v1.0 · August 2026

What a Greek
online order
actually costs

Market structure from the public record, merchant economics from three anonymised order books.

Download the PDF
01

What this is

Greek e-commerce is measured in aggregate by ELSTAT and Eurostat, and in behaviour by association surveys. What nobody publishes is the merchant's side of a single order: what it costs to win it, ship it, get paid for it, and absorb the ones that come back.

This paper assembles the public record on market size, payment and delivery, then opens the order books of three Greek merchants to show what those aggregates look like from behind the counter.

The merchants are anonymised. Three Greek e-commerce operations, referred to as Merchant A, B and C, on two different platforms. Business names, domains, product categories, platform names and the marketplace are withheld, because in a market this size any two of those together identify a business.

Codes prevent casual identification and search matching. They do not defeat a reader who already knows what the author operates. Merchant-derived findings should be read as coming from an interested party who has chosen not to be specific about which businesses.

Finding 01

The market on the public record

The 2025 ELSTAT survey, opened at source, puts turnover from electronic orders at €36.1bn: 9.5% of the total turnover of firms with ten or more employees, and inclusive of EDI and business-to-business sales. It is not the consumer market.

Finding 02

Cash on delivery disagrees with itself

Published sources put COD at 62.8% and at about 21%. The gap is probably two different bases, people versus orders, not two estimates. Neither publisher states which.

Finding 03

Delivery is leaving the home

Parcel lockers grew 36.1% in a year to roughly 5,300 units nationally, while courier-to-home remains the default at 78.7%.

Finding 04

Orders that never complete

On one merchant's book the cancellation rate tripled: 5.7% in 2022 to 18.6% in 2025. Nearly one order in five.

02

The market on the public record

The figure everyone quotes is €36.1bn, from ELSTAT, described as roughly 9.5% of Greek business turnover. Opening the survey series changes what that number means.

It counts turnover from electronic orders, and that indicator includes EDI and business-to-business sales: confirmed in the methodology of the same ELSTAT series. It is not a measure of consumer e-commerce. It covers only enterprises with ten or more employees, which excludes most of the Greek e-shop population, and the reference period runs 1 January to 15 September, not a full calendar year. The 9.5% denominator is the total turnover of those same larger enterprises, not of all Greek business.

The honest reading is therefore narrower than the headline: among Greek enterprises with ten or more employees, about 9.5% of turnover over an eight-and-a-half-month window arrived through electronic orders of all kinds, business-to-business included. That is a real and useful statistic. It is not the size of the Greek consumer e-commerce market, and it is routinely reported as though it were.

Status of these figures: opened at source. The ELSTAT release resisted four retrieval attempts across earlier drafts and was carried as press-reported. On 6 August 2026 it was retrieved directly: the announcement of 17 December 2025, reference period 1 January to 15 September 20251. Every figure in this section now comes from that document rather than from reporting of it, and the totals are confirmed exactly: €380,787m total turnover, €36,122m from electronic orders, 47,849 enterprises.

The release also settles the composition question this section was built on. Of the enterprises receiving electronic orders, 11,700 received them via a website or application and 647 via EDI-type messages: 12,347 in total, the 25.8%. The turnover figure is not split between the two channels, so €36.1bn remains a combined web-and-EDI number and cannot be reduced to a web-only one. The reading in this section is confirmed, not merely inferred.

One caveat arrives with the primary document. ELSTAT states that the 2025 results are not directly comparable with 2024, because the statistical unit changed from the legal unit to the enterprise. Any year-on-year growth computed across that boundary, and several published commentaries do compute one, is measuring a definitional change as well as a market. No growth rate is computed from this source anywhere in this paper.

On the demand side, Eurostat puts the EU average at 78% of internet users having bought or ordered online in the last twelve months, on a 2025 reference year. No Greek figure appears in that publication. An earlier version of this section claimed Greece sits below a roughly 75% EU average; both halves of that were wrong: the EU figure is 78%, and Greece was not in the source at all. A Greek penetration figure will be taken from the Eurostat dataset directly or left out.

03

Many sellers, half the revenue

Greece ranks eleventh in the European Union for the share of companies that sell online, and twenty-seventh for what that selling is worth. The same country, the same survey, the same year, sixteen places apart. That gap is the most important structural fact about Greek e-commerce, and it is almost never reported.

The figures come from Eurostat's harmonised enterprise survey, the European instrument of which the ELSTAT release discussed above is the Greek arm. Every member state asks the same question of the same population, enterprises employing ten or more people, which makes the comparison usable in a way that national press coverage of national statistics is not.

2,3
Two rankings of the same 28 territories, 2025. Eurostat isoc_ec_eseln2 and isoc_ec_evaln2, enterprises with 10 or more persons employed.
GreeceEU27Greek rank
Enterprises selling via a website or app24.5%20.7%11 of 28
E-commerce as a share of total turnover9.5%19.5%27 of 28

Greek firms are more likely than the European average to sell online, and the selling produces less than half the European revenue share. Only Cyprus scores lower on the value measure. On the participation measure Greece is ahead of Germany, the Netherlands, Czechia and Estonia.

Eleventh for how many sell. Twenty-seventh for what it is worth. Greek e-commerce is unusually wide and unusually shallow.

This is not a one-year artefact. Both series run consistently in the same direction across the whole published window.

2,3
Greek and European e-commerce, 2021 to 2025. Eurostat, enterprises with 10 or more persons employed.
Measure20212022202320242025
Greek enterprises selling via web20.017.319.622.824.5
EU enterprises selling via web19.019.419.720.620.7
Greek e-commerce, % of turnover5.86.97.18.49.5
EU e-commerce, % of turnover19.817.618.319.119.5

Greece has been above the European average on participation since 2021 and has never been within nine points of it on value. The good news in the table is real: Greek e-commerce turnover share has risen from 5.8% to 9.5% in four years, a faster relative gain than the European average, which has essentially been flat. The gap is closing. At the rate of the last four years it closes in roughly a decade.

Why the two measures disagree

Three structural facts, all from the same survey, account for most of the divergence.

One: Greek e-commerce is consumer business, European e-commerce is trade business. Eurostat splits the value of web sales into B2C and B2B/B2G, and the two economies are close to mirror images.

3
Composition of web-sales value, 2025. Eurostat isoc_ec_evaln2, as % of e-commerce turnover.
GreeceEU27
Web sales to consumers (B2C)45.8%20.8%
Web sales to businesses and government (B2B, B2G)27.1%22.3%
Enterprises making EDI-type sales1.4%5.6%

Nearly half of Greek web-sales value is B2C against a fifth across the Union. And the EDI layer, the automated, high-value, machine-to-machine order flow that dominates industrial e-commerce in northern Europe, barely exists in Greece: 1.4% of enterprises against 5.6%.

