The Greek SME
digital bill

Discharges claims F001, F002, F003.

This paper tries to answer a question a Greek small business owner asks before starting: what does it cost, per year, to sell online?

The method is deliberately narrow. Every figure comes from a published price on the provider's own page, retrieved on 7 August 2026, with the URL recorded. No quotes were requested. No salesperson was called. Nothing is estimated from experience or from what "people usually pay".

That constraint produced the paper's actual finding, which is not a number.

Half the bill is not published

Of thirty-two cost lines a Greek online business unavoidably carries, twenty-six are published openly and six cannot be priced from any public page at all.

The split is not random. The published half is infrastructure sold to consumers and small businesses by companies competing on price. The unpublished half is business software and logistics sold through a sales process.

Specifically, and each verified on the day:

• Four major Greek business-software vendors publish no price of any kind. SoftOne, Epsilon Net, Entersoft and Pylon carry no euro amount anywhere in their served pages. softone.gr/pricing and softone.gr/plans return the homepage; entersoft.gr/pricing returns a JPEG. There is no pricing page to find.
• The two national couriers are opposites, and an earlier draft of this paper got one of them wrong. ACS publishes a complete, dated, versioned tariff, as a PDF linked from its price page rather than as page content, which is why an HTML-only check missed it. ELTA Courier publishes no price at all: its τιμοκατάλογος page lists services and invites you to request a quote, and contains no amount even when rendered in a browser. Section 04 carries the ACS numbers and section 08 records the correction.
• The dominant Greek marketplace refuses automated clients entirely. Skroutz returns HTTP 403.

So a would-be Greek online seller can price their domain, their hosting, their card processing and their parcel delivery in an afternoon from public pages, and cannot price their accounting software, their courier contract or their marketplace commission without entering a sales conversation.

The unpriceable half is disproportionately the mandatory half. Accounting software is not optional for a Greek business; myDATA e-invoicing is a legal obligation.

What the published half costs

For the smallest real configuration: one .gr domain, shared hosting, a free TLS certificate, self-hosted WooCommerce, one business mailbox:

One hundred and fifty-seven euro and fifty cents a year buys a working, encrypted, self-hosted Greek online shop with a business mailbox. That is the floor, and it is remarkably low.

The hosted alternative is roughly two and a half times that before a single order: Shopify Basic at €19/month is €228/year on its own, against €75.90 for domain and hosting.

Section 04 builds the per-order costs on top, and section 05 says plainly which lines the model cannot fill.

What this paper is not

It is not a business plan. Stock, marketing, salaries, accountancy fees, social insurance and tax are all outside it. This is the digital bill: the cost of the technical means to sell.

It is not a recommendation of any provider. Providers appear because they publish a price, which is precisely the selection bias section 05 describes.

It is not a survey of what Greek businesses actually pay. Published list prices are not transaction prices, and in the unpublished half the gap between them is the entire point.

It is a companion to two other papers in this programme: one measured what a Greek online order costs per unit from published tariffs, and one measured what 900 Greek shops technically look like. This one prices the year rather than the order, and finds that the market will only tell you half of it.

Discharges claims F004, F005, F006, F007.

Thirty-two cost lines were attempted. Each was recorded as published, not published, not machine-readable, or blocked. The method for each was identical: open the provider's own site as an ordinary visitor with an identifying user agent, and look for a euro amount.

The result

The four vendors with no price

This is the finding worth checking, so here is exactly what was done.

Each vendor's homepage was fetched and the served HTML searched for any euro amount: € next to a digit, a digit next to €, or the word ευρώ. All four returned zero.

Then conventional pricing paths were probed: /pricing, /timokatalogos, /el/pricing, /plans, /τιμές. Several returned HTTP 200, which initially looked like a pricing page existed. They do not:

• softone.gr/pricing and softone.gr/plans return the homepage: same <title>, same content. A catch-all route, not a page.
• entersoft.gr/pricing returns a JPEG.

So the 200s were soft-404s, and the conclusion stands in its narrow form: no published price for accounting or ERP software could be found on any of the four vendors' own sites.

The honest limit of that claim: it does not prove no price list exists anywhere. It proves that a business owner who wants to know what accounting software costs cannot find out by looking, which is the thing that matters to the person asking.

The couriers turned out to be opposites

Both ACS and ELTA Courier maintain a page titled τιμοκατάλογος: price list. Neither contains a euro amount in the HTML the server sends, so an HTML-only check records both as unpublished. That check was wrong about one of them, and rendering the pages in a real browser showed why.

