Touchless AP in wholesale distribution: processing 1,000 invoices/month without losing margin

Wholesale distribution is a volume and thin-margin business. Whoever sells hardware to local stores, food to hospitality, school supplies to schools or healthcare equipment to clinics moves thousands of SKUs and hundreds of suppliers with typical gross margins of 2-5%. A poorly managed margin point on purchases turns directly into lost EBITDA.

A mid-sized distributor's AP processes between 800 and 1,200 invoices per month. Most are recurring, with volume discounts, quarterly or annual rebates, partial deliveries and constant returns. Manual AP does not scale. And, worse, semi-manual AP with a generic OCR that does not understand lines does not scale either: it detects the header total and lets through the real line-level deviations.

This article is for whoever runs administration or procurement at a Spanish SME distributor and is tired of finding out late that margin has slipped away through price increases nobody noticed. Thesis: touchless AP in distribution does work, but only if the system cross-checks line by line against the contract and reconciles rebates separately from the invoice.

Why wholesale distribution is different from the rest of AP

According to interior commerce data published by INE and sector benchmarks compiled by AECOC (Spanish Commercial Coding Association), wholesale commerce moves very high volume with compressed margins. What is relevant for AP is cost composition: 75-90% of a distributor's total cost is cost of goods, bought from 100-300 suppliers with daily or weekly frequency.

The five traits that make this case different:

• High and dispersed volume. 800-1,200 invoices/month in a distributor with two or three branches is usual. Each branch receives daily deliveries and each supplier invoices on their own pace: weekly, fortnightly, monthly.
• 2-5% margins on sales. No cushion. A 1.5% price rise on a high-rotation SKU not detected eats the whole margin on that SKU for the period.
• Catalogues with thousands of SKUs. A hardware distributor handles 8,000-15,000 active SKUs. The agreed price lives in an annual framework agreement with family tariff and negotiated discounts. Nobody reviews each invoice line by line.
• Constant partial deliveries. The supplier ships 60% of the PO today, 30% next week and 10% never arrives. If the invoice arrives before the complete delivery note, amounts do not reconcile and pending adjustments pile up.
• Quarterly and annual rebates. Supplier agreements include accumulated volume discounts: 2% rebate if you buy more than EUR 100,000/quarter, scaling to 3.5% above EUR 200,000. Claiming them on time is admin work many SMEs do poorly or late.

The result is that an admin team of two or three people spends 80-120 hours per month entering invoices and disputing differences, and errors still slip through. AP benchmarking figures from IOFM (Institute of Finance & Management) put the average cost of manually processing an invoice at EUR 12-18 in mid-sized companies. At 1,000 invoices/month, that is EUR 12,000-18,000 per month in capture and reconciliation labour alone.

Classic problems in distribution AP

The table summarises the five most frequent blockages seen by distributors arriving at ininvoice. None is solved with classic OCR or a generic approval workflow.

Problem	Typical frequency	Margin impact
Silent price rises on recurring SKUs	3-7% of lines/quarter	Direct leak, hard to reverse
Partial deliveries billed as complete	10-20% of POs	Overpayment + later manual adjustment
Accumulated rebates unclaimed or late	20-40% of rebate suppliers	Deferred liquidity, sometimes lost
Duplicates across branches	1-3% of invoices	Double payment if not detected in time
Substituted SKUs without price adjustment	3-8% of lines	Payment for the more expensive product

Impact on margin and EBITDA

The math is brutal in its simplicity. A distributor with EUR 6M annual revenue and a 4% gross margin has EUR 240,000 of margin to cover staff, rent, logistics and finance. If undetected billing errors take between 1% and 3% of purchases — a reasonable range per sector data and IOFM's AP benchmarking — the estimated leak is between EUR 50,000 and EUR 150,000/year on annual purchases of EUR 5M. That is between 20% and 60% of gross margin.

For a smaller distributor with annual purchases of EUR 1.5M, the estimated leak sits in the EUR 15,000-60,000/year range. It is still money the business owner is leaving on the table every year without seeing it in any dashboard, because it is hidden inside cost of goods sold.

The silent bias is always in the supplier's favour. When the price rises, they reflect it on the invoice without notice. When delivering partially, they invoice as complete and issue a credit note "when they can". When applying a rebate, they wait for you to claim it. This is not structural bad faith; it is information asymmetry. Touchless AP closes it from the receiver's side.

