Insights & Whitepapers
A source-by-source audit of the published evidence for buyer-side (AP) and supplier-side (AR) savings, drawn only from jurisdictions that have already implemented mandates. Every headline figure traced to its origin and graded — and where it does not survive the trace, we say so.
The question. Countries have been running mandatory e-invoicing for a long time — Chile since 2003, Brazil since 2008, Korea since 2011, Mexico universally since 2014, Italy since 2019. Between them they have processed hundreds of billions of invoices. So: what has been published, with sources, about what e-invoicing actually saved the businesses involved? Specifically, separately, for buyers processing payables and suppliers issuing receivables?
The short answer is that almost nothing has. After tracing every figure we could find across Latin America, Europe, Asia-Pacific, the Middle East and Africa, back to its originating document wherever the document could be obtained, the position is this:
On the buyer side, we found no measured, post-implementation study of AP invoice processing cost, receipt-to-approval cycle time, exception rates, query rates or archiving effort attributable to an e-invoicing mandate — in any jurisdiction, at any level of rigour. Not a weak one. Not a single-country one. The category is empty. Every AP figure in circulation is either a projection made before implementation, a survey asking businesses to estimate, a vendor model, or an extrapolation from one of a very small number of consultancy studies.
On the supplier side there is slightly more, but not much, and what exists is mostly about payment infrastructure rather than invoicing cost: Chile's electronic invoice-financing volumes, a single New Zealand agency's payment-time comparison, and one small Peruvian field study of issuing cost. Nothing on days sales outstanding. Nothing on dispute rates. Nothing on late-payment incidence — despite that being among the most common supplier-side claims made for mandates anywhere.
The analysts do not have it either, and that surprised us. Forrester, Gartner, IDC, Hackett, Ardent Partners, Spend Matters and the Big Four are in the business of quantifying precisely this sort of thing, so we checked all ten. There is a great deal of research in the neighbourhood, some of it very good — and not one study across the ten isolates e-invoicing exchange from the accounts-payable automation it sits inside. Section 06 is the firm-by-firm account, including the Forrester study that circulates as e-invoicing evidence and turns out to be about something else.
What is well evidenced is the tax authority's benefit, and it is evidenced to a genuinely high standard: peer-reviewed, quasi-experimental, published in serious journals. That research is real, and Section 07 covers it. But it answers a different question from the one a CFO is asking, and it is routinely presented as though it answers theirs.
One structural observation explains most of the asymmetry, and it is worth stating before any of the detail. Every credible study of e-invoicing measures outcomes visible in tax administration data — declared sales, declared purchases, VAT paid. That is not an accident: it is what tax administrations hold, and tax administrations funded most of the research. Firm-side AP and AR costs are invisible in that data. The gap in the evidence base is therefore structural rather than temporary, and it will not close on its own. It would take a firm-level survey with a control group, and nobody has commissioned one.
What counts as evidence here. We spent August 2026 working through academic databases, tax administrations, supreme audit institutions, multilateral publishers, national ministries and analyst houses, in English, Spanish, Portuguese, Norwegian, Danish, Italian and Korean. One rule governed the whole exercise: a citation must specifically support the claim attached to it. Whenever a secondary source quoted a figure, we went and found the primary document and checked the number was actually in it. Where we could not get hold of the primary document, we say so in the text rather than quietly leaning on the restatement — and that happens more often than it should.
The distinction that governs everything below. Most circulating “e-invoicing ROI” numbers do not measure e-invoicing. They measure full accounts-payable automation or an end-to-end procure-to-pay redesign, of which e-invoicing is one component. This matters enormously to anyone building a business case, because a mandate integration delivers the structured data; it does not by itself change how AP works. Every figure below is therefore labelled e-invoicing only or bundled, and where a source does not make the distinction clear, that itself is recorded as a finding.
| Grade | Means | How to use it |
|---|---|---|
| A | Measured, primary, attributable to a named organisation and a specific retrievable document | Citable as a fact, with its scope stated |
| B | Published by a credible body, but the number is unattributed within it, or modelled rather than measured | Usable with the modelling caveat attached |
| C | Single case, single organisation, small non-random sample, or perception rather than measurement | Illustrative only. Never a benchmark |
| D | Assertion with no traceable basis, or a vendor model presented as a finding | Do not use |
This is the shortest section in the report, and its length is the finding. Everything we could find on buyer-side outcomes from an implemented mandate is in the table below, strongest first. Read down it and notice what is missing: not one entry is a measured before-and-after of AP processing cost.
