E-Invoicing: The Hidden Toll on France’s Dormant Businesses

One number ought to haunt the corridors of the French Treasury. In the second quarter of 2025, according to figures released by URSSAF in late January, only 49.8% of administratively active micro-entrepreneurs reported any revenue at all. Out of 3.19 million registered accounts, roughly 1.6 million invoiced nothing whatsoever during the period. Half the scheme is an intermittent France: trial ventures, side income, businesses in transition, companies that exist legally but live economically for only a few weeks a year.

This dormant half is what I want to write about. France’s e-invoicing reform, whose true nature as real-time tax surveillance I have described elsewhere, was designed for businesses that invoice. It was never designed for businesses that exist without invoicing. And that is precisely where it breaks.

First, Let’s Dispose of the Easy Argument

Let’s start by dismantling the objection you read everywhere, one I was tempted to make myself: sole traders will have to pay a certified platform every month, so they will simply shut down.

That is wrong, at least in its crude form. The Great Invoice Toll, an investigation published by l0g on August 22, audited nineteen certified platform offerings and found thirteen with an unambiguous free tier. Qonto, Dougs, Indy, Abby, and Tiime all advertise free plans, some with no volume cap at all. A micro-entrepreneur issuing three invoices a year fits comfortably inside them. Anyone who builds a critique of the reform on the mandatory-subscription argument will have that audit thrown back at them, and rightly so.

The problem lies elsewhere, and it runs deeper. The toll is not monetary. It is structural.

A Permanent Obligation on Empty Shells

Consider the timeline. From September 1, 2026, every VAT-registered business, including micro-entrepreneurs below the VAT threshold, must be able to receive electronic invoices. Issuing them follows for the smallest businesses on September 1, 2027. Being able to receive means being connected to a certified platform and listed in the central directory, the one part of the public portal the state kept when it gutted the rest.

Now read that again with the dormant half in mind. A business that has invoiced nothing for eight months, purchased no services, and may be waiting for work that never comes must nonetheless pick a private provider, open an account, and appear in the directory. Not in order to do anything, but in order to be capable of doing it, just in case.

Let’s be precise about the nature of this burden, because this is where the analysis has to be honest. For a completely inactive business, with no purchases and no incoming documents, the receiving obligation amounts to a single upfront step: signing up somewhere so as to appear in the directory. As long as no invoice circulates, nothing happens. The cost, then, is not operational. It is existential and cognitive. You have to work out what a certified platform even is, compare offerings you have no basis for judging, open an account with a provider you do not need, and commit to an ecosystem whose rules keep shifting, all for an activity that produces nothing. This is not a recurring burden but a pure barrier to entry: you pay it once, in attention, in uncertainty, and in a sense of absurdity, at exactly the moment you are wondering whether the registration is still worth keeping. And barriers to entry never filter out the large players. They filter out the hesitant, the intermittent, the maybe-one-day. Which is to say, precisely our population.

There is no fallback, either. By gutting the public portal in October 2024 and reducing it to a directory and a tax-data concentrator, the state did more than withdraw a service: it handed the very function of access to the private market. There is no longer any point of entry that is not commercial. The intermediary requirement is locked in by design, and the dormant business has no option of dealing with the tax authority rather than with a vendor.

The reversal here is easy to underestimate. Until now, the micro-entrepreneur scheme had a certain elegance: nothing cost nothing. No revenue, no contributions, a quarterly declaration of zero, and that was that. The registration could sleep in a drawer, ready when needed. The reform adds mandatory membership in a commercial ecosystem, imposed on an activity that does not exist. The obligation never lapses, even when the activity does.

Free Tiers Don’t Love You When You Sleep

The obvious reply is that the free tier settles the matter: the dormant business signs up for a zero-euro plan and forgets about it. That reply ignores the economics of that zero.

The l0g investigation makes the model plain: these free tiers are loss leaders. Qonto wants to sell you a business account, Dougs accounting services, Indy tax filings and support, Abby banking and payment chasing, Kolecto distribution for Crédit Agricole. Electronic invoicing has three rare commercial virtues: it is mandatory, it recurs every month, and it carries exactly the data that feeds accounting, payments, and lending. Giving away the regulated transport layer is a cheap way to buy a customer relationship and sell everything else.

So what, in that equation, is a customer who never invoices worth? Nothing. No flows, no upgrade to the business account, no accountant on retainer. Pure support and infrastructure cost with no return. The dormant business is structurally the customer that freemium will eventually push out or start charging.