ACS publishes properly. Its price page links to a PDF: Τιμοκατάλογος Υπηρεσιών Ταχυμεταφορών / Μεταφορών της ACS, August 2026, version 7.5, carrying the company's registry number and a full rate table by destination, service level and weight. That is a more durable form of publication than an HTML table: it is dated, versioned, archivable and citable. It fails a machine check and passes every test that matters to a buyer.

ELTA Courier publishes nothing. Its page lists service names, door-to-door, EMS, SPM, Fast Parcel Cyprus, and a heading reading "ask for an offer and work with ELTA Courier". There is no rate table in the HTML, none in the rendered page, and no tariff PDF linked from it.

So the two incumbents sit at opposite ends of this paper's scale, and the newer locker operator BOX NOW sits in the middle: it puts headline from-prices directly in the HTML, which is the most machine-checkable of the three and the least detailed.

The lesson is about the instrument, not the couriers, and section 08 records it: a rule that reads only served HTML will call a properly published PDF tariff "unpublished". The rule was kept, it is the right rule for the question, but its output must be read with the failure mode in mind, and a browser check is the cheap correction.

The pattern

Sort the thirty-two lines by who sells them and the split is almost perfect.

Published: domains, hosting, certificates, e-commerce platforms, business email, card processing, locker delivery. Every one is sold self-service, to consumers and micro-businesses, by companies competing on a price page. Several are international.

Not published: ERP, accounting, myDATA compliance software, one courier, marketplace terms. Every one is sold through a Greek sales process: a call, a demo, a quote sized to your business.

The consequence for the buyer is asymmetric in one direction. Where prices are published, a small business pays roughly what everyone else pays and can compare in minutes. Where they are not, the price is a negotiation, and the smallest buyer negotiates worst.

Why this belongs in a paper about cost

Because it is the reason the question "what does it cost to sell online in Greece?" has no published answer, despite every component being purchasable today.

A first-time seller can assemble the technical half, €157.50 a year, section 01, from public pages in an afternoon. The mandatory half, accounting and e-invoicing, requires entering a sales process before learning the price of a legal obligation.

The absence is the measurement. Sections 03 and 04 price what can be priced; section 05 states what is missing and does not estimate it.

Discharges claims F008, F009, F010, F011.

Three configurations, priced entirely from published pages. Each assumes one business, one domain, one mailbox, and no orders yet: this is what the year costs before anyone buys anything.

Configuration A: self-hosted, minimum

Configuration B: self-hosted, a real small business

Two mailboxes, a VPS instead of shared hosting, and a .com alongside the .gr so the brand is not squatted.

Configuration C: hosted platform

The same business on Shopify rather than self-hosted. Hosting, TLS and the platform collapse into one subscription; the domain and mailboxes do not.

What the comparison says

The technical floor for selling online in Greece is about €13 a month. That is the entire published fixed cost of a working, encrypted shop with a business mailbox.

The hosted platform costs €192 a year more than the equivalent self-hosted configuration: €408.70 against €216.70. Over five years that is a little under a thousand euro, which buys a meaningful amount of the developer time that self-hosting requires and hosted platforms do not.

That trade is the honest way to read the difference, and it is also the most plausible explanation for a result in a companion paper in this programme: a technical audit of 900 Greek domains found Shopify on 0.8% of them, against WooCommerce at 16%. Greek online retail is self-hosted, and at these prices the reason is not mysterious.

Where the fixed bill is misleadingly small

Hosting at €5.70/month is a "from" price, and shared hosting is where a shop lives until it has traffic. The published entry price is real; the price of the tier a growing shop actually needs is not knowable from the same page without configuring a basket.

Google Workspace is per user per month, and it is the only line that scales with headcount. Going from one mailbox to five takes the email line from €81.60 to €408: on its own more than configuration A entire. For a business with staff, email is the largest published fixed line, which is not where anyone expects the money to go.

Nothing here includes the mandatory software. Accounting, ERP and myDATA e-invoicing carry no published price, so no configuration in this section includes them. Every total above is therefore incomplete by an unknown amount, and section 05 refuses to guess it.

That is worth stating in the plainest possible way: the fixed annual bill for a Greek online business cannot be computed from public information. What can be computed is the part sold by companies that compete on a price page, and that part comes to between €157 and €409.

Discharges claims F012, F013, F014, F015.

