Telco Automation — Fighting Robocall: examining ARCEP decision 0881’s impact for telco professionals


Content Strategist

23 min reading time

Picture of a red cross over a hand on a phone and robot

Why do telco regulations fail? Enforcement is frequently undersized, inefficient and overlooked — Proof by example with ARCEP decision 0881 fighting robocalls

How to manage a common good efficiently? While the question might seem more like an environmental concern, it’s a very real one in the world of telecommunications. Standard resources shared among providers are commonplace, we find them everywhere: consider, for instance, phone numbers, URLs or IP addresses.

📑 On the go? Download the PDF version (1.4 MB)

Such situations usually lead to a detrimental scenario where each actor, driven by his best interest, tend to over-exploit the resource —a disservice the whole group including himself. It is known as The Tragedy of the Commons.1

It was about time for the ARCEP to have a look at the numbering plan, the main focus of decision 0881. In the English press release announcing the decision3, the ARCEP states its goals: to foster innovation and support new uses, and to improve how the scarcity of numbering resources is managed.

ℹ️ The ARCEP (“Mail and Communication Regulatory Body”2) is the French government agency tasked with drafting communication standards. Like the FCC (Federal Communications Commission), its US equivalent, its primary mission is to regulate the market and ensure fair competition between all the service providers and actors.

Hence, the decision addresses the future shortages in some regions (French Réunion) and adds additional restrictions while using geographic numbers. However, the decision is broader than a mere numbering policy update: it also instates additional protections for end-users (-> requirements for businesses) against automated calls and text messages.

Since we won’t delve into the details of the numbering plan change, let’s cover the main elements here:

  • The previous framework was severely outdated (from 2001/2005)

  • The need for phone numbers keeps increasing, especially with the rise of “Machines to Machines” (M2M) traffic.

  • It was possible, somehow legal, and too easy to conceal the caller identity (businesses) for automated phone calls.

1/ The two sides of a call

Regulating telecommunications is a game of balance. Indeed, the regulatory agencies must find an equilibrium between on the one hand the needs of the business and on the other hand the expectations (and frustrations) of citizens and customers. The tensions between the two are numerous. Moreover, businesses are obviously better organized (lobbies) and funded that citizens, which gives them more weight to make their point to the regulators.

Looking for a VoIP service provider?


The dichotomy is real though, and quite often businesses must be protected… from themselves. Indeed, without any regulation we would quickly land in another tragedy of the typical scenario: companies would be flooding automated phone calls and text all day long, to the point where people will find any way to avoid the harassment: robocalls blocking apps, whitelist apps, switching numbers, or even getting rid of phone communications (voice/SMS) altogether.

tragedy of the commons - robocall, telco automation, arcep decision 0881

So, instead of falling in the trap of portraying the interests of the businesses and the citizens as opposed, we try our best to see the shared interest: in matters of telecommunications, citizens’ and businesses’ interests are aligned, up to an extent.

Indeed, when we map the main concerns of the two parties, the middle of the spectrum representing shared concerns is quite apparent:

user and business interests mapping - telco industry - robocall, telco automation, arcep decision 0881

Mapping businesses and users interest in the telecom industry

Let’s not forget the third party, the intermediary: operators

Finally, it’s worth remembering that there is a third side to a call: the telecom operators themselves. They have a commercial relationship both with the end-users of their networks (consumers) and the businesses using their network for legitimate reasons as well as spamming.

Considering they make money on both sides, it could be tempting to jump to the conclusion that operators are not eager to do much to fight robocalls as it would hinder their revenues. In reality, the telecom operators are like a scale, they must manage their common (and shared) infrastructure. Indeed, excesses on one side drive excesses to the other, as will see below in section 1-B/ Rage Against the Machine.

1-A/ Fostering Innovations for Businesses

Quite often, the needs and wants of businesses and consumers align. Consider automation for instance: while companies and consumer see different benefits behind the term, both tend to agree on its interest.

  • For consumers, automated support, for instance, means they will be able to get an answer in a timely fashion, at any time of the week or day. There might also enjoy not having to talk to a human to solve their issue. Finally, automated support is usually used not to kill the agent-based one, but to optimize it. With proper implementation of automated support technologies, a business can free more time of its agents to handle edge cases, where their help is necessary.

  • For businesses, automated support comes with many advantages. First, it means that everything goes according to a script — it’s a tremendous upside, especially for large brands who are very sensitive about managing their image. Secondly, since everything happens on a digital support (chatbot, IVR, others) and follows a script, it means that the whole customer support activity becomes traceable and measurable.