This has a practical consequence for reading the headline. The ELSTAT €36.1bn figure is routinely dismissed in commentary because it “includes B2B and EDI”. That caveat is correct and it is smaller than it sounds: with EDI-type selling at 1.4% of Greek enterprises, the B2B contamination of the Greek figure is materially lower than it would be for the same indicator in Germany or Denmark. The Greek number is a poorer measure of consumer e-commerce than a consumer survey, but it is a better one than the equivalent figure would be in most of Europe.

Two: participation is cheap and revenue is not. Adding a checkout to an existing business is a weekend of work and a monthly fee. Moving a meaningful share of revenue through it is a supply-chain, pricing and fulfilment project. A country can lead Europe on the first and trail it on the second indefinitely, and Greece does.

Three: the Greek business population is smaller. Firms of ten to forty-nine employees dominate the Greek count far more than the European one, and small firms generate a smaller share of turnover through any channel. Some part of the 9.5% is arithmetic about firm-size distribution rather than about digital capability. The survey does not publish the decomposition, so this paper does not attempt to quantify it: it is offered as a caution against reading the whole gap as underperformance.

What this means for a merchant. The competitive claim “Greek business has not gone online” is false and has been for years: a quarter of Greek firms with ten or more staff sell through a website, above the European rate. The accurate claim is that they sell little through it. If you are competing in a Greek category, assume your competitors have a store and assume it is a small fraction of their revenue, which means the incumbent is beatable on execution rather than on presence.

04

The demand side has already caught up

While the supply side lags, the Greek consumer has closed a ten-point gap with Europe in four years and, in 2025, crossed it. This is the fastest convergence in the dataset, and it changes which side of the Greek market is the constraint.

4
Individuals who bought or ordered online in the last three months. Eurostat isoc_ec_ib20, % of all individuals aged 16-74.
20212022202320242025
Greece46.845.848.354.262.9
EU2756.956.058.160.262.4
Gap−10.2−10.2−9.8−6.0+0.5

Greece gained 16.1 points in four years against a European gain of 5.5, and ended 2025 marginally ahead of the European average for the first time in the series. A half-point lead is within sampling noise and should not be read as Greece overtaking Europe. What is not within noise is the trajectory: three times the European rate of change, sustained over four waves.

But the population is more polarised

The headline conceals a split that matters commercially more than the average does.

4
Online purchasing behaviour, 2025. Eurostat isoc_ec_ib20, % of individuals.
GreeceEU27Difference
Bought in the last 3 months62.962.4+0.5
Bought in the last 12 months68.073.6−5.6
Has never bought online19.015.3+3.7
Bought 6 or more times in 3 months27.220.1+7.2
Bought 3 to 5 times in 3 months20.320.9−0.6
Bought once or twice in 3 months15.421.3−5.9

Two facts sit either side of the Greek average. Nineteen per cent of Greeks have never bought anything online, against 15.3% across the Union: a larger permanently-offline population. And Greek buyers who do buy, buy far more often: 27.2% made six or more purchases in three months against 20.1% in Europe, while the light-buyer segment of one or two purchases is six points smaller in Greece.

The Greek online market is therefore narrower and deeper than the European one. Fewer participants, each transacting more. That shape rewards retention economics over acquisition economics: in a market where the marginal new buyer is disproportionately someone who has resisted for a decade, the cheaper growth is almost always more orders from existing customers.

One structural ceiling limits how far the headline can run. Internet use itself, the denominator of everything above, stands at 89.2% of Greek individuals against 93.8% across the Union6. Roughly one Greek in nine is not online at all, so the addressable population is smaller than the European one before any commercial question is asked.

05

What Greeks buy online, and what they refuse to

On physical goods, the Greek consumer is indistinguishable from the European one. The divergence is entirely in the intangibles, and it runs in both directions, hard, in a pattern that is legible once you sort it.

5
Online purchases in the last three months, Greece against the EU27, 2025. Eurostat isoc_ec_ibgs, % of individuals. Sorted by the Greek excess.
CategoryGreeceEU27Difference
Subscriptions to internet or mobile connections26.514.0+12.4
Tickets to events38.926.8+12.1
Transport services from a transport enterprise33.725.4+8.3
Film, series or sport streaming subscriptions35.830.7+5.2
Subscriptions to electricity, water or heating supply13.38.4+4.9
Accommodation from hotels or travel agencies25.721.7+4.0
Cultural services: music, film, books, press39.536.8+2.7
Any physical goods for private use60.759.3+1.5
E-books or audiobooks as downloads9.48.9+0.5
Gaming streaming subscriptions4.65.8−1.2
Games as downloads9.111.6−2.4
Health or fitness app subscriptions1.35.4−4.1
Software as downloads7.612.4−4.8
Subscriptions to other apps1.36.6−5.2
Online news or newspaper subscriptions0.96.8−5.9
Music streaming subscriptions10.421.6−11.2

Sorted this way the table divides almost perfectly at the physical-goods line. Everything Greeks over-index on is a payment for access to the physical world or to a household utility: the phone bill, the concert, the ferry, the electricity account, the hotel room. Everything they under-index on is a recurring subscription to a digital good: music, news, apps, software, fitness.

And in between, at +1.5 points, sits the entire category most Greek e-commerce commentary is actually about: physical goods, where Greece is at European parity.

Greeks have adopted the internet as a way to pay for the physical world. They have not adopted it as a way to rent digital things.

The music-streaming figure is the sharpest single number in the table: 10.4% against 21.6%, less than half the European rate, in a category with no logistics, no customs, no delivery and no language barrier. Whatever is suppressing it is not friction in the ordinary e-commerce sense. The same shape repeats for news subscriptions at 0.9% against 6.8% and app subscriptions at 1.3% against 6.6%.

Three explanations are consistent with the data and this paper cannot choose between them. Price relative to disposable income bears hardest on discretionary recurring spend. Card-on-file behaviour is weaker in a market with unusually high cash-on-delivery use, and every one of these products requires a stored card. And substitution to free or informally-obtained alternatives is plausible but unmeasured here. Naming three candidate explanations and declining to pick one is the honest position; the pattern itself is not in doubt.

What follows for a Greek merchant

  1. If you sell physical goods, ignore the “Greeks don't buy online” premise entirely. At 60.7% against a European 59.3%, your category has already converged. The constraint is your economics, not the market's readiness.
  2. Recurring digital revenue is structurally harder here. A subscription model that works in Germany faces roughly half the base rate of subscription behaviour in Greece. That is a pricing and payments problem before it is a marketing problem.
  3. The over-indexed categories are the ones with an unavoidable real-world leg. Tickets, travel, utilities and telecoms: where the online transaction is the only convenient way to obtain a physical outcome. Where an offline alternative is genuinely convenient, Greek online share drops.
  4. Depth beats width. With 19% of the population never having bought online and existing buyers transacting 35% more often than the European average, the marginal euro is better spent on frequency than on conversion of the resistant.
06

How Greeks pay, and how they receive

The cash-on-delivery disagreement

Cash on delivery is the defining feature of Greek e-commerce in every account of it, and the published numbers do not agree.

