Which AR Platform Forecasts Collections at 30/60/90 Days?

TL;DR: A 30/60/90-day collections forecast says how much of what customers owe you will arrive in the next 30, 60 and 90 days. Most are built by sorting open invoices by age, and most miss for four reasons: disputes, payments nobody has matched, customers who always pay late, and part-payments. This guide explains what a good forecast uses, how to test any tool against your own last quarter, and how four platforms compare: Monk, HighRadius, Tesorio and Upflow. For each one, starting with Monk, we list what to ask to see on your own invoices.

Sponsored Listing by Monk

Key takeaways

• A forecast is about dates. An aging report says what is owed and how late it is. A forecast says when the money is likely to arrive.

• Four things break most AR forecasts. Disputes, payments that arrived but were never matched to an invoice, customers who always pay late, and part-payments.

• A good forecast reads each customer's own history, plus promise-to-pay dates, dispute status and matched cash.

• The fair test is your own last quarter. Run the tool as of 90 days ago and compare its forecast with the cash that really came in.

• Monk leads this guide for the groundwork a forecast depends on. Its pages cover collections, cash application and portal work. They mention forecasting in one line and describe no 30/60/90-day view, so ask Monk to show you one.

• HighRadius, Tesorio and Upflow describe forecasting on their own pages, from three starting points: treasury, finance planning and day-to-day collections.

What a 30/60/90-day collections forecast is

Your open invoice report tells you what you are owed and how late each invoice is. A collections forecast answers a different question: when will the money land?

It sorts expected cash into three windows: the next 30 days, days 31 to 60, and days 61 to 90. The first window covers payroll and bills. The later two support decisions that need more notice, such as hiring or drawing on a credit line. It is a forecast of cash in from invoices you have already sent, not a sales forecast.

Here is a made-up example. Three customers each owe $10,000 on an invoice due today. Customer A pays within a few days, every time. Customer B has paid about six weeks late on each of its last six invoices. Customer C has a dispute open over a missing purchase order number. The aging report shows three identical invoices. A useful forecast puts A in the first window, B in the second, and C in none until the dispute is cleared.

For scale, the average company's Days Sales Outstanding rose to 59 days in 2023, according to Allianz Research. That is an average across many companies, so treat it as background and not as a target. It does show that for a typical company the 60-day window is not a distant one.

Why forecasts built on AR alone miss

The simple method sorts the aging report into buckets and assumes each bucket pays on a fixed schedule. It is quick to build and wrong in predictable ways.

Disputes. An invoice in dispute will not be paid on its due date, and nobody can say when it will be. A forecast that ignores dispute status keeps counting the money in the next window, month after month.

Payments nobody has matched. A customer pays, but nobody applies the money to the invoice because the remittance did not say which invoices it covered. The aging report still shows the invoice as open, so the forecast counts cash that has already arrived as cash still to come.

Customers who always pay late. An average across all customers hides the fact that one pays on day 10 and another on day 70. A forecast built on payment terms or one company-wide average is wrong for both, in the same direction every month. This is the most fixable of the four, because the history is already in your books.

Part-payments. A customer pays most of an invoice and holds back the rest, often over a deduction. Some forecasts count the whole invoice as collected, and others count the whole thing as open. Neither is right.

A fifth cause matters for larger customers: invoices that never reach the person who pays them. Many large buyers require submission through a vendor portal, and an invoice emailed to a shared inbox may just sit there. The money then arrives when the portal submission is made, not on the due date.

These causes share a root. The forecast is using the age of the invoice, when the facts that decide timing sit in the customer's habits and in the notes your collectors write down.

What a good forecast uses

Four inputs matter most, and each is something you can ask a vendor about.

Some tools add more. Tesorio's page lists seasonal patterns and invoice terms. Upflow's page says it layers in upcoming billings beyond your current open invoices. Neither replaces the four above.

A forecast can only read what someone recorded. If collectors keep promise dates in their inboxes and disputes on a separate spreadsheet, no tool can use them.

How to test a tool's forecast against your own last quarter

Do not judge a forecast on a demo with sample data. Test it on yours.

1. Pick a cut-off date about 90 days ago. Call it day zero.

2. Freeze what was open on day zero: every open invoice with its amount, due date, customer, dispute flag and any promise note, as they stood then and not as they stand today.

3. Ask the vendor to forecast from that list, using only what was known on day zero. If the tool cannot be run as of a past date, ask for a pilot that runs forward for 90 days. It is slower, but it is the same test.

