How to Estimate AP Accruals: 4 Methods by Expense Type

It is day two of the close. Legal reviewed a contract last week, the cleaning crew finished the month's service on the 30th, and the hosting bill hasn't arrived yet. The services were delivered, the expense belongs in the period, and you need a number on the books before the review meeting tomorrow morning. The question is not whether to accrue - it is which estimation technique produces a figure your controller and your auditor will accept. This guide is a working reference on AP accrual estimation methods for staff accountants who need to book uninvoiced services this month and defend the numbers later.

Estimating accruals for uninvoiced services is harder than it sounds. Vendors deliver on their own cadence and bill on another, so the expense is real before any document lands in AP. Contracts behave differently by expense type: some are fixed monthly retainers, some are usage-metered, and some are project-based with milestone billing. Prior invoices are a useful anchor but can mislead when usage shifts, rates change mid-period, or the vendor's billing cycle straddles your close. Materiality thresholds also shape the work - a $300 courier accrual does not deserve the same rigor as a $40,000 legal estimate.

Four methods cover the vast majority of routine accruals:

Prior-period actual - use the most recent closed period's expense for the same vendor and GL account.
Contract rate - multiply the contracted rate by the proportion of the period elapsed.
Most-recent-invoice - use the dollar value of the vendor's last received invoice.
Rolling average - average the vendor's last three to twelve invoices.

Which method fits depends on the expense type and the audit evidence you can produce, which is why there is no universal default. The sections that follow define each method in practical terms, map methods to the six expense types you actually encounter, walk through the audit evidence each method requires, address materiality, and cover the true-up discipline that turns month-over-month variance into a signal rather than noise. Accrual estimation is one step inside the full AP month-end close process, and the quality of the estimate shows up in how quickly the rest of the close moves.

The Four AP Accrual Estimation Methods

Four estimation techniques cover the vast majority of month-end accruals for services delivered but not yet invoiced. Each has a specific information requirement and a natural fit. Pick one per accrual line based on what the vendor's billing pattern makes available, then document the choice on your worksheet so reviewers can follow the logic.

Prior-period actual

Definition: Use the vendor-and-GL-account's posted expense from the immediately prior closed period as the current period's estimate.

Inputs: The prior period's GL posting for that vendor and account. Nothing else is strictly required, though known volume or activity signals for the current period should flag whether the prior amount remains a reasonable proxy.

Fit: Stable expenses where month-to-month variance is typically small, or situations where no executed contract exists and vendor invoice history is too thin to support a rolling calculation. This is also the fallback when a new vendor delivered service in the current period but has only one prior billing cycle on file.

Example: If last month's freight charge from the same carrier was $4,820 and shipment volume this month was comparable, book $4,820 subject to adjustment if a known volume signal (a large outbound shipment, a paused lane) suggests otherwise.

Contract rate

Definition: Apply the contracted fixed rate to the proportion of the period during which the service was delivered.

Inputs: An executed contract or purchase order, the contracted rate or monthly fee, and the elapsed service period, typically the full month for a monthly retainer, or a pro-rata fraction when service started or ended mid-period.

Fit: Fixed-fee arrangements without consumption variability. This is the method for SaaS subscriptions billed at a flat monthly rate, retainer professional services (legal, tax, advisory), and contracted recurring services such as janitorial, security, pest control, or managed IT.

Example: If the signed cleaning contract is $1,200 per month and the cleaner delivered service for every scheduled day in the month, book $1,200. If service began on the fifteenth of a thirty-day month, book $600.

Most-recent-invoice

Definition: Use the dollar value of the vendor's most recently received invoice for the same recurring service as the current period's estimate.

Inputs: The previous invoice from that vendor for the same service line. A quick check that the service scope has not changed since the prior bill, same meter, same lanes, same headcount, protects the estimate from becoming stale.

Fit: Recurring variable services where the last bill is a reasonable proxy for this period and no contract rate governs the amount. Useful when vendor history is too short for a rolling calculation or when the last bill already reflects the current cost structure.

Example: If last month's utility bill for the same meter was $3,640 and consumption patterns are comparable, book $3,640.