| Finding | Country | Scope | Grade | What it actually is |
|---|---|---|---|---|
| Receiver saves 3 minutes per invoice | Norway | E-invoicing only | B | Vista Analyse for the Norwegian Tax Directorate, 2025. Revised down from Oslo Economics' 2020 range of 3–5 minutes, explicitly because PDF and paper handling has itself become more automated since 2020. Interview- and model-derived, not measured[1],[2] |
| Receipt cost €0.10–0.40 per e-invoice by firm size | EU-8 | Unstructured e-invoices | B | DG TAXUD invoicing-rules evaluation, 2019, survey of 2,007 businesses across eight member states. The best-sampled European work of its kind — but the per-invoice figures relate to unstructured e-invoices (essentially PDFs), which is not what a mandate requires[3] |
| €8.40 saved per e-invoice received | EU-27, B2G | Bundled | B | European Commission, 2024. The Commission's own wording is savings from “automating the invoicing process”, valued at €46 per labour hour — a time valuation, not an observed cash saving. See Section 05 for the trace[4] |
| 60% of a A$21.69 per-invoice saving falls to AP | Australia | Bundled | D | The 60/40 split is an ATO working assumption applied to a shared sender-and-receiver figure, not a measurement. The underlying study is a non-public 2016 consultancy report. See Section 05[5] |
| Invoicing is 50.6% of corporate VAT compliance cost | South Korea | Pre-mandate baseline | A | Kim & Park (2007), via the World Bank's Korea case study. 67.6% for individual businesses. Establishes that invoicing is the dominant compliance cost — and that the burden is heavier on smaller firms. Says nothing about what e-invoicing then saved[6] |
| Invoices paid in 4.2 days vs 7.1 days | New Zealand | E-invoicing only | C | MBIE, one government agency, Peppol receipt into an unchanged finance system. Unusually clean scope; measurement period not stated. A case study, not a benchmark[7] |
The nearest thing to an authoritative statement on the buyer side is a negative one, and it comes from the European Commission's own evaluation of its own directive, nine years after mandating B2G e-invoicing across all 27 member states: “For the specific objective of lowering operating costs for economic operators, the EU intervention has not been particularly successful… The lack of data does not allow to estimate whether operating costs have increased or decreased after the Directive's implementation.”[8] That is the most honest sentence written about e-invoicing ROI by anyone, and it is buried in a staff working document.
Thin, and weighted toward payment and financing infrastructure rather than the cost of issuing an invoice. The Chilean material is the strongest thing in this report on the commercial side — and it still does not support the causal claim usually made from it.
The best-documented supplier-side benefit anywhere. Chile's platform includes a public register of credit assignment, which makes an e-invoice legally assignable to a factor. The tax authority's own figures: 4,310,911 documents ceded electronically in 2018, up almost 15% year on year, financing more than US$32.8 billion to firms, with over 20,000 companies accessing electronic factoring monthly.[9]
Two caveats that matter. No counterfactual exists — nobody has estimated how much invoice financing Chile would have had without e-invoicing, and Chile separately legislated 30-day payment terms and has a mature non-bank factoring industry. The defensible claim is “this much financing flows through the platform”, not “e-invoicing caused it”. And the most recent official release we could find is January 2019, covering 2018 — anyone quoting Chilean factoring figures today is quoting an eight-year-old data point.
Colombia built the same mechanism (RADIAN) and publishes current data. Invoices registered as negotiable title instruments to 31 December 2025: 1,766,898. Against a national e-invoice volume in the billions.[10]
This is the single most useful check on the “e-invoicing unlocks SME finance” claim. Six years into Colombia's mandate, penetration of the financing register is marginal. Making an invoice electronic does not make it financeable at scale — that requires a separate registry, separate legal machinery, and separate adoption effort.
The only Latin American per-invoice issuing-cost fieldwork we located. Cost per invoice fell from S/2.28 traditional to S/0.83 outsourced electronic, roughly a 64% reduction, decomposed into emission, distribution and storage.[11]
C 111 firms in one region (San Martín), self-reported, no control group, no counterfactual. Genuinely useful as an order of magnitude for the issuing side only — it covers emission, distribution and storage, and nothing on the receiving side. Do not extrapolate to Peru, let alone anywhere else.
Sender saves approximately one minute per invoice, mainly through issuing fewer payment reminders, plus around NOK 8 of distribution cost per paper invoice avoided. Paper invoice cost NOK 25–30; EHF e-invoice transaction around NOK 4.[2],[1]
Note the asymmetry, which is consistent across every Norwegian study since 2008: the receiver gains roughly three times what the sender does. Any business case that splits savings evenly between the two sides is not reflecting the literature.