Once again, the devil lies with the details. Real-world implementations of automated support technologies are often lead to a different approach. Some businesses might be sorely focusing on automating most/all customer support activities, or reducing the cost of their support department and need for human agents. Such implementations are absurdly common, despite being detrimental to both the brand’s image and customer satisfaction.

Besides, we’ve considered support so far (=inbound calls). If we want to paint the full picture, let’s not forget about the automated outbound call campaigns that usually concentrates the frustration of consumers.

1-B/ Rage Against the Machine?

For consumers, automation is a very different reality depending on the modality. When it’s inbound (the user reaching to the bot/server), it’s usually welcomed: it means better accessibility (24/24, 7/7) and the service is generally offered alongside regular support, with human agents.


Even in this optimal scenario we already see many frustrations emerge. Users tend to find tricks to avoid interacting with the rigid and demanding machines. For instance, if it’s an IVR system (Voice), they yell “Agent, Agent, Agent” or “Human”, sometimes “0” or keep smashing the pound (#) button — whatever works to get them to a real human, usually better suited to help them. We’re now seeing resources emerge to help customers in need of help to find the phone number of the service in question, and how to bypass the automation to reach a human agent.

Manage incoming calls in real-time to make the most of your agent's time.


Picture of a red cross over a hand on a phone and robot

Automated Outbound Campaigns (Robocall)

Outbound automated call and text campaigns usually make for the bulk of end-users’ frustrations. Indeed, the techniques are now more diverse than ever (phone call + recorded voice message, text message campaigns, ringless voicemail) and accessible. Considering the historical decrease in the price of a minute/text, we had a recipe for a perfect storm of automated phone and messaging activities.

Finally, it would be misleading to ignore certain businesses’ trickeries as they concentrate the frustration of end-users:

  • “Flash calling”: calling a number for just a ring, so that the prospect would call back.

  • Obfuscation: different methods are used to make a phone call feel local while, in reality, it’s coming from abroad (More on this below)

  • Repetition: once a number lands into a given prospect database, it tends to get shared profusely. Some end-users receive up to 5-10 automated calls or messages a day. Surprisingly, even telecom carriers themselves are true adepts of the practice.3

1-C/ Do we even need phone numbers anymore?

Before we dive into the analysis of the decision, we need one more element of context. The world of communications evolved rapidly over the last couple of decades, let’s account for these changes with a quick overview. Indeed, OTT messaging apps such as Wechat, Whatsapp or Messenger (Facebook) are quickly gaining traction both in terms of users and usage:

ℹ️ Over-the-Top (OTT) messaging apps refer to apps allowing direct communication directly through the Internet, therefore bypassing telecom providers.

messaging-app-users - robocall, telco automation, arcep decision 0881

While some of these apps are still tied to a phone number (you can’t use Whatsapp without one for instance), others use email or nicknames for identification, like Messenger (Facebook account), Kik (nickname), or Hangouts (Google Account).

Direct messaging app are only one part of the new world of communications, let’s not forget supports used by communities, either small or large. Discord is probably leading the band, along with Slack (used by some communities despite its work focus), Gitter, and many others. Indeed, Discord celebrated its third birthday last year with great numbers demonstrating a very fast-paced growth:

discord 3 years stats - robocall, telco automation, arcep decision 0881

While a life without a phone number was highly unpractical 10 years ago, it’s now not only totally doable, but also a reality, especially for the younger generations: they still have a mobile plan, but a large share of their usage is data-based, and they don’t even need to share their numbers anymore. Indeed, the service offering is now dense enough for them to have several ways to chat (text based), call, or video call anyone in their network:

mobile is the new prime time - robocall, telco automation, arcep decision 0881

Source: Think With Google – Gen Z: A Look Inside Its Mobile-First Mindset

ℹ️ We are obviously overlooking MANY huge players in this overview, like Stream or Snapchat. The goal is more to give you a general sense of what’s happening with OTT messaging apps rather than being exhaustive.

With enough context, we can now jump in to the overview of ARCEP decision 0881 before discussing its potential consequences.

Smarten your business communications


2/ Decision n°0881 Overview

ARCEP Decision 0881 is quite a large package: the original document is over 100 pages long. If your business is a contact center or using automated voice or text marketing campaigns and active in France, we strongly recommend you to read the full paper. We won’t dive into each item; instead, we will analyze a set of decisions which embodies the spirit of the text.

2-A/ A tighter framework surrounding caller/sender identification

ARCEP Decision n°0881 - Article 7.4.

ℹ️ Sender ID / Caller Line Identification (CLI) refers to the number/string displayed on the callee’s phone. So far, businesses were able to use several tricks to mislead end-users by spoofing their numbers. For instance one of the most common practices was to use a local number to deceive and increase pick-up rates, even when calling from abroad.