Rolling average

Definition: Calculate the average of the vendor's last three to twelve invoices for the same service and use the average as the estimate.

Inputs: A clean history of the vendor's prior invoices for the same expense, plus a documented policy on the averaging window. A three-month window suits short history or a fast-changing cost structure; a twelve-month window smooths seasonal variation.

Fit: Variable recurring services where a single-period outlier could distort a most-recent-invoice estimate. Telecom bills with usage spikes, utilities with seasonal swings, and freight with volume fluctuation are typical candidates.

Example: If the last six monthly telecom bills averaged $2,115, book $2,115 and note the window and invoice set on the worksheet.

The principle behind method selection

The accuracy each method can credibly achieve tracks the information the vendor's billing pattern exposes. Contract rate produces the tightest estimates because the variable is already fixed in writing. Rolling average and most-recent-invoice produce tighter estimates where billing is frequent and consistent enough that recent history reliably predicts the current period. Prior-period actual is the fallback when vendor history is thin and no contract anchors the number. It is still defensible, but it usually carries the widest expected variance at true-up. Once you have selected a method and calculated a number, the journal-entry mechanics of recording and reversing an accrual covers the recording and reversal side of the entry.

Match the Method to the Expense Type

The four vendor-billing patterns from the previous section repeat across expense categories. A utility bill behaves like a utility bill whether you are in manufacturing or professional services, and a software subscription behaves like a software subscription regardless of industry. Matching the method to the expense type is the fastest route to a defensible number because you are reusing a pattern that already fits the underlying billing behavior.

Quick scan before the detail:

Utilities: Primary method is most-recent-invoice when usage is stable, with rolling average when seasonality matters. Fallback is prior-period actual. Watch for seasonal swings, rate changes, and meter anomalies.
SaaS and subscriptions: Primary method is contract rate. Fallback is most-recent-invoice. Watch for mid-period upgrades, overages, and proration.
Professional services: Primary method is contract rate for retainers and prior-period actual for time-and-materials work. Fallback is rolling average. Watch for scope changes, milestone billing, and uneven cadence.
Telecom and connectivity: Primary method is rolling average. Fallback is most-recent-invoice. Watch for overage, roaming, and late carrier reconciliations.
Freight and logistics: Primary method is prior-period actual when volume is comparable and rolling average when it is not. Fallback is a volume-weighted rate-card calculation. Watch for fuel surcharges, peak pricing, and accessorials.
Contracted recurring services: Primary method is contract rate. Fallback is prior-period actual. Watch for one-off extras, annual escalators, and variable pass-throughs.

Utilities (electricity, water, gas)

Primary method: Most-recent-invoice when meter usage is stable month over month. Rolling average when seasonality matters, using a seasonally-matched window so your reporting periods line up (a December accrual compares against prior Decembers, not against July).
Fallback method: Prior-period actual.
Key inputs: The last bill for the specific meter, or the prior three to twelve bills for that meter depending on how you size the rolling window.
Specific risks: Seasonal swings in HVAC load, published rate changes from the utility commission, and consumption anomalies from equipment failures or facility shutdowns.
Example: A warehouse with a stable electricity baseline accrues at last month's kWh charge; an office with heavy summer cooling uses a twelve-month rolling average to smooth the July spike.

For the full mechanics of splitting a bill that straddles the close date, see daily-rate proration for utility bill accruals, which walks through the day-count math in detail.

SaaS and subscriptions

Primary method: Contract rate, pulled from the executed subscription agreement.
Fallback method: Most-recent-invoice where the subscription is tiered and last month's seat count or usage volume held steady.
Key inputs: Executed subscription agreement, the stated monthly fee, and current seat count or tier level.
Specific risks: Mid-period upgrades that push you into a higher tier, overage charges for API calls or storage, and proration when seats are added or removed partway through the billing cycle.
Example: A CRM platform billed at a flat $4,800 monthly subscription fee accrues at contract rate; a data warehouse billed on query volume accrues at last month's invoice because usage has flattened.