The fixed bill is small. The per-order costs are where a Greek online business actually spends, and two of the three components are published.

Card processing

The third row is the interesting one. Greek law caps the fee on domestic consumer-card transactions up to €20 at 0.50%: Law 5167/2024, Article 50, cited on the provider's own pricing page.

That is a genuine and under-discussed feature of the Greek market. On a €15 order the statutory rate costs €0.075. The standard online rate would cost €0.57. The cap is roughly seven times cheaper on a small basket, and it applies automatically.

It also inverts the usual advice. Card fees are normally regressive: a fixed component of €0.24 hurts small orders most. In Greece, below €20 and for domestic consumer cards, they are the opposite.

Worked, on the published rates:

The cap only reaches domestic consumer cards on orders up to €20. Business cards, foreign cards and anything above €20 pay the standard rate, so a shop's realised blended rate depends entirely on its basket and customer mix, which this paper cannot observe.

Delivery

The incumbent courier, now priced

ACS publishes a full tariff as a dated PDF. Express Next Day, parcels up to 2 kg, August 2026, version 7.5:

Set against the locker operator, the gap is large and it is the whole reason lockers exist:

A locker parcel costs less than half the cheapest same-city courier movement and roughly a quarter of a mainland door delivery. The €3.00 figure used throughout this paper is therefore the floor of a real and wide range: the same domestic order costs €3.00 to a locker and up to €14.40 to a door in a hard-to-reach area.

ELTA Courier publishes no rate at all, so no third comparison is possible.

The EU figure is worth noting for anyone reading the cross-border growth stories: most EU destinations carry the same €15 headline: France, Germany, Italy, Belgium, the Netherlands, Spain, Austria, Czechia, Portugal, Denmark, Finland, Ireland and Poland all show it. Cyprus at €5 and Bulgaria at €3 are the exceptions, and both are proximity cases.

Domestic delivery costs €3 and the rest of the single market costs €15. For a Greek shop with a €25 average basket, that is the arithmetic that decides whether cross-border is a strategy or a slogan.

What one order actually costs

Two illustrative orders, using only published rates:

A €15 domestic order, paid by Greek consumer card, delivered to a locker:

A €60 order to Germany, standard card rate, locker delivery:

Neither figure includes the goods, packaging, returns, marketplace commission or labour. They are the payment-and-delivery floor, and on a small domestic order that floor is already a fifth of the order value: almost entirely because of delivery.

Card processing is 2.4% of the domestic order's cost and delivery is 97.6% of it. On the cross-border order, delivery is 91%.

The policy conversation in Greece is about card fees, which are capped and small. The measurable cost is delivery, and it is not capped.

What is missing here

Marketplace commission is absent because the dominant Greek marketplace returns HTTP 403 to an identified research client and publishes no rate card reachable from a public page. For a shop that sells through it, this is likely the largest per-order line of all, and this paper cannot price it.

Returns are absent. A companion paper in this programme models them from operational data; nothing about return rates can be sourced from a published page.

The from-prices are floors. Every delivery figure above is a "from", and the realised price depends on weight, dimensions and contract volume: the terms of which are, as section 02 establishes, exactly the terms nobody publishes.

Discharges claims F023, F024, F025, F026.

Fixed costs matter at zero orders and stop mattering quickly. This section runs the published prices across order volumes to find where each decision actually changes.

Total published cost per year, by volume

Configuration B from section 03: two domains, VPS, self-hosted WooCommerce, two mailboxes, €216.70 fixed, with domestic locker delivery at €3.00 and a €15 average basket paid by Greek consumer card under the statutory 0.50% cap.

At 1,000 orders a year the entire fixed technical bill is 7% of published digital cost. At 5,000 it is 1.4%.

Per-order cost falls from €5.24 to €3.12 and then stops, because it is asymptotic to delivery plus payment: €3.075 on these assumptions. No amount of volume gets a Greek shop below three euro per domestic order on published prices, because €3.00 of that is the delivery floor.

That is the number to carry: the marginal cost of a domestic order is delivery, and delivery does not scale down.

Where the platform decision breaks even

Shopify Basic (€228/yr) against self-hosted (VPS €36 + WooCommerce €0 = €36/yr). The difference is €192 a year, and it is fixed: it does not move with volume.

Below roughly 500 orders a year the hosted platform premium is material: €1.92 per order at 100 orders is 13% of a €15 basket. Above 1,000 it is rounding.

