Goods Received Note Template: Match Deliveries to Orders With AI

A goods received note should record the supplier, date and time, the order and delivery note numbers, each item with the quantity ordered, delivered and accepted, its condition, and who received it. AI can read a photo of the delivery note, match it to the purchase order and flag short, over and damaged lines, but a person must count and inspect first.

The note protects you in two directions. It stops you paying for goods that never arrived, and it gives you dated evidence when you ask a supplier to credit a damaged or short delivery. The template and routine below suit a small business that receives a handful of deliveries a day, with no warehouse system.

Follow me on Instagram@sagnikteaches

What a goods received note has to hold

Keep the note to one page. A receiving clerk standing at a loading door with a pallet behind them will not complete a long form, so each field needs a reason to be there. The table lists the ones worth keeping.

Connect on LinkedInSagnik Bhattacharya

The line columns "ordered", "delivered" and "accepted" carry most of the value. Delivered is what the delivery note claims or what you counted; accepted is what you are willing to pay for. A line with 10 ordered, 10 delivered and 8 accepted has told you a story in four numbers.

Subscribe on YouTube@sagnikteaches

The three documents and which one wins

Every delivery generates three pieces of paper, and they disagree more often than people expect.

1. Purchase order (PO). Your request: items, quantities, agreed prices.
2. Delivery note. The supplier's statement of what they put on the vehicle. Quantities here are not proof that they arrived.
3. Invoice. The supplier's demand for payment, which should agree with the order and with what you received.

Matching all three is called a three-way match. The goods received note is the fourth document that makes the match meaningful, because it is the only one that records what was physically at your door. When the invoice arrives, you pay for the accepted quantities at the PO price, not the delivery-note quantities. If you want the invoice side automated as well, see matching supplier invoices to purchase orders automatically; this tutorial stops at the door.

Set up the sheet once

A spreadsheet is enough to start. Make two tabs. The first, "GRN header", has one row per delivery: GRN number, date, time, supplier, PO number, delivery note number, received by, outcome. The second, "GRN lines", has one row per item per delivery: GRN number, PO line, item, unit, ordered, delivered, accepted, condition, batch or use-by, note, photo reference.

Keep your open purchase orders on a third tab in the same layout (PO number, line, item, unit, ordered quantity, agreed price, expected date). The match is then a comparison of two tables with the same columns, which is the kind of job an assistant does well and which you can check with a formula.

Add data validation on the unit column ("each", "case of 6", "kg", "litre") and on the outcome column. Unit of measure is where most mismatches start, as the worked example shows.

Step 1: photograph the delivery note properly

The quality of the photo decides the quality of the extraction. Teach everyone who receives goods the same four habits:

• lay the note flat on a surface, not held in the air or folded over a clipboard;
• photograph each page separately, with the whole page in frame and the corners visible;
• make sure any handwritten changes, stamps and signatures are in frame, because a driver's amendment is part of the document;
• name or tag the photos with the GRN number straight away.

If your phone has a document-scan feature in its files or notes app, use it; it straightens the page and sharpens text. For a printed note with small figures, scan at full resolution rather than a screenshot of a zoomed-out photo.

Step 2: turn the photo into a table

Use a business plan of an assistant that accepts images, and attach the photos. This prompt gives you a table you can paste into the "GRN lines" tab:

This is a photo of a supplier delivery note. Extract every line into a table with columns: line number, item description as printed, supplier code if shown, unit as printed, quantity delivered as printed. Then list separately: delivery note number, date, supplier name, and any handwritten notes, stamps or amendments you can see, with their position on the page. Rules: copy numbers exactly; if a digit is unclear write UNCLEAR and say which; do not correct spelling; do not add items; do not match to any order yet.

An illustrative reply for a catering company's dry-goods delivery:

Delivery note: DN 88170, dated 09/02/2027, Supplier: [name as printed]
Line | Description | Code | Unit | Qty
1 | Plain flour 16kg | FL16 | bag | 10
2 | Basmati rice 10kg | RC10 | bag | 6
3 | Olive oil 5L | OL5 | case of 4 | 2
4 | Tinned tomatoes 2.5kg | TT25 | case of 6 | 8
Handwritten: next to line 3, "1 short" with initials, bottom right.
UNCLEAR: line 2 quantity could be 6 or 8 (smudged).