True-ups are common here because monthly billing frequently reconciles overage from the prior period, so your accrual and the eventual invoice will differ by the overage amount almost every month.

Professional services (legal, consulting, audit)

Primary method: Contract rate for fixed retainers; prior-period actual for time-and-materials engagements.
Fallback method: Rolling average for time-and-materials where monthly hours are reasonably stable.
Key inputs: The engagement letter or retainer schedule for fixed arrangements, and prior invoices for time-and-materials work.
Specific risks: Uneven billing cadence (a law firm may bill twice in one month and skip the next), phase-based invoicing tied to deliverables, end-of-project billing spikes, and mid-month scope changes that invalidate a prior-period baseline.
Example: A monthly legal retainer of $12,000 accrues at contract rate; a consulting engagement running 80 to 120 hours a month at a blended rate accrues at the three-month rolling average.

Telecom and connectivity

Primary method: Rolling average over three to six months.
Fallback method: Most-recent-invoice when carrier plans are stable and monthly usage is flat.
Key inputs: Trailing invoice history for the specific account, broken out by line or circuit if the carrier bills that way.
Specific risks: Variable overage charges, international roaming or long-distance usage, equipment and installation one-offs, and billing reconciliations that arrive a cycle or two late.
Example: A multi-line corporate mobile account with fluctuating international usage accrues at the six-month rolling average; a stable domestic mobile plan with flat usage accrues at last month's invoice.

If the telecom service is a fixed-fee circuit with no usage component, treat it like any other fixed contract and use contract rate rather than last invoice.

Freight and logistics

Primary method: Prior-period actual when shipment volume is comparable to the prior month; rolling average when volume fluctuates meaningfully between periods.
Fallback method: A volume-weighted calculation using current-period shipment counts multiplied by vendor rate card pricing, where you have both.
Key inputs: Prior invoices from the carrier, internal shipment logs, and the current carrier rate card.
Specific risks: Fuel surcharges that move with diesel prices, peak-season pricing from mid-November through December, one-off lanes outside your normal routing, and accessorial charges for liftgate, residential delivery, or detention.
Example: A distributor with flat weekly LTL volume accrues at last month's carrier invoice; a manufacturer with variable outbound freight accrues using shipment counts from the WMS multiplied by the contracted per-mile rate.

Freight frequently sits alongside received-but-not-invoiced physical inventory at month-end, which is where the handoff to the GRNI accrual pattern for goods received but not invoiced becomes important. If the goods have been received but neither the freight nor the product invoice has arrived, both sides need to be accrued.

Contracted recurring services (janitorial, security, maintenance, landscaping)

Primary method: Contract rate.
Fallback method: Prior-period actual where contract terms include variable or pass-through components (consumable supplies, overtime call-outs).
Key inputs: The service contract, the stated monthly fee, and any variable pass-through items billed on top of the base fee.
Specific risks: Extra one-off services billed separately (emergency cleanups, additional security coverage for events), annual CPI or flat escalators that take effect mid-period, and performance-based adjustments tied to service-level metrics.
Example: A janitorial contract at $3,200 per month accrues at contract rate; a landscaping contract with a base fee plus separately-billed snow removal accrues at the prior-period actual during winter months.

One more consideration sits upstream of estimation. Getting the AP cutoff procedure for late and post-cutoff invoices right determines which invoices you post to the current period as actual expense versus which ones you accrue, and a cutoff error can double-count or omit an item regardless of how defensible your estimation method is. Confirm cutoff first, then estimate what remains.

Audit Evidence for Each Estimation Method

Accrual estimates draw disproportionate scrutiny because the assumptions behind them are often informal, carried in a staff accountant's head rather than filed alongside the journal entry. According to a Journal of Accountancy analysis of auditing accounting estimates, the International Forum of Independent Audit Regulators has reported that roughly 29% of all audit findings relate to accounting estimates, with most findings stemming from auditors failing to assess the reasonableness of the underlying assumptions. Translated for your close: the number itself is rarely what fails review. What fails review is the absence of a contemporaneous record explaining how the number was derived and why the inputs were appropriate. Audit defensibility for AP accruals is a documentation problem first and a methodology problem second.