Chile's tax authority surveyed firms twice, in 2009 and 2013, on the benefits of e-invoicing. In both, cost saving ranked lowest among the benefits firms valued — behind process speed, and behind simply not having to visit a tax office to have invoices stamped.[12]
C These are 1–5 importance ratings, not measured savings, so they cannot be converted into money. But the ranking is a finding in itself, from the world's most mature e-invoicing jurisdiction: firms experienced e-invoicing primarily as friction removal, not as cost reduction.
The Swedish Companies Registration Office found that a minimum of five e-invoices per month makes e-invoicing financially viable for a business.[8]
Rarely quoted and unusually practical. It is the clearest published statement that below a certain volume the arithmetic does not work — which is precisely the population every mandate struggles with, and precisely the population that gets told about per-invoice savings.
So if the measured evidence is that thin, what are all those confident per-invoice figures? We chased the four most widely cited back to where they came from. None survives the journey intact, and two turn out to be worse than they look.
A note on vendor material, because it is not where you would expect. Billentis figures — the source of most European per-invoice claims — are vendor-sponsored cost models rather than measurement studies, which is well known. What is less well known is how far they have travelled into official documents. The Commission's 2010 communication on e-invoicing sourced its market-penetration figure to Billentis and its headline €240 billion savings claim to a Capgemini paper. A UN regional commission distributes a cross-border e-invoicing guide whose filename literally ends _billentis_0.pdf. So “traced to an official source” is not the end of the trace: you have to check what the official source was citing.[19],[20]
This is the first question anyone asks, and fairly so. Forrester, Gartner, IDC, Hackett, Ardent, Spend Matters and the Big Four are in the business of quantifying exactly this sort of thing. Surely one of them has done the work? We went and checked all ten. None of them has.
Let us be precise about the claim, because it is easy to overstate. There is a great deal of analyst research in this neighbourhood, some of it very good. What there is not, anywhere across these ten firms, is a study that isolates e-invoicing exchange from the accounts-payable automation it sits inside. Every quantified benefit number attaches to a bundle. That is not an accident of how the research was funded — it is what these firms are actually paid to measure, which is software categories and business functions, not one component of an integration.
| Firm | The relevant research | What it actually measures | ROI figures? |
|---|---|---|---|
| Ardent Partners | The State of ePayables 2025, Andrew Bartolini, 20th annual edition, underwritten by Bottomline. 204 AP and finance leaders surveyed March–May 2025[21] | The whole AP function. Industry averages: $9.84 all-inclusive cost per invoice, 8.2 days, 18.4% exception rate, 35.4% straight-through | For AP, yes. For e-invoicing, none — mandates are discussed qualitatively, with no figure attached |
| The Hackett Group | Digital World Class® Matrix: 2025 Accounts Payable Provider Perspective, 19 Nov 2025. 15 providers across 17 criteria[22] | AP automation software outcomes: 60% average touchless rate, 59% post-implementation cycle-time improvement, 3.5× productivity above 30% touchless, 79% satisfaction | Outcome figures, yes. Nothing isolating e-invoicing. Their cost-per-invoice quartiles are not public at all |
| Gartner | Magic Quadrant for Accounts Payable Applications — first edition March 2025, current edition 18 June 2026[23] | Vendor positioning within the AP applications market | None. Magic Quadrants are positioning research, not ROI research. No Gartner category dedicated to e-invoicing compliance appears to exist |
| IDC | The Business Value of SAP Business Network — for Buyer Organizations, doc #US52679524, March 2025, sponsored by SAP. Eight organisations interviewed[24] | A whole business network — orders, invoices, forecasting, inventory, partner onboarding. Invoice-specific: 60% faster approval cycles, processing 16.3 to 6.7 days | Yes, and rigorously — 404% three-year ROI. But it cannot be attributed to invoice exchange alone |
| Forrester | Total Economic Impact™ studies, each commissioned by the vendor evaluated. Basware AP Automation (Jan 2024): 158% ROI[25] | AP automation across all invoice formats — paper, EDI, XML, PDF. Network and exchange benefits are not separately quantified | Yes, for AP automation. See below for the study that is widely cited as e-invoicing evidence and is not |
| Spend Matters | SolutionMap. Acquired by The Hackett Group in May 2025[26] | Vendor capability and customer-satisfaction assessment | None of any kind. Spend Matters does not publish quantified ROI research |