4. Record the cash that really arrived in days 1 to 30, 31 to 60 and 61 to 90.

5. Compare the two, in total and for your ten largest customers, and run the same test on your current spreadsheet. A tool that does no better than the spreadsheet has told you something.

6. Study the pattern of the misses. Is it always early? Wrong on disputed invoices? Wrong on the same three customers?

Two warnings. Make sure the tool has not already seen the payments you are testing it against, and ask for that in writing. And be wary of any single accuracy figure. Ask what was measured (dollars or invoices), over which window, and for how long. A percentage with no method behind it tells you very little.

How we evaluated these tools

We asked four questions of each platform. Do its own pages describe a forecast of when invoices will be paid? Which of the four inputs does it name? Does it compare forecast with actual cash? What kind of team is it built for?

We read each company's public pages and describe only what they say. Where a page is silent we say so. Silence does not prove a feature is missing, but it puts the question on your demo list. Every Monk figure below is the company's own and has not been independently audited. We do not list prices, because none of the pages we read published them.

The four

1. Monk

Best for: finance teams that want collections, cash application and portal work in one platform, and want to see the forecast run on their own data.

Monk is an AI-based accounts receivable platform that sits on top of the systems you already use. Its platform page covers the groundwork that AR collections forecasting depends on. It says Monk handles "every step of collections to cash application and reporting," including payment portals and purchase order mismatches, with escalations to your team "only when necessary or per the rules that you set." Your data is mapped one to one and kept in sync.

Monk's cash flow velocity playbook says "Monk brings together collections, cash application, and forecasting." Its public pages don't spell out which windows the forecast covers or what it reads, so the best way to judge it is on your own invoices.

What the pages do describe is groundwork. The playbook tells readers to track promise-to-pay signals and calls them "the most undervalued signal in AR," though it does not say whether Monk's own forecast reads them. It recommends standardizing dispute workflows. Cash application, which means matching payments to invoices, is a core part of the platform. And Monk's vendor portal figures matter for a forecast, because an invoice that has not been submitted will not be paid on the date a forecast assumes.

In a demo, ask Monk to show you, using your own invoices:

1. A cash-in forecast for 30, 60 and 90 days, in total and for each customer.

2. Which inputs it reads: payment history, promise dates, dispute status and matched cash.

3. The same forecast run as of a past date, using only what was known then.

4. The forecast set against the cash that actually arrived.

5. What happens to an invoice not yet submitted to a customer's portal.

Monk reports the figures below for its own platform. They describe its customers overall and several are averages, so treat them as what happened across Monk's customers and not as a forecast for yours. None of them measures forecast accuracy.

A lower DSO is not the same as a better forecast, because a team can collect faster and still be surprised by when the cash arrives. Use the back-test to find out whether the forecast is any good.

Where it does not fit: a business whose main need is a standalone treasury forecast across bank accounts, payables and payroll, or one with so few customers that a single person already knows when each will pay.

Also be clear about what the pages do not say. They do not describe the forecast, its windows or its inputs, or any way to compare it with actual cash. Ask Monk to answer each on your own invoices.

2. HighRadius

Best for: treasury and finance teams at mid to large-sized companies that want a forecast of total cash flow with scenarios, and receivables as one input.

HighRadius has two relevant pages. Its cash flow forecasting page is written for treasury and calls itself "tailor-made for mid to large-sized companies." An AutoML feature lets you choose from multiple models, based on your bank and ERP data. The page offers a scenario builder and variance analysis for each cash flow category. One line is specific to receivables: the product can upload invoices to more than 100 AP portals and capture promise-to-pays and disputes.

Its collections management page describes a score for each customer, built from past-due amounts, credit risk and payment history, which predicts late payers. A dispute prevention agent flags likely disputes early.

The trade-off follows. The forecasting page is about cash flow for the whole business, not a customer-by-customer collections forecast in 30, 60 and 90-day windows, and it does not claim to be. It also quotes accuracy figures, which are the vendor's claims to check with the back-test. HighRadius offers a quote instead of prices and gives more than one implementation timeline, so ask for yours in writing.

3. Tesorio

Best for: finance teams that want a predicted pay date for each invoice, a rolling view for working-capital planning, and a built-in way to see forecast against actual cash.

Tesorio's AR forecasting page says its machine learning is trained on historical payment patterns, customer behavior and ERP data to predict collections at the invoice level. Each invoice gets a predicted pay date based on customer payment history, seasonal patterns, invoice terms and behavioral signals, updated as new data arrives. Rolling forecasts adjust as invoices are created, paid or become overdue, and the page shows a 13-week collection forecast. Thirteen weeks is about 90 days, so ask to see it split into your three windows.