Cash on delivery, as published
SourceCOD share
GRECA E-Commerce Survey 202562.8%
Later payment-mix reporting~21%

A threefold difference, and not a rounding problem. But it is probably not a disagreement either.

The GRECA figure comes from a survey of 800 Greek online shoppers conducted by online questionnaire and presented in May 2025. What could not be established, after going to the association, the university lab that ran it and the announcement page, is what the 62.8% is a percentage of. Share of consumers who use cash on delivery at all? Share who prefer it? Share of transactions settled that way? The press-release PDF carrying the payment-method breakdown has since been retrieved, and it does not answer the question. It reports headline behavioural findings and names the survey's sponsors, and it states no base, no sample size and no methodology note for the payment split17. Four attempts, and the honest conclusion is now that the base is not published rather than that it could not be found.

That matters more than the gap itself. A share-of-consumers figure and a share-of-transactions figure are not competing estimates of one quantity: they are answers to different questions, and averaging or choosing between them is meaningless. It is the same error that runs through this paper's sister volume on AI statistics, where an “ever purchased” question and a “past twelve months” question were reported as though they measured the same thing.

So the honest position is not “sources disagree, take the range”. It is: one of these numbers is probably about people and the other about orders, and until the base is published neither can be used to size a merchant's exposure.

A band of 21% to 63% is uninformative as a forecast and highly informative as a warning.

Any merchant strategy, policy argument or market model resting on a single COD figure is resting on a number its own sources do not agree about.

Delivery is shifting out of the home

Courier-to-home remains the default at 78.7%, and out-of-home is scaling fast: as reported from the postal regulator’s 2024 market review, automated parcel lockers grew 36.1% in a single year to roughly 5,300 units nationally, and 51.8% of consumers report using them often. The regulator publishes the review; its data files resisted retrieval, so these figures are reported rather than read at source. The postal regulator's data shows why carriers care: envelopes were 78% of items but only 43% of revenue, while parcels were 22% of items and 57% of revenue.

Verification status

Resolved. Eurostat's e-commerce indicator, the structure of ELSTAT's electronic-orders series, and now the ELSTAT 2025 release itself, totals, enterprise counts, the web/EDI split and the reference period, are all opened at primary.

Still press-sourced: the locker and parcel figures. The EETT annual report for 2024 was retrieved and read on 6 August 2026.16 It is an account of the regulator's activities, and it does not contain the parcel volumes or locker counts quoted here; those come from EETT's separate market review, which remains unretrieved. So the 36.1% locker growth, the 5,300 units, the 78.7% courier-to-home share and the item/revenue splits stay as reported and are marked as such wherever they appear. Opening a document and finding the number absent is a different and better-founded state than failing to open it, but it is not verification.

07

The China channel

A fifth of the Greek e-commerce market has left for platforms with a structurally different cost base. Any paper about what an order costs a Greek merchant has to start there.

How many Greeks buy this way, and what that figure actually measures

The European Central Bank's Consumer Expectations Survey, April 2026 wave, asked euro-area consumers a single question: “Have you ever purchased products from Chinese e-commerce platforms (e.g. Temu, AliExpress, Shein, Banggood)?” More than 70% answered yes in Greece, Portugal and Spain: the highest in the euro area, and the three countries are reported together rather than ranked.

The reference window is “ever”. Greek coverage of this study reported it as eight in ten Greeks having ordered in the last twelve months. That is a different question, and the eight-in-ten figure comes from a different study entirely. Two surveys were merged into one claim across several outlets. Both are cited separately below.

The ECB also found what the headline hides: roughly one in five euro-area respondents shop on these platforms at least monthly, use skews toward lower and middle-income households, and almost half of respondents do not use them at all. Wide reach, concentrated intensity.

Greece, Portugal, Spain
>70%
Euro area, monthly+
~20%
Euro area, non-users
~50%

Source: ECB Consumer Expectations Survey, April 2026 wave. Question: “Have you ever purchased products from Chinese e-commerce platforms?” The three countries are reported jointly, not ranked.

What it is worth

The Greek-specific numbers come from a nationwide survey conducted by NielsenIQ for SELPE, the Hellenic Retail Business Association: 85% of Greek online shoppers have bought from at least one non-EU platform, Temu and Shein together hold 15-20% of the Greek e-commerce market, and average annual spend directed to non-EU marketplaces is €244 per shopper.

Read those with the interest attached. They were commissioned by the association whose members are losing the volume. That does not make them wrong, nobody else has measured it, but the coverage discloses no sample size, no fieldwork dates and no question wording, and the association's own publication was not reachable. They are the best available figures and they are not independent ones.

Why price is only half the answer

The ECB attributes the pull to low prices and product variety, especially in clothing, household goods and electronics. But the same Greek survey work that measures the platforms also finds that 55% of Greek online shoppers would not buy from a retailer they do not trust regardless of price.

Those two findings only look contradictory. Price wins where trust is not at stake: a €4 phone case carries no risk worth pricing. It stops winning at the point where a return, a warranty or a data-protection question becomes plausible. That boundary, not the price gap, is where a Greek merchant still competes.

What changed on 1 July 2026

The €150 customs duty exemption for low-value consignments ended. In its place, Council Regulation (EU) 2026/382 introduced a flat €3 customs duty on distance sales up to €150, applying until 1 July 2028.

Two details in the Commission's guidance decide who actually pays. The duty is charged per item, not per consignment: a basket of six cheap items carries €18, not €3. And liability sits with the declarant: the seller, importer, IOSS holder or their indirect representative, not the consumer at the door.

Per item, not per consignment. On a basket of low-value goods, that is the whole margin.

What the duty does to a basket

Because the charge attaches per item rather than per consignment, its effect scales with how many things are in the parcel, which is the defining characteristic of the channel it targets.

1 item
€3
3 items
€9
6 items
€18

Derived from Council Regulation (EU) 2026/382 and the Commission’s implementing guidance: a flat €3 per item on distance sales up to €150. Arithmetic, not an estimate.

This is the first structural change to the channel's economics since it arrived. Whether it closes the gap is an open question this paper does not answer: the duty is nine weeks old at the time of writing and no post-implementation volume data exists yet. What can be said is that the mechanism targets exactly the pattern the platforms depend on: many cheap items, declared low, shipped individually.