| Deloitte | Business benefits of e-invoicing; Finding the Business Benefits in the E-Invoicing Transition (HBR, Oct 2025)[27] | Nothing measurable. Benefits are stated qualitatively and the scope is explicitly bundled with ERP and compliance technology | Zero figures, zero sources. The HBR piece is labelled sponsor content throughout |
| PwC | Invoice automation for finance operations[28] | The whole invoice lifecycle — intake, validation, matching, approval, cash application | Figures given (40–60% less manual processing time, 2–5 days off approval cycles) with no method, no sample and no source, hedged as “may” and “could” |
| EY | E-invoicing Developments Tracker; Tax and Finance Operations Survey 2025 (1,600 respondents, fielded by Oxford Economics)[29] | Mandate tracking; and, in the survey, the tax and finance function as a whole | None. The trackers carry no figures at all. E-invoicing appears once in the survey, in passing, inside a broader category |
| KPMG | E-invoicing & digital reporting services[30] | Service offering | Zero numerical data of any kind. |
Two honest observations before the traces. First, the good research here is genuinely good: IDC's SAP Business Network study names its analysts, publishes a document number, describes its eight interviews and gives its figures in full for free. That is more transparency than most of the sources in Section 05 manage. Its problem is scope, not rigour. Second, every one of these firms sells into this market — the Big Four sell the implementation, Deloitte now sells the software too, and Forrester's TEI studies are commissioned by the vendor being evaluated. None of that disqualifies the work. It does mean the scope questions are worth asking twice.
If you do use the Ardent benchmarks, use them properly — there are two 2025 editions and they disagree. The State of ePayables 2025, underwritten by Bottomline and fielded March–May 2025, gives $9.84 per invoice, 8.2 days and an 18.4% exception rate from 204 respondents. Ardent Partners' Accounts Payable Metrics That Matter in 2025, sponsored by Pagero, gives $9.40, 9.2 days and a 14% exception rate from 212.[21],[35] Same firm, same year on the cover. The reconciliation is in the fielding dates: the Pagero edition says 2025 but was fielded March–May 2024, so it is the older wave. Cite the edition and the fielding period, not just “Ardent Partners, 2025”, or your reader will find the other one and conclude you got it wrong. Useful footnote: the Pagero edition is the one that actually defines all-inclusive cost — “receipt, processing, and approval, as well as salaries, benefits, technology, overhead” — a definition the State of ePayables edition omits.
One thing we could not check, which is itself worth saying. We wanted to quote Gartner's own market definition for the AP Applications Magic Quadrant, because how a market is defined tells you what the analysts think belongs in it. We could not get it. The report is paywalled, reachable only through vendor reprint pages that publish the vendor's positioning rather than Gartner's framework. So we are not going to characterise it. It does mean that the single most influential map of this software market is one that neither we nor most of its subjects' customers can actually read.[23]
The research quality here is genuinely high: peer-reviewed, quasi-experimental, published in journals that reject most of what they receive. It is also, without exception, about the money the state collects rather than the money a business saves. Both facts deserve to be stated together.
| Country | Finding | Design | Grade | Source |
|---|---|---|---|---|
| Peru | Reported taxable sales +6.6–7.4%, purchases +4.5–5.6%, VAT liabilities +8.1% in year one | Difference-in-differences on staggered mandate waves, 78,000 firms, >80% of domestic VAT | A | IMF WP/19/231; Journal of Public Economics 2022[36] |
| Peru | Spillover: unmandated small firms whose partner was mandated reported +11% sales, +17% VAT | Network analysis on the same reform | A | IMF WP/2022/057[37] |
| Peru | Independent replication on newer data: effective VAT payments +5–8% in year one | ~200,000 firms | A | Consejo Fiscal del Perú, Nov 2025[38] |
| Italy | “E-invoicing decreased the Italian VAT loss in 2019 by about €2.2 billion to €2.6 billion compared to 2018” | DiD using reverse-charge products as controls; trade data gap on intra-EU imports | A | Heinemann & Stiller, Int. Tax & Public Finance 32(1)[39] |
| Rwanda | E-invoicing alone: +US$10,913 net VAT per firm. Audit alone: +US$26,843. Combined: +US$33,372 | Universe of Rwandan tax filings, 2012–2019 | A | Journal of Development Economics 172 (2025)[40] |
| Argentina, Ecuador, Uruguay, Mexico | Positive VAT effects in all four: Argentina taxable sales 0 to +10%; Ecuador VAT +17.7% (2015) and +24.7–28.1% (2016); Uruguay VAT payments +3.7%; Mexico tax accrued +11 to +16% | DiD, propensity matching, regression discontinuity | A | IDB discussion papers 561–563; ITESM for SAT[41] |
Three qualifications the authors themselves draw, which almost never travel with these figures.