Notice the two things the assistant surfaced without being asked to guess: a handwritten amendment by the driver, and a smudged figure. Those are exactly the details a quick glance at the paper would miss, and exactly why the extraction rules say "UNCLEAR" instead of "best guess".

Step 3: match the table to the purchase order

Next, give the assistant the extracted lines and the PO lines, and ask for a comparison. Keep this as a separate step so you can see what each stage produced.

Compare the delivery lines with the purchase order lines below. For each PO line report: item, unit ordered, quantity ordered, quantity on the delivery note, difference, and a flag from this list only: MATCH, SHORT, OVER, UNIT MISMATCH, NOT ORDERED, NOT DELIVERED, UNCLEAR. Match on item description and supplier code, not on line order. If the unit on the delivery note differs from the PO (for example "case of 6" versus "each"), flag UNIT MISMATCH and do not convert. Finish with a one-line summary of the number of lines flagged. Do not decide whether to accept anything.

The "do not decide" line matters. The assistant's job is to compare; accepting or rejecting is a decision for the person at the door. An illustrative result:

Two of the five lines need no more thought from the buyer. Three need a human: the smudged rice figure, which you will settle by counting; the olive oil shortage, which the driver has already admitted in writing; and the sunflower oil, which is missing from the note entirely and might be on a later delivery.

Step 4: the check at the door, before anything is signed

This is where the extraction ends and the physical work begins. The person receiving does the following, using the flagged table on a phone or a printout:

1. Count against the table, not against the note. Count cases, bags or pallets as they come off the vehicle and write the count in the "delivered" column.
2. Inspect each line. Look for torn or crushed packaging, leaks, broken seals, wrong sizes and, where relevant, use-by dates that are shorter than your standard.
3. Take temperatures of chilled and frozen goods with a probe and record them; reject outside your limits. Use the limits in your own food-safety plan.
4. Photograph any problem with the item, label and surrounding packaging in the frame.
5. Write the exception on the driver's copy before signing: "2 bags torn, 1 case olive oil not delivered, received subject to count." The driver signs it as well.
6. Set the accepted quantity line by line and the outcome for the delivery as a whole.

If a delivery is too big to count at the door, sign "received subject to count" and finish the count within a stated time, such as 24 hours. Photograph the stack as it arrives to show its condition at the time.

Who signs, and what the signature means

A signature on a goods received note says "I counted and inspected these goods and this is what I found". It does not say "the AI agreed". Write that on the template if it helps. Set two rules:

• The signer is the person who physically counted. A manager who was not at the door should not sign for goods they did not see.
• Above a value you set, say anything over the cost of a normal day's purchases, or for goods that need a safety check such as chilled meat, a second person checks and initials too.

Keep a short list of who is authorised to receive, and brief temporary staff before they take a delivery alone. A new starter who signs a driver's tablet without looking is a common reason a credit claim fails, because the signature says everything arrived.

The email to the supplier, drafted from the flags

With the flagged table in front of you, an assistant can draft the shortage and damage message in under a minute. Give it the GRN, the flags and the photos' descriptions, and ask it to keep to the facts.

Draft a short email to a supplier about delivery DN 88170 (our PO-2231, received 9 Feb 2027 at 07:40, our GRN-0412). Facts: olive oil, 3 cases ordered, 2 delivered, driver noted "1 short" on the note; sunflower oil 20L, 2 ordered, none delivered; 2 bags of flour torn, photos attached. Ask them to (1) credit or deliver the olive oil, (2) confirm when the sunflower oil will arrive, (3) credit the 2 torn bags. Polite, 100 words, no blame, one clear ask per item.