What this means for the cost model

The chapters that follow price a Greek merchant's order. The China channel sets the ceiling that pricing runs into. A merchant is not only competing on landed cost against another Greek e-shop: they are competing against a channel that until July 2026 paid no duty at all, and that still operates on a different fulfilment and returns economics.

The honest version of the competitive question is therefore not “how do we match those prices”. It is: which orders are we losing to that channel, which are we keeping because trust is priced into them, and what does each of those two groups actually cost us to serve? The second half of that question is what the rest of this paper measures.

08

Orders that never complete

This section uses Merchant A's order book: approximately 39,000 orders over roughly four and a half years. Volumes are reported as rates and indices only.

The cancellation rate has more than tripled

2022
5.7%
2023
4.6%
2024
9.7%
2025
18.6%
2026 to June
18.3%

From under one order in twenty to nearly one in five, across three years of rapid growth.

What cannot be said yet

The order table records a single status: cancelled. It cannot distinguish a customer who changed their mind before dispatch from a parcel refused at the door from an order killed for stock.

In a market where cash on delivery is somewhere between a fifth and two-thirds of transactions, refusal on delivery is the obvious hypothesis. It is exactly that: a hypothesis, testable only with payment-method metadata and the status-history table.

If it holds, it puts a number on something the Greek trade press discusses constantly and never quantifies: what COD actually costs a merchant, not in fees, but in orders that evaporate after they have been picked, packed and driven.

Orders are falling, like for like

Same five calendar months, year on year
PeriodOrders (indexed)Cancellation rate
January to May 2025100.016.5%
January to May 202685.918.4%

Down 14.1%, with cancellations up 1.9 points. An earlier cut of this comparison suggested a 22% decline. It was wrong: it compared five low-season months against a twelve-month average. The discarded figure is recorded here because a reader is entitled to know which way the errors ran.

Cancellation is seasonal, and the seasonality is real

January
12.4%
March
13.9%
May
23.9%
July
22.1%
September
21.0%
December
14.3%

An eleven-point swing between January and May, inside a single year, so year-mix cannot manufacture the pattern. A pooled version across all years showed the same shape and was discarded, because 2024 and 2025 carry both the volume and the elevated rates and would have produced the curve on their own.

09

Compliance is a cost line, not a footnote

Between 2024 and 2028 a Greek merchant acquires a stack of per-order and per-SKU obligations. Most coverage treats these as legal news. They are cost.

The customs duty, and why the unit matters

The clearest of them is also the newest. Council Regulation (EU) 2026/382 ended the €150 customs duty exemption on 1 July 2026 and replaced it with a flat €3 duty on distance sales up to €150, running until 1 July 2028, after which normal duties apply by product type.

Two provisions in the Commission's guidance decide who carries it.

First, the charge is per item. Not per parcel, not per order. A consignment of six low-value goods attracts six charges. For a category built on multiple cheap units in one shipment, that is not a rounding adjustment to landed cost: it is frequently larger than the margin on the order.

Second, liability sits with the declarant: the seller, the importer, an IOSS holder, a special-arrangements user, or their indirect representative. Only in narrow cases, where a member state offers a free web declaration route, does it fall to the consumer. The default is that the seller side pays.

There are exclusions: goods under preferential trade agreements or Customs Union measures, where VAT was not collected through IOSS and the goods are declared in H1. A merchant sourcing under a preference regime is in a materially different position from one importing at standard rates, and the two should not be modelled together.

What sits alongside it

The customs duty is one line among several arriving in the same window.

Product safety. The General Product Safety Regulation, (EU) 2023/988, requires that a product sold at distance has a named responsible person established in the EU, with that information available to the buyer before purchase. For a merchant importing directly rather than buying from an EU distributor, that role has to be occupied by someone, and occupying it has a price.

The same regulation places specific duties on online marketplaces, read here from the regulation itself rather than from commentary: register a single point of contact for market surveillance authorities and a separate one for consumers; design the interface so traders can supply safety and traceability information; refuse listings where that information is absent; remove content referencing dangerous products expeditiously; and use customer records to notify buyers directly when a product is recalled.

That last obligation is the one merchants underestimate. It converts a marketplace's customer data from an asset into a duty, and it means a recall is no longer a supplier's problem that stops at the warehouse.

Packaging. Regulation (EU) 2025/40 replaces the packaging and packaging waste directive with a directly applicable regulation, bringing recyclability grading, minimisation rules, labelling and extended producer responsibility registration. Its obligations phase in over more than a decade.

Application dates, read from the Official Journal. Earlier drafts omitted these because the texts retrievable at the time truncated before the final articles. Both have now been read in full.

Packaging, Regulation (EU) 2025/40. Article 71 provides that the regulation enters into force on the twentieth day following publication and applies from 12 August 2026, with Article 67(5) applying from 12 February 2029. Adopted 19 December 2024.8

Product safety, Regulation (EU) 2023/988. Article 19 has no separate commencement of its own: Article 52 applies the whole regulation from 13 December 20249, and Article 50 repeals Directive 2001/95/EC with effect from the same day. The distance-selling duties in Article 19 are therefore already in force. They require the offer itself to show the manufacturer's name and postal and electronic address; where the manufacturer is outside the Union, the name and addresses of the EU responsible person; information identifying the product, including a picture of it; and any warning or safety information, in a language easily understood in the member state where the product is made available.

That last set is a product-page specification, not a policy. A Greek merchant reselling imported goods is already required to carry it on every listing.

How to put this in a cost model

The instinct is to treat regulation as overhead: a fixed annual compliance cost, allocated across orders. For this stack that is wrong, and it produces a model that understates the cost of exactly the orders a merchant should be scrutinising.

These obligations attach at three different units:

Where each obligation attaches
ObligationUnit it attaches toBehaves like
€3 import dutyPer itemVariable cost, scaling with basket size
Responsible-person requirementPer product lineFixed per SKU, punishing long-tail catalogues
Packaging recyclability and labellingPer packaging formatFixed per format, amortised over volume
Producer-responsibility registrationPer market enteredFixed per country, punishing thin cross-border expansion
Marketplace safety dutiesPer platform relationshipFixed, but a condition of access rather than a cost

Read that column and the strategic implication is visible without any euro figures at all. A long catalogue of low-value imported SKUs, sold in small baskets across several EU markets, is the shape this stack penalises hardest. A short catalogue of higher-value goods sold domestically is barely touched.

That is worth stating plainly because it is the opposite of how the burden is usually described. The framing in trade coverage is that compliance falls hardest on small merchants. On the evidence of where these obligations attach, it falls hardest on a particular shape of merchant, and plenty of small merchants are not that shape.