So the honest advice inverts the usual assumption. A hosted platform is often recommended to the smallest sellers as the simple option, and it is precisely the smallest sellers for whom the premium bites hardest per order. At scale the €192 disappears into the noise and the decision should turn entirely on engineering time, not on subscription cost.

The five-year cumulative difference is €960: real money for a micro-business, and less than two days of developer time at a professional rate. Whoever has the time should self-host; whoever has the money should not agonise.

Where the payment cap stops helping

The statutory 0.50% cap applies to domestic consumer cards on orders up to €20. Above that, the standard published online rate of 2.19% + €0.24 applies.

The cap is worth up to €580 a year to a shop doing a thousand orders at the €20 boundary: more than twice the entire fixed technical bill.

And it produces a genuine cliff. An order of €20.00 costs €0.10 to process; an order of €20.01 costs €0.68. Seven times more, for one cent of basket. Nothing in this paper suggests engineering around that, and a shop should not price its catalogue to game a payment cap, but a merchant whose average basket sits near €20 should know the boundary exists, because it is not visible anywhere in normal reporting.

Where cross-border stops working

Domestic locker delivery starts at €3.00. Thirteen EU destinations all show €15.00.

A €15 order to Germany costs €15 to deliver. The delivery equals the order. Cross-border only becomes arithmetically sane somewhere above a €60 basket, where delivery falls to a quarter of order value, and it is genuinely comfortable only above €150.

For a Greek shop whose domestic average basket is €15 to €25, "expand into the EU" is not a strategy at published locker rates unless the basket changes with it. Cyprus at €5.00 and Bulgaria at €3.00 are the exceptions and both are proximity markets.

This is where the paper's method shows its limits, and the limit runs in the merchant's favour: these are published from-prices for single parcels. A shop with volume negotiates a courier contract, and the terms of that contract are, as section 02 establishes, exactly what nobody publishes. The cross-border numbers above are therefore a ceiling on cost for a serious exporter and an accurate floor for everyone else.

What the volume table cannot include

Every row above omits the unpublished half: accounting software, ERP, myDATA e-invoicing, marketplace commission and negotiated courier rates.

For a shop selling through the dominant Greek marketplace, commission is very likely the largest single per-order line, larger than delivery and far larger than payment, and it is the one line this paper could not price at all, because the marketplace returns HTTP 403.

So the correct reading of the €3.12 asymptote is: the published floor for a direct-to-consumer domestic order. A marketplace order costs more by an amount the market declines to state.

Discharges claims F027, F028, F029.

This is the third paper in this programme to measure Greek business infrastructure, and the three results rhyme in a way none of them was designed to produce.

Three different instruments, three different subjects, research publishing, technical hygiene, and pricing, and the same shape underneath: information that could be published, at no cost, is not.

It is not secrecy and it is not incompetence

In each case the withheld information is cheap to release and its absence is not obviously anyone's strategy.

A Greek software vendor is not hiding its prices to gain advantage over a competitor who also hides them. Neither gains. Both impose a phone call on a buyer who might otherwise have bought.

A Greek shop is not omitting a Content-Security-Policy to save money; the header is free. It is omitting it because nobody has ever looked.

A Greek trade association commissioning a survey rather than publishing its own operational data is not protecting a secret: it is doing the thing that requires no internal work.

The common factor is that publishing is optional and nothing checks it. Where something does check, the behaviour changes immediately and completely: cookie-consent banners appear on 67.7% of Greek e-shops because a regulator enforces them, while CSP headers appear on 18.4% because nobody does. The statutory payment cap in section 04 is honoured because it is law.

What this means for the price findings specifically

The transparency split in section 02 is not evidence that Greek software vendors are unusually secretive. It is evidence that the market has not made publishing a competitive necessity yet, and the comparison group in this very paper shows what happens when it does.

Every provider that published a price here, Papaki, top.host, Viva, BOX NOW, Shopify, Google, sells into a market where the buyer compares before purchasing. Every provider that did not sells into one where the buyer arrives through a relationship.

BOX NOW is the clean test. It is the newest entrant in Greek parcel delivery, and it is the only delivery provider tested that puts its prices in the HTML. The two incumbents, ACS and ELTA Courier, both publish a page called τιμοκατάλογος whose numbers are assembled in the browser and absent from the served response.

A new entrant publishes because comparison is how it wins. Incumbents do not because comparison is how they lose. That is a market dynamic, not a national characteristic, and it predicts that Greek software pricing becomes public the moment a credible entrant makes it a selling point.