The signals are already visible. Kolecto limits its free plan to the first twelve months for newly created businesses. Pennylane restricts its own to micro-enterprises under an annual ceiling. B2BRouter counts twenty-four transactions a year, incoming invoices included. Pricing tables shift, promotions expire, counters tighten. Today’s zero is a snapshot, not a commitment. And on the day a dormant user’s platform decides that segment is no longer worth serving, that user will discover that switching platforms is a regulated procedure, with notice periods, a directory entry to update, and a commercial contract to unwind. To manage zero invoices.

What the Dormant Half Will Do

Put yourself in their position. You keep a dormant registration because it costs nothing. You are now told to pick a private operator, get connected, track changes to its pricing, and start over if the free plan disappears. All for an activity that brings in a few hundred euros in a good year.

The calculation takes about ten seconds, and it has only three outcomes. They do not apply to the same people, and the distinction matters.

For the truly dormant, those who have invoiced nothing for a long time, the outcome is deregistration or abandonment. URSSAF already strikes off accounts automatically after twenty-four consecutive months of zero revenue; the new barrier will accelerate that, this time through discouragement. Hundreds of thousands of registrations will close, not because an activity ended, but because keeping the option open has become a chore. And when the registration goes, so does what it protected: the possibility of waking up again. The retiree who sold expertise three times a year, the student testing an idea, the employee keeping a door open to self-employment will all quietly shelve the project. This is the most silent form of destruction and the most expensive: the loss of the option, the trial run, the maybe.

For the barely intermittent, those taking in a few hundred euros in good years, a third path opens: going grey. They have real transactions to make, and for them the calculation tips. The occasional job that would have produced a declared invoice will now be settled hand to hand. That is the central irony of a reform sold as a weapon against VAT fraud: by raising the cost of compliance for the smallest players, it makes going off the books relatively more attractive in exactly the places where fraud was marginal. Guitar lessons, weekend IT repairs, photo retouching on the side will slip out of the system.

In all three cases, the state loses. It loses contributions, it loses VAT, it loses statistical visibility into self-employment, and it loses the very thing the 2008 scheme got right: bringing into the legal economy activities that would never have entered it otherwise.

A Laboratory for Managed Dependence

This reform is part of an architecture I have documented article after article: real-time tax surveillance whose flows pass through a hundred-odd private platforms that constitute as many targets and amount to an intelligence trove handed to our rivals, all of it operated by a state that, as the DGFiP breach demonstrated, demands everything and protects nothing. I won’t rehearse that here. What interests me is what this architecture does to the dormant case specifically.

In a system built to make flows legible, the intermittent business is an anomaly. It slips through the mesh: no regular flows, no usable data, no predictable behavior. The system does not know what to do with an activity that exists only in flashes. So it fits it with the same harness as everyone else, never mind that the harness outweighs the horse.

I don’t credit the Treasury with any intention of killing off dormant micro-enterprise. I simply note that nobody, in any impact assessment, asked what happens to the half of the scheme that doesn’t invoice. They counted businesses, not activities. They reasoned in terms of administrative stock, not economic reality. That is technocracy’s signature move: it sees the boxes, never the lives that only half occupy them.

What Should Have Been Done

A solution existed, and it was simple: a de minimis threshold. Below a certain number of invoices per year, say a dozen, a business could have met its obligations directly through the public portal, free of charge and without a private intermediary. That was, in fact, exactly the promise of the free public gateway announced in 2020, conceived as the social half of the bargain, and it is precisely what the Treasury sacrificed in 2024.

Failing that, one minimum demand remains: that the Treasury publish a standardized comparison of the available offers, spelling out exactly what the free tier covers, what the counters are, what overages cost, and what it costs to leave. The state compels you to choose a provider; it should at least give you the means to choose. It has not, and that silence says a great deal about its real priorities. The tax concentrator works, the directory works, and the rest is the market’s problem.

A Scheme That Dies at the Edges

French micro-enterprise will not die by decree. It will die at the edges, through the attrition of its silent half, the half that does not protest, does not organize, and counts for nothing in consultations. Each deregistration will be individually rational and collectively disastrous. Five years from now, people will express surprise that the scheme has a million fewer accounts, they will call it normalization, and nobody will connect it to the mandatory connection deadline of September 2026.

The genius of the micro-entrepreneur status was that it made entrepreneurship reversible. You could try, fail, go dormant, and start again without the administrative machine billing you for existing. E-invoicing ends that reversibility. From now on, existing has a price, if not in euros then in connections to maintain, accounts to open, and pricing tables to watch.

A society that taxes its entrepreneurs for sleeping should not be surprised when, one day, they stop waking up.