What this section does not do is put euro figures on those lines. Doing that requires the per-merchant contract data, courier tariffs, gateway fees, packaging costs, returns handling, that the cost model in the next section is built from, and which is not yet available.

10

What one order actually costs

Every input in this section is a published list price, taken from the tariff sheet or pricing page of the company that charges it, on 6 August 2026. Nothing here is a merchant's negotiated rate and nothing is an estimate. A reader with the same four documents can rebuild every figure.

That is a deliberate choice, and it costs something. Real merchants pay less than list, volume discounts are published and substantial, so the absolute numbers below are an upper bound. What survives the discounting is the structure: which cost lines dominate, how they respond to order value, and which decisions actually move the total. Those are the same at list price and at contract price, and they are what a merchant needs.

The four published inputs

12,13,14
Published Greek e-commerce cost inputs, retrieved 6 August 2026. All figures as printed by the provider.
InputPublished rateSource
Courier, parcel to 2 kg, door to door€10.30ACS domestic price list, valid from 1 Jan 2026. Includes 6.8% fuel surcharge and VAT
Courier, parcel to 2 kg, point to point€4.50Same sheet: collection from a shop or locker rather than an address
Cash-on-delivery collection€3.50Same sheet, “Είσπραξη Αντικαταβολής Εσωτερικού”
Card payment online2.19% + €0.24Viva.com Greek pricing page, domestic and EEA consumer cards
Card payment, domestic consumer card up to €200.50%Same page, under Law 5167/2024 Article 50
Marketplace administrative cost, order €20.01-50, 0-5 kg, from Athens€0.40Skroutz Marketplace pricing policy, published per-order table
Marketplace listing, Starter tierFree + €0.90 per orderSame page. Growth tier €400 a year plus VAT
Marketplace commissionNot publishedVisible only after merchant sign-in. See below

The last row deserves attention before the arithmetic. Skroutz publishes its administrative cost to the cent, a matrix of order-value bands against weight bands, four tiers by fifteen, and does not publish its commission at all. The single largest variable cost of selling on the dominant Greek marketplace is the one a prospective merchant cannot see before signing up.

This paper therefore does not state a Skroutz commission. Third-party integrators publish ranges of roughly 3% to 15% by category; those are secondary, uncheckable, and excluded under the same rule that governs every other figure here. The model below carries the commission as an explicit variable, and a merchant should substitute their own rate from their account.

A worked order

Take a representative Greek order: €45.00 including VAT, one item, one kilogram, shipped from Athens to a Greek address. Greek VAT at the standard rate of 24% means the merchant's net revenue before any cost is €36.29.

The same €45 order under four fulfilment and payment choices. All costs at published list price; product cost held at 55% of net revenue for comparability. Marketplace commission excluded because it is not published.
Own site, card, to doorOwn site, COD, to doorOwn site, card, to lockerMarketplace, card, to door
Gross order value45.0045.0045.0045.00
VAT at 24%−8.71−8.71−8.71−8.71
Net revenue36.2936.2936.2936.29
Cost of goods, at 55% of net−19.96−19.96−19.96−19.96
Courier−10.30−10.30−4.50−10.30
Cash-on-delivery collectionn/a−3.50n/an/a
Card processing−1.23n/a−1.23−1.23
Marketplace administrative costn/an/an/a−0.40
Marketplace per-order fee, Starter tiern/an/an/a−0.90
Contribution before commission, overhead and returns4.802.5310.603.50
As % of gross order value10.7%5.6%23.6%7.8%

Card processing on €45: 2.19% + €0.24 = €1.23. Cost of goods is a stated assumption, not a published figure, and is held constant across all four columns so that it cannot drive the comparison. Marketplace commission is omitted from every column because it is not published; including it would reduce the fourth column only.

What the model actually shows

One: delivery is the dominant variable cost, by a wide margin. At €10.30, door-to-door courier is 22.9% of the gross order value and more than twice the contribution the order generates. It is eight times the card-processing cost. Every conversation about Greek e-commerce margin that begins with payment fees is starting with the fourth-largest line.

Moving one order from the customer's door to a pickup point more than doubles its contribution. No pricing, marketing or payments decision available to a Greek merchant has that effect.

Two: the out-of-home delivery choice is the single largest lever in the model. Point-to-point at €4.50 against door-to-door at €10.30 is a €5.80 swing on an order generating €4.80 of contribution. Contribution goes from 10.7% of order value to 23.6%: it more than doubles. This is the arithmetic underneath the locker and pickup-point growth described earlier in this paper: it is not a convenience trend, it is the difference between a viable and an unviable order.

Three: cash on delivery costs roughly half the contribution on this order. The €3.50 collection fee is a published, visible, per-order charge and it takes contribution from €4.80 to €2.53. That is before any allowance for refusal at the door, which this paper has already flagged as unquantified and which falls entirely on COD orders. A merchant with a high COD share and a door-to-door default is operating on the two most expensive settings simultaneously.

Four: the marketplace is cheaper than it looks, before commission. The published marketplace costs on this order, €0.40 administrative plus €0.90 Starter-tier fee, total €1.30, or 2.9% of order value. That is comparable to card processing. The entire economic question of selling through the marketplace therefore sits in the one number that is not published.

How the lines scale

The four costs behave differently as order value rises, and the differences decide which orders are worth taking.

Published cost lines against gross order value, 1 kg parcel from Athens, at list price.
Gross order value€15€45€120€300
Courier, door to door10.3010.3010.3010.30
… as % of order68.7%22.9%8.6%3.4%
Card processing0.571.232.876.81
… as % of order3.8%2.7%2.4%2.3%
Cash-on-delivery fee3.503.503.503.50
Marketplace administrative cost0.150.400.751.20

Three of the four lines are effectively fixed per parcel; only card processing is genuinely proportional. The consequence is stark at the bottom of the range: on a €15 order, door-to-door courier alone is 68.7% of the order value. No product margin in ordinary retail survives that. A Greek merchant shipping small baskets to the door at list price is not running a low-margin business; they are running a business whose smallest orders are structurally loss-making, and the loss is invisible in any metric that averages across order sizes.

The card-processing line has one exception worth knowing. Under Law 5167/2024, Article 50, transactions on domestic consumer cards up to €20 are charged at 0.50%10, so on a €15 order the published Viva rate falls to €0.08 rather than €0.57. It is a real saving on exactly the orders where every cent matters, and it applies automatically. It is also, at €0.49, roughly a twelfth of what moving that order to a pickup point would save.

Skroutz's administrative cost is the one line in the table that scales with order value by design: €0.15 at the bottom band, €1.20 at the top. It is the only published Greek e-commerce cost input the author found that is explicitly structured to take less from small orders.

What this model does not include

Four terms are missing and a merchant must add them from their own accounts. They are named rather than estimated.