The consequence is distributional and it is measurable

Where prices are published, everyone pays approximately the same and can find out in an afternoon. Where they are not, price is set in a conversation, and the party with least information and least volume does worst in that conversation, which is the micro-business that digital-adoption policy is aimed at.

This programme's third paper found that Greek AI adoption is rising among large firms and falling among small ones, and that the national average records the result as decline. The pattern here is a mechanism for exactly that: the fixed costs a small Greek business can discover are trivial, €157.50 a year, and the ones it cannot discover are mandatory. A large firm has a procurement function to make the calls. A one-person shop has an evening.

What would change it

Nothing in this section is a call for regulation of pricing, which would be a disproportionate response to a market that is simply young.

The cheaper intervention is the one this programme keeps arriving at from different directions: publish the thing that is already true. A rate card costs nothing to put on a page. A security header costs nothing to send. A trade body's own operational data costs nothing to release and travels roughly 70% further than the commissioned survey it currently pays for.

In all three papers the finding is the same and so is the remedy, and neither is expensive. The Greek market's information problem is not that the information is costly. It is that publishing it is optional, and optional things do not happen.

Discharges claims F030, F031, F032.

The instrument in this paper is not a crawler or a benchmark. It is a rule, and the rule is what makes the result checkable.

The rule

A price counts as published if a euro amount appears in the HTML the server sends to an ordinary visitor.

That is the whole test. It is deliberately strict and deliberately mechanical, and every consequence of it is visible in section 02.

What it accepts: a price in the page, a price in a table, a price in a "from €X" marketing line.

What it rejects: a price rendered by JavaScript after load, a price behind a configurator, a price in a PDF linked from the page, a price available on request.

Why that rule and not a looser one

A looser rule, "could a determined person find out?", makes every price published, since every vendor will tell you if you ask. It measures nothing.

A stricter rule, "is there a machine-readable rate card?", would fail almost everyone including the providers this paper found admirably transparent.

The rule chosen sits at the line that matters to the person asking the question: can a business owner find out what this costs, by looking, without identifying themselves? That is the difference between a market a newcomer can price and one they must enter first.

It has a real cost, and section 02 states it: ACS and ELTA Courier both do publish tariffs, to a human with a browser. They fail this test only because the numbers arrive client-side. That is a genuine difference from a vendor with no price anywhere, and the rule cannot see it, so the paper reports those two under a separate status rather than lumping them with the vendors that publish nothing.

The procedure, exactly

For each cost line:

1. Fetch the provider's homepage over HTTPS with an identifying user agent linking to the research. No login, no form, no crawl beyond the pages named.
2. Strip scripts, styles and tags, decode HTML entities, and search the remaining text for a euro amount: € adjacent to a digit, a digit adjacent to €, or the word ευρώ after a number.
3. If nothing is found, probe conventional pricing paths: /pricing, /timokatalogos, /el/pricing, /plans, /τιμές.
4. If a path returns HTTP 200, open it and read the <title> and first text before concluding a pricing page exists.
5. Record provider, item, amount, unit, transparency status, source URL and retrieval date.

Step 4 is the one that matters, and it was added after step 3 produced a wrong answer. Three probed paths returned 200 and none was a pricing page: softone.gr/pricing and softone.gr/plans return the homepage, and entersoft.gr/pricing returns a JPEG.

Without step 4 this paper would have reported that Greek software vendors maintain pricing pages containing no prices. The true finding is that the pages do not exist and the sites answer 200 for every path, which is a different and more accurate claim. A soft-404 is indistinguishable from a real page by status code alone, and status codes are what an automated check naturally reads.

What repeating it costs

About an hour. There is no crawl, no API key, no paid data source and no scale problem: twenty-five cost lines across perhaps fifteen providers, each one request.

The published table carries every source URL and retrieval date, so edition two is a diff rather than a repetition.

Repeating it in another market

Nothing in the method is Greek. The cost lines generalise directly, domain, hosting, certificate, platform, business email, card processing, delivery, accounting and compliance software, marketplace commission, and the interesting output is not the total but the ratio of published to unpublished lines.

For Greece that ratio is 18 published to 7 not, and the unpublished lines cluster in exactly one place: business software and logistics sold through a sales process. Whether that clustering is a Greek characteristic or a general feature of small European markets is a question this paper raises and cannot answer, because it measured one country.

Anyone who runs the same rule against another market can settle it in an afternoon, and the result would be worth more than this paper: a single-market finding about pricing opacity is an observation, and a two-market comparison is evidence.