The touchless flow adapted to distribution

A generic touchless flow does not work here. Six phases must be adapted:

1. Multi-branch ingestion. Each branch or logistics centre has its own inbox or alias. Invoices arrive tagged by cost centre without anyone forwarding anything. Bundled invoices with multiple destinations are split by line.
2. OCR/IDP on invoices with multiple discounts. A typical distribution invoice includes promo discount, prompt-payment discount, family discount, all per line. The extractor must read each applied discount separately and compute the net unit price before comparing.
3. Three-way matching against contract and partial deliveries. The real match is contract/tariff vs delivery note vs invoice. When there are partial deliveries, the system accumulates delivery notes linked to the PO and compares against the invoice only for lines effectively received. Tolerance is applied line by line, not on the header.
4. Risk scoring per supplier. There are stable suppliers and suppliers who change prices every quarter. The system learns which supplier generates more discrepancies and applies stricter controls where needed. Stable suppliers pass touchless with normal tolerance; conflicting ones are reviewed in detail.
5. Routing to procurement or warehouse. A price variance goes to procurement (renegotiate). A quantity variance goes to the warehouse manager (verify what was received). A duplicate goes straight to administration. Not every human resolves every exception.
6. Export to distribution ERP. Clean data flows into the ERP and feeds margin dashboards by family, supplier and branch. Administration stops generating reports by hand and procurement gets real visibility to renegotiate.

This flow puts humans where they add value: procurement renegotiating, warehouse verifying receipts, administration only reviewing exceptions the system could not close.

Specific tolerances by product category

One of the most misunderstood decisions is applying the same tolerance to the whole catalogue. In distribution it does not work. A box of screws does not vary; a box of fruit does.

Reasonable tolerance framework as a starting point (always line by line, never header):

• Perishable products (fresh food, frozen). Quantity tolerance 3-5% for losses and variable weight. Price tolerance stays tight.
• Durable packaged product (hardware, stationery, equipment). Quantity tolerance at most 1% (ideally zero). Units are closed and weight is exact.
• Promotional and high rotation. Price tolerance 2% or EUR 1.50 in OR mode — either side covers. It is the base tolerance we recommend in three-way matching for stable high-volume SKUs.
• Low-rotation SKU with volume-step price. Zero price tolerance. If there is a difference, there is a tariff error or an unapplied step.

These numbers get tuned in the first weeks with real business data. The universal rule: never use the same tolerance for a box of fruit and a box of light bulbs.

Rebates and volume discounts: the special case

Rebates are the block that generates the most discrepancies between billed and actually owed. They work backwards from the normal flow: the supplier bills at standard prices during the quarter and, at close, issues a credit note for the accumulated rebate. If nobody cross-checks accumulated purchases against the agreed tier table, the credit note falls short or never arrives.

Typical cases:

• Monthly rebate on monthly purchases (rare, but exists in some sectors).
• Quarterly rebate scaled by tier: 2% over EUR 50k, 3% over EUR 100k, 4% over EUR 200k.
• Annual rebate with year-end settlement. The most frequent and the easiest to lose.
• Mixed rebate with additional promo bonus by category or brand.

Touchless AP does not solve the rebate automatically, but it does feed the calculation: it keeps accumulated purchases per supplier and family with clean data, and alerts when a tier is crossed or when the received rebate credit note differs from the calculated one. It is an exception to the touchless flow: rebate credit notes are routed to procurement for validation before booking. Without this, distributors typically lose between 5% and 15% of the agreed rebate — not because the supplier denies it, but because nobody claims it in time.

Integration with distribution ERP

AP does not live in isolation. The invoice arrives, gets reconciled, and clean data feeds the ERP that already manages stock, sales and margin. Distributors arriving at ininvoice typically combine:

• Horizontal or vertical distribution ERP with products like Holded, Sage 50/200, Distrito K, Solmicro or sector vertical packages.
• AP automation layer between the supplier inbox and the ERP. ininvoice lives here: ingestion, reading, three-way matching, export.
• BI or dashboard on top of the ERP for margin by family, supplier and branch. The BI drinks from clean data, not from invoices with dragged-in errors.

The idea is not to duplicate masters. Suppliers and catalogue live in the ERP, ininvoice reads them. The AP layer just cleans, matches and exports.

Verifactu and distribution

When food, consumer goods and wholesale suppliers progressively become obliged to Verifactu and B2B e-invoicing, distributors will receive invoices with signed structured data. This changes AP's input quality: less by-eye OCR, more QR and structured XML reading.

For a distributor, the practical effect is twofold: the incoming invoice is easier to process automatically, and at the same time the AEAT receives the same data from the supplier in real time. Any divergence between what your books show and what the supplier has declared becomes detectable. [VERIFY WITH TAX ADVISER] the calendar applicable to your suppliers.

Estimated case: distributor with 100 suppliers

Indicative figures for an SME distributor with 100 active suppliers, 800 invoices/month and two people dedicated to administration.