Peru's authors say the effect is reporting, not production. Measured value-added grew around 15% among adopters while employment grew about 2%. Their conclusion is that the growth “is likely to capture an increase in the share of output that is reported”. Anyone citing “+7% sales” as a commercial benefit of e-invoicing has inverted the paper's meaning. The same paper notes that e-invoicing enables lower administrative and compliance costs for firms — and provides no numerical estimate of any firm-side cost reduction. The most rigorous e-invoicing paper in the literature does not quantify the taxpayer's savings.[36]
Rwanda's authors found e-invoicing alone to be a weak instrument. The working-paper version puts it plainly: “while e-invoicing adoption per se has increased firms' net VAT payments, this impact is quantitatively limited, as firms seem to re-adjust their expenses so to keep VAT payments low.” Firms offset sales-side transparency by inflating the input side. The paper's real finding is that e-invoicing's value is as an enabler of effective audit — the only audits that significantly improved compliance were those on firms already using e-invoicing. Worth knowing: that sentence was removed from the published journal abstract, which reads more positively. If you quote it, cite the working paper.[40]
Italy's paper measures cross-border only. The €2.2–2.6bn is a reduction in VAT loss, extrapolated from a trade-data-gap effect on intra-EU imports, over twelve months. It makes no claim about domestic VAT compliance. The headline 7% gap reduction is the baseline estimate; the paper's own weighted specifications give 4.8–5.6%.[39]
And a number worth sitting with. The European Commission's own impact assessment for VAT in the Digital Age models total benefits of €371.9 billion over 2023–2032. Of that, €335.6 billion is VAT collection and C-efficiency, and the line item labelled “E-invoicing benefits” is €5.6 billion — 1.5% of the total. Set against €79.1 billion of modelled administrative burden falling on businesses.[42] Read plainly, Europe's own business case for ViDA is a revenue case, not a business-efficiency case. That is a legitimate basis for public policy. It is a very different proposition from the one businesses are usually sold.
Findings that cut against the standard narrative, all from primary sources, none of which appears in vendor material. A business case that anticipates these is stronger than one that gets ambushed by them.
Chile reached 99% of invoices electronic by April 2017 and universal mandate in February 2018. Its tax authority's own 2025 study puts VAT non-compliance at 18.9% for 2018–2020, equal to 1.9% of GDP, with a corporate income tax gap of 46.4%.[43]
Near-universal e-invoicing coexisting with a ~19% VAT gap is the strongest single corrective to “e-invoicing closes the VAT gap”. Note too that Chile has never had its mandate econometrically evaluated for revenue impact — the IDB/CIAT synthesis explicitly excludes it. The country most cited as the success story is the one with no impact evaluation.
5% of firms use ghost invoices annually, and among those firms ghost transactions account for 14% of purchases — concentrated among larger firms. Published in AER: Insights, using the transaction-level data that e-invoicing itself generates.[44]
Structured, validated, real-time invoice data does not equal genuine transactions. E-invoicing makes fraud visible; it does not make it impossible.
Korea has the world's longest-running B2B mandate. Its tax authority reports a 900 billion KRW annual reduction in invoice preparation, storage and filing cost.[45] But national taxpayer compliance cost measured by the Korea Institute of Public Finance reached 15.44 trillion KRW for 2021, 0.73% of GDP, and a domestic academic argument put to the Korean Tax Association is that costs rose from around 7 trillion KRW in 2007 to 15 trillion in 2022 while the state's own cost of collection fell.[46]
The contention is that digitalisation transferred cost from the administration onto taxpayers, concentrated on small firms. Korea's continuing per-invoice tax credit for individual businesses is itself evidence that someone priced that burden and thought it needed compensating.
The most important methodological caveat we found anywhere. Oslo Economics' 2020 report for the Norwegian government states that earlier studies overestimated gains by assuming fully manual invoice handling, when non-electronic invoices are already largely processed digitally via scanning — and that further accounting automation will reduce the marginal per-invoice gain from e-invoicing over time.[2]
Vista Analyse acted on it in 2025, revising the receiver's time saving down from 3–5 minutes to 3, explicitly citing improvements in PDF and paper handling since 2020.[1] This undermines the entire class of “paper costs €11, e-invoice costs €1” comparisons: the baseline they measure against no longer exists.
