A deduplication method for digital access documents

SAP digital access is priced per document, and the document count is the single largest variable in the settlement. Yet most buyers' first-pass document counts are dramatically inflated — sometimes by a factor of three or more — because the underlying transaction history contains duplicates that the buyer's measurement method has not removed. Cancelled and recreated sales orders. Reversed and reposted invoices. Test transactions that survived the cutover from quality to production. Pipeline triggers that fired twice on the same business event. Each of these patterns inflates the raw document count without representing a distinct billable consumption event. The deduplication method we use across engagements removes the inflation in a defensible way and is documented here.

Why document multiplication happens

The SAP transaction history is a record of every document creation, reversal, cancellation, and amendment. It is not a record of unique business events. A sales order that is created, cancelled, and recreated against the same customer for the same material on the same day generates three documents in the system but only one underlying business event. SAP's standard digital access counting method counts the three documents. The buyer's defensible counting method counts the one event. The difference is the deduplication delta, and across our engagements it averages 18 to 34 per cent of the raw count.

The document multiplication traps article covers the most common inflation patterns. This article focuses on the deduplication method that removes them.

The four deduplication categories

Deduplication runs across four categories. Each requires a different method, and each contributes a different share of the typical delta.

• Cancellation pairs — a document created and subsequently cancelled. Counted once, not twice.
• Reversal pairs — a document posted and subsequently reversed, with a recreated successor. Counted once.
• Test and pre-production residue — documents created in test contexts that landed in the production extract. Excluded.
• Pipeline-replay duplicates — two documents that record the same underlying business event from two integration sources. Counted once.

The aggregate delta varies by SAP module. Sales-document deduplication typically delivers 15 to 25 per cent reduction. Invoice deduplication delivers 8 to 12 per cent. Purchase-document deduplication delivers 12 to 18 per cent. Service-document deduplication delivers 5 to 10 per cent. Manufacturing-document deduplication is highly variable, ranging from negligible to over 30 per cent depending on the shop-floor architecture.

The cancellation-pair method

The cancellation-pair method identifies documents where a creation and cancellation share the same business key — customer, material, posting date, and net value. The pair is treated as a single business event for digital access counting purposes. SAP's audit team typically accepts this method when it is documented in a methodology note, applied consistently across the measurement period, and supported by the underlying transaction reason codes.

The defensive arguments are straightforward. The cancellation entry exists to correct an erroneous creation. The combined pair represents one customer interaction, not two. Counting the pair twice would penalise the buyer for using SAP's standard correction workflow rather than rejecting the transaction at entry. The pattern is documented in the digital access measurement white paper.

Where the method can be challenged

SAP can challenge the method where the cancellation occurs outside a defined window — typically more than 48 hours after the creation — or where the cancellation reason code suggests a substantive business event rather than a correction. The defensive approach is to apply the method within a window (24 to 72 hours) and to exclude cancellations with reason codes indicating customer-initiated returns or post-shipment cancellations, which are arguably separate business events.

The reversal-pair method

The reversal-pair method is similar but applies to documents that have been posted and subsequently reversed, often with a recreated successor that carries a different document number. The method identifies the triple — original, reversal, successor — and counts the triple as one event. The defensive arguments are the same as for cancellation pairs, with the additional point that the SAP system itself ties the reversal to the original through document references.

Reversal pairs are common in financial documents — postings made in error and reversed in the same period — and in manufacturing documents where confirmation reversals are part of the standard production workflow. The reversal-pair deduction typically represents 5 to 15 per cent of the raw count in those modules.

The test-residue method

Test residue is the category of documents that exist in the production extract but were created in a test, training, or sandbox context. The most common sources are quality-to-production cutover errors, training-organisation documents that were not flagged as test, and integration-test documents that flowed through the production landscape during a UAT cycle. The method identifies these documents through the company-code, plant, or organisational-unit assignment, supported by the document creator's user ID and the document creation timestamp.

The defensive position is that test documents are not billable digital-access events because they do not represent business consumption. SAP's audit team typically accepts the exclusion if the test units are documented in the buyer's organisational structure and the exclusion is applied consistently. The pattern is described in the baseline measurement method article.

The pipeline-replay method

The pipeline-replay method addresses the case where the same business event creates two documents because the event flowed through two integration paths. The most common pattern is a third-party order entry system that creates a sales order in SAP, and a downstream integration that also creates a sales order in SAP from the same upstream event. Without deduplication, the buyer pays for both. With deduplication, the buyer pays for one.

The method requires a business-key match across the two documents — customer, material, net value, posting date, and a deterministic identifier from the upstream event. The defensive position is that the second document is a technical artefact of the integration architecture, not a separate business event. The argument is stronger when the integration is documented in the architecture map and the second document carries a reference to the first. The pattern is described in the EDI traffic article.

The methodology note

The deduplication method is only defensible if it is documented. The methodology note has six sections. The deduplication categories applied. The matching criteria used in each category. The time window or reason-code constraints. The sampling approach (full population or sample with confidence interval). The exclusion summary by category. And the documented deltas by SAP module. The note is attached to the digital access position paper and is the document that the SAP audit team reviews.

Without the note, the deduplication argument is harder to sustain in negotiation. With it, the typical settlement runs at 75 to 90 per cent of the deduplicated count, against an opening claim built on the raw count.

The tooling question

The deduplication runs through SQL queries against the SAP transaction tables or through a dedicated digital access measurement tool. Either approach works. The SQL approach is more transparent and defensible because the queries can be inspected by both parties. The tool approach is faster but requires the buyer to be able to demonstrate the underlying logic, which not all vendor tools support clearly. The tooling choices are discussed on the digital access negotiation service page and on the S/4HANA topic page.

If you are about to present a digital access measurement to SAP — or to respond to an opening claim built on SAP's raw count — the deduplication exercise typically delivers a 15 to 30 per cent reduction with no commercial concession on either side. Our standard methodology note is shared with engagement-letter clients and is the work product covered on the digital access negotiation service page.