The obvious extension, and why it was not done

The largest gap here, accounting, ERP and e-invoicing pricing, is closable by requesting quotes as an ordinary small business and publishing what comes back with dates.

That was deliberately not done, for a reason worth stating rather than hiding. Requesting a quote under a research pretext means entering a sales process without disclosing the purpose, and publishing the reply means publishing a commercial communication the sender did not expect to see printed. Both are defensible with disclosure and neither is defensible without it.

Doing it properly means identifying as a researcher up front, which changes the quote and therefore measures something different from what a real buyer would be offered. That is a genuine methodological problem, not an excuse, and it is why the gap in this paper is left open rather than filled badly.

The honest version of the extension is to ask the four vendors directly for a published rate card, on the record, and print whatever they say: including "we do not publish one". That is a request, not a pretext, and it is the intended next step.

Discharges claims F016, F017, F018, F019.

The bill is incomplete, and by an unknown amount

Every total in this paper excludes accounting software, ERP and myDATA e-invoicing, because no price for any of them is published. Those are not optional purchases; e-invoicing is a legal obligation for Greek businesses.

This paper does not estimate the missing lines. It would be easy to write "budget €300 to €800 a year" from experience and it would read as authoritative. It would also be exactly the kind of unsourced figure this programme's first paper was written to criticise. The gap is left open and labelled.

So the correct reading of the €157.50 floor is: the published portion of the fixed annual bill. Not the bill.

Published prices are not paid prices

Every figure here is a list price on a provider's own page. What a business actually pays differs for reasons this method cannot see: promotional first-year rates, multi-year discounts, bundled domains, negotiated card rates at volume, courier contracts priced on parcel count.

The direction of the error is not uniform. In the published half, list prices are close to paid prices: domain and hosting are transactional. In the unpublished half, the gap between list and paid is the whole reason there is no list.

Two specific distortions worth naming:

• "From" prices. Hosting at €5.70/month, VPS at €3.00, every BOX NOW figure: all are entry prices for the smallest configuration. Real usage costs more and the increment is not published in a comparable form.
• First-year pricing. Domain registration in particular is frequently discounted on first purchase and renews higher. Papaki's own footnote says the prices apply to new purchases. The €7.50 is a first-year price and the renewal is not stated on the same page.

VAT is not handled consistently

Greek business pricing is quoted sometimes with and sometimes without VAT, and the pages measured here do not follow one convention. Papaki footnotes that its prices exclude certain charges; Google quotes ex-VAT for business plans; Shopify's Greek page quotes in euro without stating the treatment prominently.

No VAT normalisation was applied, because doing so would require assuming a treatment the source does not state. A Greek business recovers VAT on business inputs, so the ex-VAT figure is usually the economically relevant one, but the totals in section 03 are sums of figures whose VAT treatment is not uniform, and that is a real limitation rather than a rounding detail.

One provider per line

Papaki for domains, top.host for hosting, Viva for payments, BOX NOW for delivery, Google for email. Each was chosen because it publishes a price, not because it is cheapest or typical.

That is a selection bias with a clear direction: providers that publish prices are competing on price, so the published figures are likely at the low end of the market. Combined with the "from" problem, the totals in this paper are floors twice over.

Prices move, and this is one day

Every figure was retrieved on 7 August 2026 and each carries its URL and retrieval date in the published dataset. Card rates change with regulation, courier tariffs change annually, and platform pricing changes without notice.

The statutory 0.50% cap in section 04 is the most durable line here because it is set by law, Law 5167/2024, Article 50, and the least durable are the platform subscriptions.

What the missing half would take to fill

Not much, and that is the point:

1. Request quotes from the four software vendors as an ordinary small business, and publish what comes back with dates. This is the obvious next step and it converts the paper's biggest gap into its most valuable table.
2. Render the courier tariff pages in a browser rather than fetching them, which would capture ACS and ELTA and make the delivery comparison real rather than one-sided.
3. Price the marketplace commission from a merchant-facing document rather than a blocked public page.
4. Add the registry and chamber fees, which are set administratively and should be citable to a public schedule.

Each of those is a phone call or a browser away. None of them is a phone call the person deciding whether to start a business should have to make before learning the price.