  1. Marketplace commission: not published, and on a €45 order a 10% rate would remove €3.63 of the €3.50 contribution shown. The marketplace column is not a claim of profitability; it is a claim about the published costs only.
  2. Returns: the correct treatment is a provision equal to the return rate multiplied by the round-trip shipping plus handling, applied to every order rather than to returned ones. At a 10% return rate and a €10.30 return leg, that is €1.03 off every order in the first column, or 21% of its contribution.
  3. Packaging and pick time: small, real, and merchant-specific.
  4. Discounts and marketing: the merchant-borne portion of any promotion, and customer acquisition cost, neither of which has a published list price.

Adding those four to the first column would take a €4.80 contribution close to zero. That is not a rhetorical flourish; it is the reason the second and third columns matter so much. On published Greek list prices, the difference between a viable and an unviable order is mostly the delivery method and the payment method, not the product.

On the honesty of list prices. No merchant of any size pays €10.30 to ship a parcel. ACS publishes volume discounts on the same sheet, from 10% at eleven monthly shipments up to 30% at the top tier, and negotiated e-commerce contracts run below the published discount ladder. Skroutz publishes administrative-cost rebates of 10% at 2,500 monthly orders and 15% at 5,000.

Applying a 30% courier discount to the first column moves contribution from €4.80 to €7.89: a 64% improvement from a single negotiated rate, and larger than any other lever in this model including the pickup-point switch. That is itself the finding: in Greek e-commerce, the courier contract is the P&L. It also means small merchants, who cannot reach the volume tiers, face a structurally worse cost base than their published-price competitors on precisely the line that matters most, which is a plausible mechanism for the wide-and-shallow pattern documented in section 03.

11

Sources, and how to check them

Every figure in this paper is either a published statistic opened at its source, a published list price taken from the company that charges it, or a labelled assumption. This section says which is which.

Official statistics

SourceUsed forStatus
ELSTAT, Survey on the Use of ICT and E-commerce in Enterprises 2025, released 17 Dec 2025€36.1bn, €380.8bn, 25.8%, the web/EDI split, the reference periodOpened at source 6 Aug 2026, after four earlier failures
Eurostat isoc_ec_eseln2, isoc_ec_evaln2Enterprises selling online, e-commerce turnover share, B2C/B2B composition, EDIPublic API, extract of 6 Aug 2026
Eurostat isoc_ec_ib20, isoc_ec_ibgs, isoc_ci_ifp_iuBuyer penetration, purchase frequency, category mix, internet usePublic API, extract of 6 Aug 2026
Regulation (EU) 2025/40 and (EU) 2023/988, Official JournalPackaging and product-safety obligations and their commencement datesRead in full from the Official Journal PDFs
EETT annual report 2024Nothing. It was opened and does not contain the courier and locker figures quoted from EETT elsewhere in this paperOpened; figures absent. Those figures remain as reported
GR.EC.A / ELTRUN e-commerce survey 2025Consumer behaviour figuresPress release opened on the fourth attempt. It states no base, no sample size and no method for the payment split

Published commercial prices

ProviderDocumentRetrieved
ACS CourierDomestic services price list, cash tariff, valid from 1 Jan 20266 Aug 2026
Viva.comGreek pricing page, card-acceptance fees6 Aug 2026
SkroutzMarketplace pricing policy, partner support6 Aug 2026

ACS was used because it is the only one of the four major Greek couriers publishing a current, dated, nationwide retail tariff in a form that can be cited. Geniki Taxydromiki, Speedex and BOX NOW were checked and no equivalent 2026 national list price was found. That is a limitation of the model and also a small finding about the market: three of the four largest Greek parcel carriers do not publish a price a prospective merchant can read.

Stated assumptions

Three, and they are the only unsourced numbers in this paper.

  1. Cost of goods at 55% of net revenue in the worked order. Held constant across all four columns so it cannot drive any comparison. A merchant should substitute their own.
  2. A representative order of €45, one item, one kilogram, Athens to a Greek address. Chosen to sit in the middle of the published Skroutz value band; the scaling table shows what happens either side of it.
  3. A 10% return rate in the returns illustration, used only to show the shape of the provision, not as a claim about any market.

Everything else in the paper carries a source. Where a figure could not be sourced, the Skroutz commission, the GRECA payment base, the EETT parcel volumes, it is named as missing rather than estimated, and no argument here rests on it.

12

References

Every source below was opened by the author at the URL shown, on the date shown. Two entries record sources that were opened and found not to contain the figure sought; those are kept in the list because a failed retrieval is part of the evidence.

Official statistics

  1. Hellenic Statistical Authority (ELSTAT). Survey on the Use of Information and Communication Technologies and E-commerce in Enterprises, 2025. Piraeus, 17 December 2025. Reference period 1 January to 15 September 2025. Enterprises employing 10 or more persons, NACE Rev. 2 divisions 10-63, 68-82 and group 95.1. https://www.statistics.gr/documents/20181/4d030e6d-89df-280f-e3ae-7471cf769816 Opened at source 6 August 2026 after four earlier retrieval failures. Source of €36,122m, €380,787m, 47,849 enterprises, the 11,700 / 647 web-versus-EDI split, and the statement that 2025 is not comparable with 2024.
  2. Eurostat. E-commerce sales of enterprises by NACE Rev. 2 activity. Dataset isoc_ec_eseln2. Data published 15 June 2026. https://ec.europa.eu/eurostat/databrowser/view/isoc_ec_eseln2/default/table Extracted 6 August 2026. Source of the enterprises-selling-online ranking and the 2021-2025 series.
  3. Eurostat. Value of e-commerce sales by NACE Rev. 2 activity. Dataset isoc_ec_evaln2. Data published 27 February 2026. https://ec.europa.eu/eurostat/databrowser/view/isoc_ec_evaln2/default/table Extracted 6 August 2026. Source of e-commerce turnover share and the B2C / B2B composition.
  4. Eurostat. Internet purchases by individuals (2020 onwards). Dataset isoc_ec_ib20. Data published 17 April 2026. https://ec.europa.eu/eurostat/databrowser/view/isoc_ec_ib20/default/table Extracted 6 August 2026. Buyer penetration, never-bought share and purchase frequency.
  5. Eurostat. Internet purchases: goods or services (2020 onwards). Dataset isoc_ec_ibgs. Data published 17 April 2026. https://ec.europa.eu/eurostat/databrowser/view/isoc_ec_ibgs/default/table Extracted 6 August 2026. The category-level Greece-versus-EU comparison.
  6. Eurostat. Individuals: internet use. Dataset isoc_ci_ifp_iu. Data published 17 April 2026. https://ec.europa.eu/eurostat/databrowser/view/isoc_ci_ifp_iu/default/table Extracted 6 August 2026.
  7. Regulation (EU) 2019/2152 on European business statistics. Official Journal L 327, 17 December 2019. https://eur-lex.europa.eu/eli/reg/2019/2152/oj The legal basis making the enterprise figures comparable across member states.