Before and after, in miniature: the vague version ("some items were missing from our last delivery, please sort") invites a slow reply. The specific version, with PO, delivery note and GRN numbers and one ask per item, tends to be answered faster because the supplier's office can act on it without a phone call. For damage that is the carrier's fault, handling damage claims and disputes covers the evidence a claim needs.

Worked example: a catering company's Tuesday delivery

An illustration with invented figures. A catering company with 12 staff receives a weekly dry-goods and chilled delivery for the week's events. The purchase order has 14 lines. The delivery arrives at 07:40, and the kitchen manager has 20 minutes before the driver's next drop.

1. 07:40. The kitchen manager photographs both pages of the delivery note and drops them into the assistant with the extraction prompt. It returns 13 lines (one PO line is not on the note) plus a handwritten "1 short" beside the olive oil.
2. 07:44. The comparison prompt returns: 10 MATCH, 1 SHORT, 1 UNIT MISMATCH, 1 NOT DELIVERED, 1 UNCLEAR.
3. 07:45 to 07:58. While the driver unloads, the manager counts against the table. The UNCLEAR rice is 6, matching the order. The UNIT MISMATCH is lemons: ordered by the kilogram, delivered as "1 case", which turns out to hold 10 kg. Accepted. The chilled items read 2.8 and 3.4 degrees C on the probe, inside the company's limit of 5.
4. 07:58. The manager notes "received subject to count" on the driver's copy for the frozen items, which will be counted after the freezer is loaded, and signs.
5. 09:30. The frozen count finds one box of puff pastry damaged. A photo is added, the line is accepted at one fewer, and the supplier email is drafted from the flags.

The delivery took 18 minutes at the door and about 8 more for the paperwork. The company ends the morning with a record that shows four exceptions and what was done about each, ready to compare with the invoice when it arrives.

A unit mismatch that cost a week's margin

An illustrative mistake: a purchase order asked for "cling film, 6", meaning six rolls, and the supplier's delivery note showed "6" under the unit "cases of 12". The invoice was for six cases. Because the receiving clerk had counted six boxes arriving and signed, the match looked clean, and the invoice was paid. The company then held 72 rolls of cling film it did not need, at roughly twelve times the expected cost.

A unit check would have caught it. The comparison prompt above flags UNIT MISMATCH whenever the delivery note's unit differs from the order's, and a data-validation list on the unit column in the PO sheet makes the buyer state the unit when ordering. Ask the assistant to flag any quantity more than double the ordered quantity as well; it costs nothing and catches both unit errors and keying errors.

Partial deliveries, back orders and substitutions

Not every gap is a shortage. Suppliers split deliveries, back-order items and substitute products, and each needs a different entry on the note so the order does not stay open forever or get paid twice.

• Partial delivery. Record what arrived against the PO line and leave the rest "open" with the date the supplier promised it. When the second part comes, create a second GRN that references the same PO line, so the two add up to the order.
• Back order. The supplier confirms they cannot supply an item yet. Mark the line "back-ordered" with the date, and ask the assistant to list every open back-ordered line each Friday, so a missing item does not wait unnoticed until an event needs it.
• Substitution. A different brand or pack size arrives in place of the one ordered. Never accept it silently. Flag it NOT ORDERED on the table, photograph the label and ask the buyer, who may accept it, ask for the original or reject it. Record the decision on the note, because the invoice will carry the substitute's price.
• Over-delivery. More arrives than you ordered. Accept up to the ordered quantity unless the buyer says otherwise, and note the excess as "held for collection" or "rejected", so you do not pay for it by accident.

The status of every open PO line is easier to follow if your orders also go out in a consistent format. Automating purchase orders and supplier emails shows a way to raise them that leaves a clean record to match against.

Paper-only receiving next to the photo-and-match routine

Before: the driver hands over a delivery note; whoever is nearest signs it; the note goes into a tray; at the end of the month someone searches the tray for the one that matches an invoice the accountant is asking about. A missing case of oil is noticed on the day the kitchen runs out of it, nine days after the delivery, when the supplier's office says it was delivered because the signed note says so.

