GP2 rewards low-touch support. It’s the same fake-EBITDA logic that runs through the GPSR and DPP pieces on GP1, relocated one layer down the waterfall. Low ticket volume, a high self-service resolution rate, thin support headcount relative to order volume — all of it reads as operational efficiency in a QoE model. Often it is. Sometimes it isn’t efficiency at all. It’s the customer giving up.
When “Low-Touch” Support Means “No Route Out”
The pattern is specific: no phone number listed anywhere on the site. A chatbot whose decision tree has no terminal branch that reaches a human, a return request, or a cancellation — every path loops back to the same three menu options. A contact form that submits into an inbox nobody is resourced to clear. From a support-ops dashboard, this is indistinguishable from genuine self-service efficiency: cost per order is low either way. The difference only exists outside the P&L, in what a customer actually experienced trying to exercise a right they’re legally entitled to.
The Legal Hook: Two Separate EU Provisions, Same Underlying Problem
Two distinct rules bear on this pattern, and they’re worth keeping separate rather than treating as one claim.
The first is Article 6(1)(c) of the Consumer Rights Directive, which requires a trader to disclose the means of communication it makes available. The Court of Justice settled a live question about this in 2019: in Bundesverband der Verbraucherzentralen und Verbraucherverbände v Amazon EU (Case C‑649/17), the Court held that a trader is not obliged to provide a telephone number specifically. What it did hold is that whatever channels the trader offers must let the consumer contact it “quickly” and communicate “efficiently” — a functional standard, not a channel mandate. A chatbot can lawfully replace a phone line. A chatbot engineered with no path to a human, to a return request, or to a cancellation is on considerably weaker ground claiming it meets a “quick and efficient” standard it was specifically built to defeat.
The second is Article 9 of the Unfair Commercial Practices Directive — the original 2005 text, not an Omnibus Directive addition, contrary to how an earlier draft of this piece framed it. Article 9 treats “disproportionate non-contractual barriers” a trader imposes on a consumer who wants to exercise a contractual right — including the rights to cancel or to switch to another trader — as a factor pointing toward an aggressive commercial practice. The European Commission’s 2021 interpretive guidance on the UCPD applies this concept directly to switching-barrier conduct. A support structure engineered so that reaching a human, lodging a cancellation, or requesting a return is functionally impossible fits the description the provision was written to catch.
Both citations above have been checked against the primary text. What still needs a lawyer’s sign-off before publishing is narrower: how national courts and consumer authorities are currently applying the “quick and efficient” and “disproportionate barrier” standards to chatbot-only support structures specifically — neither provision was drafted with chatbots in mind, and enforcement on this exact fact pattern is still developing.
Why Automation Makes This Structural, Not Incidental
Whether the missing human branch in a bot flow was deliberately engineered to suppress cost, or simply never built because nobody prioritized it, doesn’t change the exposure. Liability under both provisions above turns on effect, not intent. A support structure “optimized” by removing its most expensive branch — a human who can actually resolve something — is optimizing precisely the thing these rules target.
Where the Hidden Cost Actually Lands
- Live, developing enforcement exposure. Neither underlying provision is new, but their application to chatbot-only, human-free support structures is an active and still-developing area — recent CJEU case law and current Commission guidance both bear directly on this fact pattern, not a fringe theory.
- A GP3 bleed, not just a GP2 one. Reviews describing unreachable support suppress conversion and inflate CAC on the marketing layer — an obstruction pattern taxes two waterfall layers simultaneously, not one, and neither shows up labeled as the other’s cause.
- A margin cliff at close, not before it. Once the acquirer staffs real, escalatable support — because carrying the obstruction is now a risk they personally own, and because it’s reputationally untenable under new ownership — the GP2 support cost the target modeled rises. The “efficiency” was a deferred liability whose maturity date just moved to whoever owns it next.
What Outside-In Analysis Can Detect Before the Data Room
- Calling any listed number, at the hours the site claims to be staffed
- Running the chatbot flow specifically trying to reach a human, request a return, or lodge a cancellation, and recording whether a terminal path to any of those exists at all
- Reading reviews filtered for “no response,” “can’t reach anyone,” “bot loop,” “ignored my email,” “no phone number”
- Checking whether a functioning cancellation or return-initiation mechanism is actually reachable in the live flow, not just described in policy text
- Cross-referencing marketplace or app-store reviews (where the brand also sells) against the support channels claimed on its own site, for consistency
The Pre-LOI Question Every PE Fund Should Ask
Not “what’s the support cost per order” — that’s exactly the number a target running this pattern will show you looking its best. The question is: can a customer who wants to cancel, return, or complain actually reach a human being, and how many attempts does it take to find out?
A target with genuinely low-touch, well-resourced support has a real efficiency. A target with a structurally unreachable one has a liability wearing an efficiency’s clothes — invisible in the cost line, and fully inherited at close.
This analysis is part of Tronvik’s GP2 Fulfilment & Service Margin pillar. Nothing in this article constitutes legal advice. To initiate an outside-in customer-service obstruction screen on a specific acquisition target, contact info@tronvik.com.