Two errors in this paper's method, recorded

The HTML-only rule called a properly published tariff "unpublished". An earlier draft recorded ACS and ELTA Courier identically, as couriers publishing a price you cannot read. Rendering both pages in a real browser showed they are opposites. ACS publishes a complete, dated, versioned rate table as a PDF linked from its price page: a more durable form of publication than an HTML table, and one this paper's rule cannot see. ELTA publishes no rate at all and invites a quote.

The correction moved eight ACS rate lines into the published column, changing the headline split from 18-of-25 to 26-of-32, and it added the only real comparison in section 04: a locker parcel at €3.00 against ACS point-to-door at €5.70 same-city and €11.20 mainland.

The rule was kept, because "is there a euro amount in what the server sends" is still the right question for a machine check and still the right proxy for comparability. What changed is that a browser pass is now part of the method for any provider the rule marks as unpublished. A rule with a known failure mode and a cheap correction is better than a looser rule with an unknown one.

The soft-404 error, recorded

The pricing-transparency check initially treated an HTTP 200 on /pricing as evidence that a pricing page existed. Three of the probed paths returned 200 and none was a pricing page: two returned the vendor's homepage and one returned a JPEG.

Had that not been checked, this paper would have reported that Greek software vendors publish pricing pages that happen to contain no prices: a weaker and wronger claim than the true one, which is that the pages do not exist and the sites return 200 for everything.

The check that caught it was opening the response and reading the <title>.

Discharges claims F020, F021, F022.

For someone deciding whether to start

The technical floor is about €13 a month. A .gr domain, shared hosting, a free certificate, self-hosted WooCommerce and one business mailbox come to €157.50 a year at published prices. Whatever is stopping a Greek business from selling online, at that number it is not the cost of the technology.

Budget separately for the half nobody publishes. Accounting software, ERP and myDATA e-invoicing carry no public price and are not optional. Get quotes before committing, because you cannot compare them any other way, and the smallest buyer negotiates worst against a sales process.

Self-hosting is €192 a year cheaper than the hosted equivalent: €216.70 against €408.70. Over five years that gap is close to a thousand euro, which is a real amount of developer time. Whether that trade is worth it depends on whether you have the time; it is a genuine trade rather than an obvious answer.

Email is the line that grows. It is the only published fixed cost that scales per person: €81.60 a year for one mailbox, €408 for five. At five staff, business email costs more than every other published fixed line combined.

For someone already selling

Check whether your card processing is using the statutory cap. Greek law caps domestic consumer-card fees at 0.50% on transactions up to €20 (Law 5167/2024, Article 50). On a €15 order that is €0.075 against €0.57 at the standard online rate: roughly seven times cheaper. If your basket skews small and domestic, this is the single largest published saving available to you and it applies automatically where it applies at all.

Delivery, not card fees, is your per-order problem. On a €15 domestic locker order, payment costs €0.075 and delivery costs €3.00: delivery is 97.6% of the combined cost, and the combined cost is a fifth of the order value. The public conversation in Greece is about card fees, which are capped and small. The uncapped cost is delivery.

Cross-border is a delivery arithmetic problem. Domestic locker delivery starts at €3. Thirteen EU destinations all show the same €15 headline. On a €60 order to Germany, delivery is 91% of the payment-and-delivery cost. Cyprus (€5) and Bulgaria (€3) are the exceptions, and both are next door.

For the sector and its regulators

The finding worth carrying out of this paper is not a price. It is that a Greek business cannot discover the cost of its legal obligations from public information.

Domain, hosting, certificates, platform, email, card processing and parcel delivery are all published, comparable and checkable in an afternoon. Accounting software, ERP and e-invoicing, the mandatory half, are not published at all by any of the four major Greek vendors tested. Neither incumbent courier publishes a machine-readable tariff. The dominant marketplace blocks automated access entirely.

That asymmetry has a distributional consequence. Where prices are published, everyone pays roughly the same and can compare in minutes. Where they are not, price is a negotiation, and the smallest buyer negotiates worst, which is precisely the buyer public policy on digital adoption is aimed at.

The intervention is not a subsidy. It is that the mandatory half should be as priceable as the optional half, and nothing about publishing a rate card for accounting software is technically difficult. The market has already demonstrated it works: the providers who publish are the ones competing.

For this paper

Section 05 lists four gaps and every one is closable: request quotes from the four vendors as an ordinary small business and publish the replies with dates; render the courier tariff pages in a browser rather than fetching them; find a merchant-facing marketplace rate card; add the registry fees from their administrative schedule.

The first of those would convert this paper's largest weakness into its most useful table: the first published comparison of what Greek business software actually costs. That it does not already exist is the reason this paper ends where it does.