Metric	Manual AP	Touchless with ininvoice
Hours/month on capture and reconciliation	~100 h	~25 h (exceptions only)
Average invoice → ledger time	10-15 days	<48 h
% discrepancies detected	~35%	>95%
Imputable AP staff cost	~EUR 2,000/month	~EUR 500/month
AP software cost	EUR 0	EUR 249/month
Estimated monthly net saving	—	~EUR 1,250-1,400

This does not include recovered margin. If the distributor captures 1% leak on EUR 5M/year of purchases, that is EUR 50,000 additional typically lost in the "we do not know exactly where the margin goes".

Checklist for the procurement director

Ten concrete actions a procurement director at a distributor can kick off this quarter without touching the ERP:

1. Inventory the real volume: how many invoices/month the company receives, how many active suppliers and how many different SKUs.
2. Identify the top 20 suppliers by amount and the top 20 by invoice count. They usually do not overlap.
3. List suppliers with agreed rebate and review whether the last received credit note reconciles with real accumulated purchases.
4. Measure how much time administration spends on capture vs on discrepancy resolution. The ratio reveals the problem.
5. Audit a sample of 50 invoices from the last quarter at line price: how many have a silent rise vs agreed tariff.
6. Compute current gross margin by family and supplier. The dispersion is usually revealing.
7. Define tolerances by category (perishables, durables, promotionals) before configuring the flow.
8. Confirm with the accounting team or firm how clean data flows into the ERP (Holded, Sage, Distrito K, sector vertical).
9. Define exception routing: what goes to procurement (price), what to warehouse (quantity), what to administration.
10. Pilot with one branch and the top 10 suppliers before extending. Measure KPIs for 4-6 weeks.

Frequently asked questions

Does this work if my suppliers still send poorly structured PDF invoices?

Yes. The IDP engine is trained to extract lines with description, quantity, unit price and per-line discount, even if layout changes between suppliers. When the supplier moves to Verifactu or structured e-invoicing, input quality improves automatically.

What about partial deliveries billed as they arrive?

The system accumulates the delivery notes linked to the PO and compares the invoice against the lines effectively received up to the invoice date. Pending lines stay open and are reconciled with the next invoice. No duplicate payment and no piled-up manual adjustments.

How are quarterly and annual rebates managed?

The system keeps accumulated purchases per supplier and family with clean data. When an agreed rebate tier is crossed, it raises an alert for procurement to claim it. When the credit note arrives, it is matched against the calculation and an exception is raised if it differs. The final figure is validated by procurement before booking.

Do I need a formal PO for every order to every supplier?

Not strictly necessary. What you need is a framework agreement with tariff and family discounts. Matching is done contract vs delivery note vs invoice. A formal PO per line is only common in large long-lead purchases or heavily regulated sectors.

How does the system separate per-line discounts from header discounts?

Line by line, reading each discount applied on the invoice. Each discount is allocated to its line, header discounts are prorated by line amount. The net unit price per line is what is compared to the agreed tariff. Header totals are never compared.

What integrations does it have with distribution ERPs?

ininvoice exports clean data via CSV/API to your ERP. Holded and Sage 50/200 are the most common in the Spanish distribution SME ICP. Integration with Distrito K, Solmicro or other verticals is discussed case by case.

How long does it take to be operational at a distributor?

It starts when you connect the inbox. Extending to the rest of suppliers and branches depends on volume. Most distributors are fully covered in 4-6 weeks.

Does it replace the ERP or accounting?

No. ininvoice is pure AP automation. It is not ERP, accounting or stock management. It sits between the supplier inbox and your ERP. Clean entries are exported to Holded, Sage 50/200, Distrito K or equivalents.

Three things to remember

Wholesale distribution is not just another AP case. With 2-5% gross margins and thousands of active SKUs, it is one of the sectors where manual AP eats most margin silently — and where there is most room for good automation.

1. Manual AP at distributors with 100-300 suppliers eats between 1% and 3% of purchases in undetected errors. On EUR 5M/year that is EUR 50,000-150,000 of lost margin per year.
2. Generic touchless will not do. You need to match line by line against the tariff, manage partial deliveries, compute accumulated rebates and route exceptions to whoever decides in each case.
3. The stack that works is light: supplier inbox → AP automation layer → ERP. The BI feeds the margin dashboard in parallel. No big projects, no six-month consultancy.

If you want to see how your flow looks with one real branch, try ininvoice with your own invoices. You can also check pricing and features in detail.

Related content

• Three-way matching: cross-checking invoice, PO and delivery note line by line
• Touchless accounts payable: what it means and how to measure it
• Managing invoices in distributors: base guide by supplier
• IDP vs classic OCR: why OCR is not enough for distribution