What follows is a method-by-method evidence map, in the same order as the four methods introduced earlier. For each, you get the primary evidence an auditor will expect to see attached to the accrual schedule, the secondary evidence that strengthens the position, and the single question most likely to land in the review meeting.

Prior-period actual. Primary evidence is the prior period's GL extract showing the vendor and account's posted expense, accompanied by a short note explaining why current-period activity is comparable. That note is the load-bearing element: without it, you have copied a number rather than estimated one. Secondary evidence consists of operational signals confirming that volumes or rates did not materially change during the period, such as shipment counts, headcount reports, or system usage logs tied to the expense driver. The likely auditor question: "What would have prompted you to pick a different number?" If you cannot answer that cleanly, the method was applied on autopilot.

Contract rate. Primary evidence is the executed contract or purchase order showing the rate and term, paired with a calculation tying the booked accrual to the contracted rate for the elapsed portion of the period. The calculation should be reproducible by a reviewer working only from the contract and the days-elapsed figure. Secondary evidence is proof the service was actually delivered, which depending on the vendor might be a sign-off email, a service ticket, a utilization report, or a delivery confirmation. Without delivery evidence, you have accrued for a promise rather than a performed service. The likely auditor question: "Was there any variable component in the contract that the rate alone does not capture?" Usage-based overages, tiered pricing, and true-up clauses are the common traps.

Most-recent-invoice. Primary evidence is the prior vendor invoice itself, with a visible tie from its amount to the current-period accrual. "Visible tie" means the invoice is attached or referenced by document number, not recalled from memory. Secondary evidence is a one-sentence note explaining why last month's bill is representative of this month, usually citing stable usage, the same seat count, the same headcount of licensed users, or the absence of a rate change. The likely auditor question: "What in this period is different from the period that invoice covered?" This method collapses quickly when pricing changed mid-period or a contract renewed with a step-up.

Rolling average. Primary evidence is the actual invoice set used in the average, the calculation worksheet that produced the average, and a written policy on the averaging window (three, six, or twelve months) and why that window fits the expense pattern. The policy matters as much as the math, because an averaging window selected per-accrual rather than by rule looks like hindsight. Secondary evidence is a variance check comparing the rolling average to the most recent month, so a recent outlier surfaces rather than being buried inside an average. The likely auditor question: "Why this window? What would change the window?" Have a defensible answer ready: seasonality, contract cadence, or volume stability, not "it felt right."

The principle underneath all four: audit defensibility is not about using the most sophisticated method, it is about using the method best supported by the information you actually have, and keeping that support filed with the accrual schedule. A contract-rate accrual with no contract attached fails review faster than a well-documented rolling-average estimate, because the reviewer can reproduce the second and cannot reproduce the first.

Materiality and Method Choice

Materiality in this context is the dollar level below which an accrual error would not change a reasonable reader's view of the financial statements. That threshold is set by the facts of the entity, often through the external auditor's planning materiality, an internal finance policy, or a stated percentage of a base such as pretax income or total expense, so no single cutoff applies to every close.

The practical point is simple: the rigor of the method should scale with the size of the accrual.

Small and individually immaterial. Prior-period actual or most-recent-invoice is usually enough. A full rolling-average build for a minor monthly service charge is disproportionate effort, but the support still needs to exist: attach the prior invoice or prior-period detail and leave a short note on why the amount is still representative.

Moderate and near the threshold. Most-recent-invoice or rolling average is the safer choice because the number is large enough to attract review but not always large enough to justify contract-level rebuilds. This is where the method choice should be explicit on the accrual schedule: note why last invoice is representative, or name the averaging window and file the invoice set behind it.

Large and individually material. Contract rate is the cleanest method where a fixed fee exists. If no fixed rate exists, use rolling average with a named window, a variance check against the most recent month, and primary support attached, such as the contract, invoice sample, or calculation worksheet. At this tier, assume someone will ask how the number was derived and whether a different method would have produced a meaningfully different result.

