct, submeter, allocated), after-hours HVAC rate | Utilities clause |
Options and rights
| # | Field | What to capture | Typical source |
|---|---|---|---|
| 28 | Renewal options | Number, length, notice window (earliest and latest dates), rent basis (fixed, % increase, FMV with mechanism) | Options rider |
| 29 | Expansion / ROFO / ROFR | Space covered, trigger, notice window, rent basis, whether it survives assignment | Options rider |
| 30 | Termination option | Date, notice window, termination fee formula (unamortised TI + commissions + months of rent) | Options rider |
| 31 | Relocation right (landlord) | Whether landlord may relocate the tenant, cost allocation, notice | Relocation clause |
| 32 | Assignment and subletting | Consent standard, recapture right, profit-sharing split | Assignment clause |
| 33 | Co-tenancy (retail) | Named anchors or % GLA trigger, opening vs operating, remedy, cure period | Co-tenancy clause |
Obligations, improvements and insurance
| # | Field | What to capture | Typical source |
|---|---|---|---|
| 34 | Tenant improvement allowance | $ total and per sf, permitted uses, deadline to use, draw conditions, overage treatment | Work letter |
| 35 | Landlord's work / delivery condition | Itemised landlord obligations and delivery standard | Work letter |
| 36 | Maintenance and repair responsibility | HVAC, roof, structure, parking, interior: who maintains and who replaces | Repairs clause |
| 37 | Insurance requirements | Tenant coverage types and limits, additional insured wording, certificate renewal date | Insurance clause |
| 38 | Holdover | Rent multiplier (125%, 150%, 200%) and whether consequential damages apply | Holdover clause |
| 39 | Surrender and restoration | Removal obligations, restoration standard, treatment of cabling and specialty improvements | Surrender clause |
Critical dates (computed)
| # | Field | What to capture | Typical source |
|---|---|---|---|
| 40 | Critical date register | Every date derived from fields 8–39 that requires an action: option notice windows, rent steps, deposit reductions, allowance expiry, insurance renewals, reconciliation deadlines, expiration | Computed from the fields above |
Field 40 isn't really one field; it's the output of the other thirty-nine. A renewal option with a notice window of "no earlier than 12 and no later than 9 months before expiration" produces two dates that have to be on someone's calendar three years from now. If the abstract records the option but not the dates, the option will be missed.
The fields most often missed
Across migrations, the same fields come back blank or wrong. They share a cause: they live in clauses abstractors skip because the field on the template is a single box and the clause is a page.
-
Co-tenancy (field 33): Abstractors record "Yes" and move on. The remedy (reduced rent, percentage-only rent, termination) and the trigger (named anchor vs GLA percentage) are what the asset manager needs the day the anchor announces closures, and the difference between an opening and an operating co-tenancy clause is whether the risk has already passed.
-
Gross-sales definition (field 17): The abstract records the percentage and the breakpoint and skips the definition. Whether online orders fulfilled from the store count is a yes/no that changes the bill by thousands a year; it has to be a flag on the abstract, not a footnote. The wording issues are in our post on the percentage rent clause.
-
Cap type (field 24): "5% cap" gets recorded. Cumulative, non-cumulative or compounding does not, and the three produce different bills every year of the term. The comparison is in CAM caps and controllable expenses explained.
-
Base year gross-up (field 21): The base year figure is recorded; whether it was grossed up, and to what occupancy, is not. Every base-year escalation for the rest of the term depends on it.
-
Option notice windows (field 28): The option is recorded as "one 5-year option at FMV." The earliest and latest notice dates aren't computed, so nobody is warned when the window opens, and the tenant either loses the option or exercises it late and the landlord has a dispute.
-
Measurement standard (field 5): Rentable area is recorded; the standard and the remeasurement right aren't. When the building is remeasured under BOMA 2017 and the tenant's share moves, the abstract can't explain why.
-
Termination fee formula (field 30): "Termination option in year 5" is recorded. The fee, typically unamortised TI and commissions plus several months' rent, isn't, so the option can't be priced and the tenant's exercise notice starts an argument.
-
Guaranty burn-off (field 3): The guarantor is named; the conditions under which the guaranty reduces or ends (usually after N months of timely payment) are not, and the credit team is still treating a lapsed guaranty as security.
QA: the two-person abstraction rule
An abstract is only as good as its second reader. The rule that works: one person abstracts, a different person verifies every field against the lease, and the verifier signs the abstract. Not a sample of fields, every field, because the errors cluster in the fields that look simple.
Three refinements make it workable at scale:
-
Verify against the source, not the abstractor: The verifier opens the lease to the clause cited in the "source" column and reads it. If the verifier is checking the abstract against the abstractor's memory of the lease, the exercise is theatre.
-
Test the computed dates: For field 40, the verifier recomputes at least the option windows and rent steps independently and compares. Date arithmetic is where "12 months before expiration" becomes 13 or 11.
-
Reconcile to the rent roll: Before a migration goes live, the abstracted base rent for every lease is totalled and compared to the current month's actual billings. A gap of more than a rounding difference means a field is wrong somewhere, and it's cheaper to find it now than in the tenant's first invoice from the new system.
For a 200-lease portfolio, budget roughly two hours per lease for abstraction and one for verification on a standard office lease; double both for retail leases with percentage rent and co-tenancy, and for anything over 80 pages.