The bill is €157.50 and counting, and the counting is the part the market will not do for you.

The whole bill, on one page

Everything this paper could price, and everything it could not, in the order a Greek business meets them.

Published fixed annual total: €157.50 at minimum, €216.70 for a realistic small business, €408.70 on a hosted platform.

Published marginal cost of a domestic order: about €3.08, of which delivery is 97.6%.

Unpublished: two of the mandatory pre-launch lines, the incumbent courier tariffs, and the marketplace commission, which for a shop selling through the marketplace is probably the largest per-order cost of all.

That is the answer to the question this paper set out to ask, and the shape of the answer is the finding. A Greek business can price its technology in an afternoon and cannot price its obligations at all.

Every price in this paper was read from the provider's own page on 7 August 2026 by an ordinary HTTPS request with an identifying user agent. No quote was requested and no salesperson contacted. Full URLs are printed as text so the PDF is standalone.

Prices found published

1. Papaki: domain registration. .gr €7.50/yr, .com €10.00/yr, .eu €5.50/yr. https://www.papaki.com/ The page footnotes that prices apply to new purchases; renewal pricing is not stated alongside.
2. top.host: hosting. Shared web hosting from €5.70/month; VPS from €3.00/month; dedicated from €89.70/month. A free domain is included with each hosting package. https://top.host/
3. Let's Encrypt: TLS certificates. Free, automated, no charge at any tier. https://letsencrypt.org/
4. Shopify: Greek pricing. Basic €19/month, Grow €56/month, Advanced €289/month, Plus from €2,100/month. Card rates from 1.9% + €0.25, falling to 1.8% + €0.25 on the Advanced tier. https://www.shopify.com/gr/pricing
5. Google Workspace: Greek pricing. Starter €6.80, Standard €13.60, Plus €21.10, per user per month on an annual commitment. https://workspace.google.com/intl/el/pricing.html
6. Viva: card processing. Online 2.19% + €0.24 per transaction; card present 1.69% with a €0.03 minimum; 0.50% for domestic consumer cards on transactions up to €20. https://www.viva.com/el-gr/pricing The 0.50% rate is cited on the page to Law 5167/2024, Article 50, amending paragraph 1 of the preceding article. It is a statutory cap, not a commercial offer.
7. BOX NOW: locker delivery. From €3.00 domestic, €5.00 to Cyprus, €3.00 to Bulgaria, and €15.00 to France, Germany, Italy, Belgium, the Netherlands, Spain, Austria, Czechia, Portugal, Denmark, Finland, Ireland and Poland. https://www.boxnow.gr/

Prices sought and not found

8. SoftOne Technologies. No euro amount in the served HTML. /pricing and /plans return HTTP 200 with the homepage's own title and content: a catch-all route, not a pricing page. https://www.softone.gr/
9. Epsilon Net, including Pylon. No euro amount in the served HTML of either page. https://www.epsilonnet.gr/
10. Entersoft. No euro amount in the served HTML. /pricing returns a JPEG image. https://www.entersoft.gr/
11. ACS Courier: domestic tariff. HTTP 200, 24,192 characters of text, zero euro amounts: the tariff is assembled in the browser and absent from the served response. https://www.acscourier.net/el/ypiresies/taxymetafores-eswterikoy/timokatalogos/
12. ELTA Courier: tariff. HTTP 200, no euro amount in the served response. https://www.elta-courier.gr/timokatalogos
13. Skroutz: marketplace commission. HTTP 403 to an identified research client. No public rate card was reachable. https://www.skroutz.gr

Data published with this paper

14. Broikos, N. (2026). Greek SME digital bill: price and transparency table. 25 cost lines with provider, amount, unit, transparency status, source URL and retrieval date. https://broikos.gr/research/data/paper-07/greek-sme-prices.csv

Companion papers

Per-order landed cost from published tariffs is modelled in What a Greek online order actually costs. The technical state of 900 Greek shops, including the Shopify and WooCommerce shares referenced in section 03, is measured in The Greek e-shop technical audit. Both are listed with full URLs at the end of this document.

---

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. What a Greek online order actually costs https://broikos.gr/research/what-a-greek-online-order-costs.pdf What the €36.1bn e-commerce figure counts and what it does not, the convergence of Greek and European online buying, and a landed-cost model built from published tariffs rather than quoted rates.
3. 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.
4. 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.
5. 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.
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.