Law and regulation

  1. Regulation (EU) 2025/40 of the European Parliament and of the Council of 19 December 2024 on packaging and packaging waste, amending Regulation (EU) 2019/1020 and Directive (EU) 2019/904, and repealing Directive 94/62/EC. Official Journal, series L, 2025/40. https://eur-lex.europa.eu/eli/reg/2025/40/oj Read in full from the Official Journal PDF, 6 August 2026. Article 71: applies from 12 August 2026; Article 67(5) from 12 February 2029.
  2. Regulation (EU) 2023/988 of the European Parliament and of the Council of 10 May 2023 on general product safety. Official Journal L 135, 23 May 2023. https://eur-lex.europa.eu/eli/reg/2023/988/oj/eng Read in full 6 August 2026. Article 19 sets the distance-selling information duties; Article 52 applies the regulation from 13 December 2024; Article 50 repeals Directive 2001/95/EC from the same date.
  3. Law 5167/2024 (Greece), Article 50, amending the framework for card-acceptance charges on domestic consumer cards. Cited as the legal basis for the 0.50% rate on domestic consumer-card transactions up to €20, as published by the payment provider at reference 13. The provision is cited from the provider’s own pricing page rather than from the Government Gazette text, which was not opened; a reader relying on it should read the statute directly.
  4. European Commission, Directorate-General for Taxation and Customs Union. Guidance and legal text on the temporary flat-fee arrangement for low-value consignments. https://taxation-customs.ec.europa.eu/ Used for the per-item duty arithmetic in the China-channel section.

Published commercial tariffs, used in the cost model

  1. ACS Courier. Domestic Services Pricelist: cash tariff, valid from 1 January 2026. Prices include a 6.8% fuel surcharge and VAT. https://www.acscourier.net/el/ypiresies/taxymetafores-eswterikoy/timokatalogos/ PDF retrieved 6 August 2026. Shipments to 2 kg: point-to-point €4.50, point-to-door €8.80, door-to-door €10.30. Cash-on-delivery collection €3.50. Volume discounts of 10% to 30% by monthly shipment count on the same sheet. ACS was used because it is the only one of the four largest Greek carriers publishing a current, dated national retail tariff; Geniki Taxydromiki, Speedex and BOX NOW were checked and no equivalent was found.
  2. Viva.com. Pricing: Greece. https://www.viva.com/el-gr/pricing Retrieved 6 August 2026. Online payments 2.19% + €0.24 per transaction on domestic and EEA consumer cards; card-present 1.69% with a €0.03 minimum; 0.50% for domestic consumer cards up to €20 under Law 5167/2024 Article 50.
  3. Skroutz. Marketplace pricing policy (Τιμολογιακή Πολιτική Skroutz Marketplace). Partner support. https://partnersupport.skroutz.gr/hc/el-gr/articles/31775165438609 Retrieved 6 August 2026. Starter tier free with €0.90 per order; Growth tier €400 a year plus VAT; administrative cost per order by pickup location, order value and weight (€0.15 to €1.20 for 0-5 kg from Athens); rebates of 10% at 2,500 monthly orders and 15% at 5,000. The commission rate is not published on this or any public page and is visible only after merchant sign-in. It is therefore absent from the cost model, and no third-party estimate of it has been used.

Other research cited

  1. European Central Bank. Consumer Expectations Survey, and accompanying ECB analysis of euro-area purchasing from Chinese e-commerce platforms, 2026. https://www.ecb.europa.eu/press/blog/date/2026/html/ecb.blog20260713~ae38da7434.en.html Used for the “ever purchased” figures in the China-channel section, with the question wording reported alongside because it is an ever question rather than a recent-period one.
  2. Hellenic Telecommunications and Post Commission (EETT). Annual Report 2024: Summary of Actions. https://www.eett.gr/wp-content/uploads/2025/12/ANNUAL_REPORT_2024.pdf Opened 6 August 2026 and found not to contain the figures sought. It is an account of the regulator’s activities. The parcel volumes, locker counts and item/revenue splits attributed to EETT in this paper come from EETT’s separate market review, which could not be retrieved, and are marked “as reported” wherever they appear.
  3. Greek E-Commerce Association (GR.EC.A) and ELTRUN, Athens University of Economics and Business. Annual E-Commerce Survey 2025. Presented 7 May 2025. https://eltrun.org/wp-content/uploads/2025/05/DT_GR-ELTRUN.pdf Opened at the fourth attempt, 6 August 2026, and found not to state the base. The release gives headline behavioural findings and names its sponsors, and states no sample size, no base and no methodology note for the payment-method split. The conclusion recorded in this paper is therefore that the base is not published, rather than that it could not be found. Survey sponsors named on the release: Eurobank (principal), DHL, Skroutz and Worldline.
  4. NielsenIQ for SELPE (Hellenic Retail Business Association). Nationwide survey on Greek use of non-EU e-commerce platforms. https://www.selpe.gr Commissioned by a retail trade body with a direct interest in the answer; stated in the text where used.

Data published with this paper

  1. Broikos, N. (2026). Eurostat extracts for “What a Greek online order actually costs”. Five datasets, frozen at the date of writing. https://broikos.gr/research/data/paper-02/
  2. Broikos, N. (2026). Claim ledger and source log. 66 rows recording the verification status of every figure considered, including those cut. https://broikos.gr/research/data/paper-02/
  3. Merchant order data, three Greek online retailers, withheld. Identities withheld under a confidentiality undertaking; reported only as rates and indices. Not independently verifiable: see the data-availability statement. No conclusion in this paper depends on these figures.
13

Limitations

LimitationHandling
The cost model uses published list prices, which no real merchant paysStated at the top of the section and quantified at the bottom: a 30% courier discount changes the result by 64%. The model is offered for the structure of the cost stack, not for its absolute values
The Skroutz commission is not published and is therefore missing from the marketplace columnNamed explicitly. The marketplace column is a statement about published costs only and is not a profitability claim. A 10% commission would remove the entire contribution shown
Only one courier's tariff could be cited, because only one publishes a current national retail price listNamed, with the three that do not. The delivery-cost conclusion depends on the ratio between door and point delivery, which is unlikely to invert at another carrier, but the absolute figures are ACS's
Cost of goods, order size and return rate are assumptions, not measurementsAll three listed in the sources section as the paper's only unsourced numbers, and cost of goods is held constant across every column of the comparison
Eurostat enterprise data covers only firms with ten or more employees, which is a smaller share of the economy in Greece than in most comparatorsStated wherever the figures appear. It means Greek participation is probably overstated relative to the full business population, not understated
Two of three merchants are marketplace-integrated, so channel dependency reads highStated; the own-site-primary merchant is the counterweight and is the one this draft reports
The available snapshot has no values, payments, shipping or geographySections on cost are blocked rather than estimated
Merchant data is three businesses under common ownership, not a sampleLabelled a case throughout; public figures printed beside theirs
“Cancelled” is read as one event type and certainly is notStated at the point of use; the cause is labelled a hypothesis
Anonymisation is weak by constructionStated in the method box rather than left for the reader to work out
Snapshots, not live data: later status changes may understate recent cancellationsStated
Public statistics currently drawn from press reporting, not primary tablesEvery one flagged; none survives to publication unopened

The materially weakening limitations are the first two. The first means channel economics will read high. The second means the paper's central promise is not yet delivered.