AI abstraction: what it gets right and what it gets wrong
AI extraction tools have changed the economics of abstraction in the last two years, and they're worth using. They also fail in specific, predictable ways, and the two-person rule doesn't go away; it changes shape.
What AI does well: locating clauses in long documents, pulling clean numeric fields (rent schedule, area, dates, deposit), handling volume, and flagging where a clause exists at all. On a standard-form lease it will fill 30 of the 40 fields correctly most of the time, and it will do 200 leases in an afternoon.
Where it goes wrong:
- Amendments: The model reads the original lease and the first amendment as two documents and reports the original rent. Leases with three or four amendments are the norm in a mature portfolio, and the abstract has to reflect the amended state. Feed the full document set in order and verify the amended fields by hand.
- Defined terms with non-obvious definitions: "Lease Year" defined as the twelve months starting on the first day of the month after commencement; "Operating Expenses" with a two-page exclusions list; "Gross Sales" with a carve-out for online orders. The model captures the term and misses the definition.
- Conditional and nested provisions: A renewal option that is void if the tenant has been in default more than twice, or a co-tenancy remedy that applies only after a 90-day cure period. The model records the right and drops the condition.
- Cross-references: "The Breakpoint shall be as set forth in Section 4.3(b)" resolved to the wrong section, or not resolved at all.
- Confidence: The output looks the same whether the model is sure or guessing. It won't tell you which fields to check.
The practical setting: use AI for the first pass on every lease, then have a person verify every field against the source, with particular attention to fields 3, 17, 21, 24, 28, 30 and 33. The person who would have been the abstractor becomes the verifier. That's still a large saving, and it's the version that survives the first reconciliation season.
Mapping abstract fields to lease records in NetSuite
The reason to abstract into fields rather than into a Word summary is that the fields drive the system. In a property management platform on NetSuite, the forty fields above map directly onto the lease record and the billing, accounting and alert rules attached to it:
| Abstract fields | Where they land | What they drive |
|---|---|---|
| 1–7 Parties and premises | Customer record; lease header; unit/space record | Invoicing entity, pro-rata share for recoveries, exclusives conflict checks |
| 8–13 Term and dates | Lease header | Straight-line rent calculation, commencement of billing, lease-year cycle for reconciliations |
| 14–19 Rent | Rent schedule lines; percentage rent rule; deposit ledger | Monthly recurring charges, straight-line schedule, percentage rent billing, deposit liability |
| 20–27 Recoveries | Recovery setup on the lease: structure, base year, cap, gross-up, exclusions | Monthly estimates, annual reconciliation, cap tests, gross-up calculation |
| 28–33 Options and rights | Option records with computed notice dates | Critical-date alerts, renewal pipeline, termination fee calculation |
| 34–39 Obligations and insurance | Allowance commitment; work order responsibility matrix; insurance record | TI draw tracking, maintenance assignment, COI expiry alerts |
| 40 Critical dates | Generated from the above | Alerts to leasing, accounting and asset management at configurable lead times |
When the abstract is entered as data, the tests in the QA section become reports: total abstracted base rent vs billed rent, leases with an option and no computed notice date, leases with a cap and no cap type, leases with percentage rent and no gross-sales flag. That's how RIOO's leasing management on NetSuite treats an abstract: as the lease record itself, not a document filed next to it. The test for any system: pick a lease with a renewal option, and ask when the notice window opens. If the answer requires opening the PDF, the abstract hasn't been captured, it's been summarised.
Frequently asked questions
Q1. What is a lease abstract?
A lease abstract is a structured summary of a lease's key terms captured as discrete data fields: parties, premises, term, rent schedule, escalations, expense recoveries, options, obligations, insurance and critical dates. It allows billing, accounting and date tracking to run from data rather than from the lease document.
Q2. What should be included in a lease abstraction checklist?
At minimum: landlord and tenant entities, guarantor, premises and area with measurement standard, pro-rata share, commencement and expiration dates, the full base rent schedule, escalation method, free rent, percentage rent terms, security deposit, lease structure, base year or expense stop, recoverable categories, gross-up, caps and their type, exclusions, reconciliation deadlines, every option with its notice window, TI allowance terms, maintenance responsibilities, insurance requirements, holdover and surrender terms, and a computed register of critical dates.
Q3. How long does lease abstraction take?
Roughly two hours to abstract and one to verify a standard office lease; double for retail leases with percentage rent and co-tenancy provisions or for leases over 80 pages with multiple amendments. AI-assisted extraction reduces the abstraction time substantially but not the verification time.
Q4. Can AI abstract leases accurately?
AI tools reliably locate clauses and extract numeric fields such as rent schedules, areas and dates. They are unreliable on amendments, defined terms with unusual definitions, conditional provisions and cross-references, and they don't signal uncertainty. Use them for the first pass and have a person verify every field against the lease.
Q5. What is a critical date in a lease?
Any date on which a party must act or lose a right: renewal and termination option notice windows, rent step dates, deposit reduction dates, allowance expiry, insurance certificate renewals, reconciliation statement deadlines and lease expiration. Critical dates are computed from the abstract's term and option fields and tracked with lead-time alerts.
.svg)
.svg)
.svg)