After: the note is photographed at the door, compared with the order within minutes, counted against the flagged lines and annotated before signature. The shortage is reported the same morning with the delivery note number and the GRN number. The same-day email is what makes a credit straightforward to get, because the supplier can still check their own vehicle and stock.

What to do at month-end with the notes

Once a month, ask the assistant to compare the GRN lines with the invoices you have received. Two lists are useful:

• Received but not yet invoiced. Goods in your cupboard for which no bill has arrived. They belong in your month-end figures as an obligation, and your accountant will want the list.
• Invoiced but not received. Bills for items with no matching note. Do not pay these until you have asked the supplier what they are.

Paste both tables and ask for the two lists with the working shown; spot-check at least five rows in each against the paper. If your count and your system keep disagreeing, finding why your count and your system disagree goes further into the causes that begin at the receiving door.

Checking the routine is working

After a month, look at three numbers from the sheet: the share of deliveries with at least one exception, the share of exceptions that received a supplier credit or a replacement, and the average time from delivery to the supplier being told. If few exceptions ever appear, you may be signing too quickly; a small business with several deliveries a week should see some. If many exceptions never get credited, the email step is the weak link, not the receiving.

It also helps to review the extraction itself. Each week, compare two or three photographed notes with the extracted tables, line by line. If the same type of error recurs (a particular supplier's font, say, or a column the assistant mistakes for the quantity), add one sentence about that supplier to the prompt.

If you would like help setting up the photo-to-table routine around your own suppliers, that is what my AI implementation consultation covers. Start with the two or three suppliers that deliver most often and the ones that most often get things wrong, and expand from there.

Goods received note questions from small teams

What is the difference between a delivery note and a goods received note?

The supplier writes the delivery note to say what they sent. Your team writes the goods received note to record what actually arrived and in what condition. The two should agree, and where they do not, the received note is your evidence for a shortage or damage claim.

Can AI count the boxes for me?

No. It reads the paperwork and compares figures. Someone at the door must count and inspect, then enter what they found. AI is useful for turning a delivery note photo into a table and checking it against the order, not for confirming that the goods exist.

Who should sign the goods received note?

The person who physically counted and inspected the delivery, and a second person for high-value or high-risk deliveries. Signing 'received subject to count' on the driver's copy is sensible when the count cannot be finished at the door.

How long should I keep goods received notes?

Keep them at least as long as the matching invoices, since they support payment, and longer if your accountant, insurer or a customer contract asks for it. Record retention rules vary, so check your own requirements and store the notes where the invoice can be found with them.

Further reads

• AI Supplier Management: Track Prices, Lead Times, and Risk — Track which suppliers keep delivering short or late.
• Cold Chain Temperature Logs: AI Flags Breaches in Logger Exports — For chilled deliveries, check the temperature record as well.
• AI in Warehouse Operations: Trace Pick Errors Back to Their Cause — Trace errors that start at receiving back to their cause.
• Stocky Alternative for Shopify: Purchase Orders and AI Reordering — Purchase orders and reorders for an online shop.
• Supply Chain AI Software: What a Small Importer Needs First — A buying checklist for small importers covering data, lead times, supplier emails and stock sync, with one worked reorder to test any vendor against.
• Joinery AI: Build a Cutting List and Material Order From a Drawing — Extract every part from a joinery drawing into a sheet, audit it against the drawing, add waste by material and turn the totals into a supplier order.
• Product Data Enrichment for Industrial Catalogues: AI Fills Gaps — Define the attributes buyers filter on, extract them from datasheets with page evidence and flags, and audit a risk-weighted sample first.
• Proforma Invoice vs Purchase Order: Let AI Catch the Mismatches — A line-by-line check of a proforma invoice against your purchase order, with AI flagging price, quantity, terms, Incoterm and bank-detail changes.
• Purchase Order Tracking: Build a Supplier Chase Sheet With AI — One open-PO sheet, formulas for days late and lateness bands, and AI-drafted chase emails for each band plus a weekly routine.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: standard receiving and three-way-match practice; illustrative figures and a generic catering company throughout. No vendor pricing is quoted.