14

Citation, licence and interests

How to cite this paper

Broikos, N. (2026) What a Greek online order actually costs: market structure from harmonised European statistics, and a landed-cost model from published Greek tariffs. Version 1.0, 6 August 2026. Athens. Available at: https://broikos.gr/research/what-a-greek-online-order-costs.pdf (Accessed: DD Month YYYY).

Version

Version 1.0, published 6 August 2026. This is the first public version. Substantive changes will increment the version number and be listed in the change log published with the data.

Data availability

The statistical data underlying this paper is published at https://broikos.gr/research/data/paper-02/ under the same licence as the text: the five Eurostat extracts frozen at the date of writing, the 66-row claim ledger and the source log. The cost model can be rebuilt from the three published tariffs cited in the references; no proprietary input is required.

What is withheld, and what that means for the reader. A small number of figures in this paper, the cancellation rate and the like-for-like order decline, are derived from the order tables of three Greek online retailers whose identities are withheld under a confidentiality undertaking. Those figures are published only as rates and indices. No absolute revenue, order count, customer record or product-level figure appears anywhere in this paper, and no merchant, platform, marketplace or category is named.

Those specific figures therefore cannot be independently verified, and the reader is asked to take them on trust. The author considers that acceptable here for one reason: no argument in this paper depends on them. The market structure comes entirely from Eurostat and ELSTAT, and the cost model comes entirely from published list prices. Remove the merchant figures and every conclusion in this paper stands unchanged. They are included because they are illustrative, not because they are load-bearing.

Declaration of interests

The author operates e-commerce businesses in Greece, including the three anonymised retailers whose aggregate figures appear in this paper. The author therefore has a direct commercial interest in the Greek e-commerce market and in the cost lines analysed here.

Two consequences a reader should weigh. First, the merchant figures are self-reported by an interested party and unverifiable, which is why they are confined to illustration and excluded from every conclusion. Second, the author is a customer of couriers, payment providers and marketplaces of the kind priced in section 10, and has commercial relationships in that market. None of those companies was consulted, sponsored this work, or saw it before publication, and all pricing used is each company's own public list price rather than any rate available to the author.

Licence

This paper and the datasets published with it are released under a Creative Commons Attribution 4.0 International licence (CC BY 4.0). You may copy, redistribute, quote, chart and build on this material, including commercially, provided you credit the source. Licence text: https://creativecommons.org/licenses/by/4.0/, and served beside this paper at https://broikos.gr/research/LICENSE-CC-BY-4.0.txt. What the grant covers and what it withholds is itemised in https://broikos.gr/research/LICENSING.md. Analysis code is MIT: https://broikos.gr/research/LICENSE-MIT.txt

Funding and independence

Unfunded. No sponsor, client, trade body or commissioning party paid for, commissioned, reviewed or approved this research. No organisation named in it was given sight of it before publication. This matters because a substantial share of the sources graded here were commissioned, and the paper argues that a commissioning interest should be stated wherever it exists.

Corrections

If a figure here is wrong, or an organisation is described incorrectly, write and it will be corrected with the date and the reason recorded in a public change log rather than silently edited. That log is published at https://broikos.gr/research/corrections.md and already records every figure cut, corrected or downgraded during production. Every correction made during production is already recorded in the claim ledger published with this paper, including the figures that were cut.

Contact: https://broikos.gr/contact · https://broikos.gr

About the author

Nikolaos Broikos operates e-commerce businesses in Greece, works in web development and digital strategy, and builds agent harnesses. He writes from Athens.

This programme is the record of that work rather than a commentary on it. The e-commerce operations supply the transaction data behind the landed-cost model in What a Greek online order actually costs; the harnesses are the instruments measured in What actually makes an agent harness work. Where the author’s own systems are the subject, that is stated in the first paragraph of the paper concerned as well as in its declaration of interests.

No academic affiliation, no institutional backing, no funding, and no client commissioned any of this. The papers therefore ask to be judged on their published instruments, data and corrections rather than on credentials: every dataset, every claim ledger including the claims that failed, and every retraction is published alongside the text, so a reader who distrusts the author can check the work instead.


The other papers in this programme

Independent, unfunded and published free under CC BY 4.0, each with its underlying data. Read separately; they share a method, not an argument.

  1. Who publishes the numbers: a census of corporate research in Greece https://broikos.gr/research/who-publishes-the-numbers.pdf 68 Greek publication programmes coded on a published schema: who produces the numbers business decisions rest on, what motivates them, how far the press carries them, and which of them answer engines actually cite.
  2. The adoption gap: what businesses say about AI and what the statistics measure https://broikos.gr/research/the-adoption-gap.pdf Where Greek firms actually sit on AI adoption once the size classes are separated, why the skills-gap explanation does not survive the spending data, and what Greek firms bought instead.
  3. What actually makes an agent harness work https://broikos.gr/research/agent-harness.pdf Techniques, prompt engineering and measured results from three agent harnesses, two built by the author and one not, across 45 graded sessions on a single model, including three faults found in the measuring apparatus itself.
  4. The Greek e-shop technical audit https://broikos.gr/research/greek-eshop-audit.pdf 900 Greek domains measured on one day: what Greek online retail actually sends to a browser. 36.8% send no security headers, 11.1% fail a validating TLS handshake, and 5 shops out of 505 pass five elementary checks.
  5. The Greek SME digital bill https://broikos.gr/research/greek-sme-digital-bill.pdf What it costs per year to sell online in Greece, priced entirely from published pages: EUR 157.50 at the floor. Of 25 cost lines, 7 have no published price at all, and they are disproportionately the mandatory ones.
  6. What an agent actually costs https://broikos.gr/research/what-an-agent-costs.pdf Seven versions of one production agent on the same benchmark and the same model, including the two versions that got worse, the failure taxonomy, and the chart that would have shown a cost explosion that never happened.