There is also an aggregation problem. Ten individually immaterial accruals estimated with quick methods can still add up to a material balance when they hit the same account family or reporting area. Before applying a light-touch method across the board, compare the aggregate accrual population against the materiality threshold rather than judging each line in isolation.

When the tier is uncertain, default to the more rigorous method. The extra work to build a rolling average or tie the number back to a contract is usually modest compared with the time lost defending an avoidable variance after review.

True-Up Discipline — Closing the Loop Between Estimate and Actual

Every true-up has three moving parts: reverse the prior period's accrual, record the actual invoice at its real amount, and calculate the variance between estimate and actual for each vendor. The mechanics belong to the journal-entry process already covered earlier; the discipline here is what you do with the variance once the posting is complete.

Set a simple review rule and use it consistently. Many teams start by flagging any vendor whose estimate-to-actual variance exceeds 5% or 10% for method review. Variances below that threshold usually confirm the current method is fit for purpose. Variances above it mean the method, the inputs, or the vendor billing pattern deserves another look before next month.

Persistent positive variance. If actual invoices keep landing above the estimate, the method is systematically under-accruing. Shorten the rolling-average window so recent price movement carries more weight, add a modest buffer where policy allows, or check whether an annual price escalator or usage change was missed in the contract review.

Persistent negative variance. If actuals keep arriving below the accrual, the team is over-accruing and parking expense in the wrong period. That often means the rolling average still reflects an older, higher run rate. In that situation, a most-recent-invoice method may be more faithful if the trend is clearly downward, or the service level may have been reduced without the accrual worksheet catching up.

Erratic variance with no pattern. Some vendors simply bill irregularly enough that no tight estimate is available. Professional-services invoices tied to milestones and freight invoices exposed to surcharges are common examples. The right response is partly vendor-side, better reporting cadence, cleaner PO discipline, clearer contract terms, and partly methodological: a wider rolling-average window usually produces a more stable estimate than anchoring to the last invoice.

True-up discipline is what turns accrual estimation from judgment into a managed process. A worksheet that keeps each month's estimate, actual, and variance by vendor creates evidence for the next close and shows whether the method is tightening over time. That history is often as valuable to a reviewer as the current month's calculation because it proves the team is learning from its misses rather than repeating them.

How Structured Vendor-Invoice History Strengthens Your Estimates

The most-recent-invoice and rolling-average methods fail less often on theory than on input quality. Both depend on a clean history of vendor invoices that can be filtered by vendor, period, and expense type. When that history lives in PDF folders, email threads, or a general ledger export that strips away service-line detail, the estimate becomes harder to reproduce and much harder to defend.

In practical terms, structured history means invoice data arranged in rows and columns: vendor, invoice date, amount, service line, GL account, and any other fields the close team needs to sort and average. That is the real input behind a rolling-average accrual. Instead of re-keying values from a stack of prior bills, the accountant can filter the vendor's history, pick the relevant window, and show exactly which invoices were included in the calculation.

That changes the audit defense materially. A rolling-average accrual supported by a spreadsheet that shows the invoice set used, the averaging window, the resulting calculation, and a source-file reference for each row is a different-quality workpaper from the same number produced in a scratch tab with no trail back to the underlying bills. The method has not changed, but the evidence behind it has become reproducible.

That is why the category of tool matters. An accountant using an AI tool that converts invoices into structured spreadsheets can produce Excel, CSV, or JSON outputs that preserve vendor-level detail and give the close team a cleaner base for filtering, averaging, and cross-referencing source documents. In other words, the tool does not replace the accrual policy; it improves the inputs the policy depends on.

That is the framework to keep: four methods, a decision matrix by expense type, an evidence map tied to each method, and a true-up loop that tests whether the choice was right. When those pieces work together, AP accrual estimation stops being a month-end guess and starts behaving like a policy the team can explain, repeat, and improve.

Structured vendor history makes that policy stronger because it sharpens the two methods that rely most on prior billing data, while true-up discipline keeps correcting the method when the data says it should. A close package that shows both, reliable source inputs and a visible feedback loop, is usually the clearest sign that the accrual process is getting tighter month after month.