Poland's NIK reported in July 2025, after auditing January 2022 to June 2024, that KSeF's problems were apparent in 2023 and that the Finance Minister “did not take effective action”; an external audit showed an absence of effective oversight. Meanwhile average enforcement recovery per control fell threefold, and VAT audits fell 30.8% in 2023.[47]
Supreme audit institutions are unusually candid and essentially never quoted in e-invoicing marketing. We looked for equivalents in Denmark, Sweden and Italy and found none — not because the programmes were flawless, but because nobody audited them.
The 2025 APEC report's own survey — as distinct from its modelled benefits — found that 79% of respondents said implementation costs exceeded expectations, with large-enterprise costs above USD 20,000.[13]
A useful irony: the cost findings in that report are better sourced than the benefit findings, because they come from its own survey rather than from modelling. They are also the part nobody quotes.
What the evidence — as opposed to the marketing — actually supports, and what a business should do about it.
Two decades of mandates, hundreds of billions of invoices, and not one before-and-after study of AP processing cost, cycle time or exception rates attributable to an e-invoicing mandate. Every AP figure in circulation is a projection, an elicited estimate, or a vendor model. This is not an accusation against any particular number — it is a statement about a category that turns out to be empty.
And before anyone assumes the research houses have quietly done this work and put it behind a paywall: we checked all ten of them, and they have not. What they have measured, carefully and at length, is the accounts-payable function and the software that automates it. That is a different thing, and none of them claims otherwise — it is the retelling that blurs the two.
The Commission's €5.28/€8.40 is explicitly for “automating the invoicing process” at an assumed €46/hour. The ATO's A$30.87-to-A$9.18 gap is explicitly a shared sender-and-receiver figure with an assumed 60/40 split. The €25–65 figure is explicitly for the whole order-to-pay cycle. In every case the qualification is in the source document and absent from the quotation. If you take one thing from this report: read the sentence the number sits in, not the number.
Norway has measured this repeatedly since 2008 and keeps landing in the same place: about one minute for the sender, about three for the receiver. The ATO independently assumes 60/40. Every serious source points the same way. A business case that splits benefits evenly between AP and AR, or that leads with supplier-side savings, is not reflecting anything published.
The strongest per-invoice savings claims compare structured e-invoicing against fully manual paper handling. That comparison is increasingly fictional: PDF and scanned-paper processing is itself now substantially automated, which is why Norway revised its own numbers down between 2020 and 2025. The later you implement, the smaller your incremental gain — and the more of your business case has to rest on compliance rather than efficiency.
Peru, Italy, Rwanda, Argentina, Ecuador, Uruguay and Mexico all have credible causal evidence of increased tax compliance. That research is good, and this report does not diminish it. But it measures declared sales, declared purchases and VAT paid — the state's revenue, not the taxpayer's cost. Europe's own ViDA impact assessment puts 90% of modelled benefit in VAT collection and 1.5% in e-invoicing itself. Presenting compliance research as business ROI is the single most common category error in this field — and the second most common, close behind, is presenting AP-automation research as e-invoicing research.
Chile's electronic factoring is real and large, but has no counterfactual and its most recent official data is from 2018. Colombia built the same mechanism, publishes current figures, and shows penetration in the low millions of invoices against national volumes in the billions. Making an invoice electronic does not make it financeable. If SME finance is in your business case, it needs its own evidence.
Given the state of the published evidence, the only defensible per-invoice figures in your business case are yours. Before implementation, capture: cost per invoice processed on the AP side, receipt-to-approval cycle time, exception rate and its causes, cost to issue on the AR side, and time spent on tax reporting and audit preparation. Then measure the same things afterwards.
This is not a counsel of despair — it is what the evidence gap actually implies, and it has a practical advantage. A CFO will trust a smaller number you measured over a larger one you imported, and a mandate is a rare opportunity to instrument a process that most organisations have never baselined at all.
The honest framing for most compliance-driven programmes is that you are buying the right to keep trading in a market, not a return. The direct savings are real but they are unlocked by the integration and only banked if you also redesign the process — which is a separate programme with separate cost, aimed at a different part of the organisation. Conflating the two is how e-invoicing business cases end up promising a return the integration was never going to deliver.
Source-type tags: [official] government / EU / multilateral primary source · [study] peer-reviewed or working-paper research · [press] news reporting · [vendor] supplier or sponsored-analyst material. Where we could not retrieve a primary document, that is stated in the entry.