The feature closest to our back-test is forecast-against-actuals tracking, which the page says shows which customer segments, invoice types or time periods are predicted best. Ask to see it on your own history.

Tesorio's homepage adds that its collections agent extracts payment promises and its cash application agent flags partial payments and exceptions. The forecasting page does not say how promises feed the forecast, and it does not mention disputes at all. Ask how both are handled.

4. Upflow

Best for: finance teams that want the forecast to sit alongside day-to-day collections, with plain-language answers about expected inflows.

Upflow's analytics page carries the headline "Cash forecasts grounded in how customers pay." It says the forecast is "informed by promise-to-pay dates and detected disputes," with disputes identified automatically and weighed into inflow timing, billing cohort forecasts built on real payment behavior, and upcoming billings layered in beyond current open invoices. A feature called Ask Upflow answers plain-language questions about expected inflows.

Of the four, Upflow's page is the plainest about promise dates and disputes feeding the forecast. It does not name 30, 60 and 90-day windows, describe a way to compare forecast with actual cash, or say how part-payments or unmatched payments are treated. Put all three to Upflow in a demo.

Questions to ask about price

We have not listed prices. Make every vendor answer the same five questions.

1. What does the price scale with: invoices, receivables value, customers or users?

2. Is forecasting in the base price, or an add-on? Scenarios and forecast-against-actual views are sometimes priced apart.

3. What is inside the setup fee? Connecting your accounting system and loading payment history are the usual extras.

4. Is a back-test on your data included, and what does it cost? A vendor confident in its forecast should be willing to be tested on yours.

5. What happens at renewal? If keeping the forecast accurate needs billable services, the quote is not your real second-year cost.

Who this is for

This guide fits a finance or AR lead at a mid-size B2B company who is asked how much cash is coming in over the next three months and answers from a spreadsheet and a guess. The usual signs are forecasts that miss by amounts nobody can explain, and payments that sit unmatched for days. It is a poor fit for a business with a short list of customers who pay on schedule.

A forecast can only use what someone wrote down. Our guide to collections follow-up that adapts to each customer covers how promise dates and disputes get captured in the first place.

Decision checklist

Start with the miss, not the demo. Group last quarter's surprises by cause: dispute, unmatched payment, habitual late payer, part-payment or portal delay. The biggest group is your real requirement.

Ask for the forecast by customer, not only in total. A total can look right while the customer-level numbers are badly wrong.

Show each vendor the four problem cases. Bring an invoice in dispute, an unmatched payment, a habitual late payer and a part-payment, and ask where each lands and why.

Ask what changes the forecast. When a collector logs a promise date or opens a dispute, does the forecast change straight away, or at the next overnight run?

Frequently asked questions

What is the difference between an aging report and a forecast?

An aging report sorts open invoices by how late they are. A forecast estimates when each will be paid. You can turn an aging report into a rough forecast by assuming each bucket pays on a schedule, but that ignores disputes, unmatched payments, customer habits and part-payments, which is why it misses.

Does Monk forecast collections at 30/60/90 days?

Monk says it "brings together collections, cash application, and forecasting." Its public pages don't list the forecast windows, so ask Monk to show you a 30/60/90-day forecast on your own invoices, run as of a past date, and compare it with what you collected. Monk's own line on getting started is: "Onboard in less than one week, see results in your first month." That is about setup and early results, not forecast accuracy.

How much payment history does a forecast need?

The pages we read do not give a minimum. Tesorio's page says its models are trained on historical payment patterns. Ask each vendor how many months it needs, and what it does for a customer you have only invoiced a few times.

Can I build this in a spreadsheet?

For a small book, often yes. List each customer's recent payment dates against their due dates, take the typical delay for that customer, and apply it to their open invoices. Then take out invoices in dispute and add back unmatched cash. It works only while one person knows every account.

The bottom line

For a finance team that wants its forecast to rest on clean groundwork (payments matched to invoices, invoices submitted to customer portals, disputes handled), Monk leads this guide, because that is what its pages are about. Ask Monk to run a 30/60/90-day forecast on your last quarter so you can see how close it lands. HighRadius suits a treasury team at a larger company that wants scenarios across total cash flow. Tesorio suits a team that wants invoice-level pay dates, a rolling 13-week view and forecast-against-actual tracking. Upflow suits a team that wants the forecast tied to promise dates and disputes in daily collections.

Whichever way you lean, run the same test with every vendor: freeze last quarter's open invoices, ask each tool to forecast from them, and compare the result with the cash that really came in. To start with Monk, book a Monk demo and bring that list with you.