The decision gives a precise definition of what “CLI/Sender ID modification” is:

The caller line identifier is considered "modified" when the number displayed on the end-user phone (the callee) cannot be used to reach back to the caller.

So far, the framework surrounding the practice was slim, so this decision defines conditions to regulate it, the main and most novel ones being:

  1. The two-way imperative: if you use a certain number to call (or text) a prospect, the prospect must be able to contact you back on the same number, as long as the campaign is going.

  2. The principle of territoriality: while using both geographic and non-geographic numbers (for instance French mobiles 06/07 XX XX XX XX or geographic 01-5 XX XX XX XX numbers), the caller/sender must be able to guarantee that the agent is indeed calling from France. Art 7.4.2.b.

2-B/ A new approach to regulate automated calls and text messages

ARCEP Decision n°0881 - Article 7.5.

The ARCEP lists the main nuisances and abuses observed with automated campaigns and establish a framework to regulate the activity.

Additional restrictions for automated communications (robocall)

Overall the section is quite dense, so let’s focus on the most interesting dimensions:

  1. Automated Call Center Dialers are explicitly mentioned in the list of nuisances established by the ARCEP. Their legal status remains unclear. We contacted the ARCEP for clarification, but it turns out their explanation was still unclear.4

  2. Several additional rules are established, limiting what types of number can be used for automated campaigns and imposing transparency on the caller (see the previous section). Automated systems can no longer use “territorialized” numbers (see exceptions below).

  3. Finally, three main limitations are established — they are very insightful to analyze as they translate what kind of automated traffic the ARCEP sees as beneficial on the network.

The document introduces the notion of “territorialized” numbers (“numéros territorialisés”), these include:

  • Geographic numbers: 0[1-5] XX XX XX XX in France.

  • Mobile numbers: 06/07 XX XX XX XX.

  • Non-geographic numbers (⚠️ ≠ mobiles): 09 XX XX XX XX. These numbers are no tied to a specific region.

Therefore, automated traffic is now contained by default on one type of numbers: Long virtual mobile numbers — 07 XX XX XX XX XX XX (14 numbers including extension). These numbers are new and used mostly for automated or for machine communications, both machines to humans and machines to machines (M2M).

The three exceptions are the following:

  1. Small Scale: Automated systems calling/texting five different numbers or less.

  2. Inbound/Outbound Texts Balance: automated text messaging systems which receives as much or more texts as they send.

  3. Inbound/Outbound Calls Balance: automated calling systems which receives significantly more calls than they initiate.

CALLR has all the numbers and services you need to reach France, in self-serve!


The type of automated traffic the ARCEP wants to see

The ARCEP introduces an exciting new paradigm here: automated communication activities should be balanced between inbound and outbound. It makes much sense: why would you block a high volume chatbot to which users voluntary come to talk to? On the opposite, having a very law inbound/outbound ratio is usually a good tell for spam or intrusive behaviors.

It’s interesting to note inbound/outbound balance exception is more lenient with text that voice — it translates the ARCEP will to support chatbots while also cracking down on robocalls.

⚠️ Note: The decision clearly states that Calls that are initiated manually (without using an automated numbering system) by agents in a contact center ARE NOT regarded as emitted by automated systems. (Decision 0881 – Footnote 33)

While having a base principle to determine what kind of traffic is preferred by the authority, we can already see its limit. Here are two very telling situations to consider:

  1. Automated critical alerting traffic: consider a seismology monitoring system made to alert local populations if a life-threatening event is detected. Such systems generate intense outbound traffic (A LOT of phone calls to many people over a short time-frame), so they violate the outbound/inbound balance principle, despite literally saving lives.

  2. Debt collection services generally have most of their traffic outbound, since they usually need several calls for the debtor to reach back to them. The principle would make their service potentially illegitimate, despite providing real utility.

2-C/ Enabling new and innovative uses

The ARCEP made several decisions to support innovation and enable new usages. What follows is an overview:

  • Make it easier to use several phone numbers on one mobile device. It will become possible to have several numbers for one single subscription, which could, for instance, allow people to have a professional and personal number redirecting to the same line/device.

  • Enable SMS/MMS capabilities for geographic numbers. So far, SMS and MMS were restricted to non-geo numbers in France. However, the details of the implementation are not known yet.

  • Make it easier to use long virtual numbers for IoT devices, even for users out of France.

ℹ️ ARCEP Decision 0881 was published last year, July 2018. However, considering the changes made to automated phone calls (and the need for service providers to adapt), the effective date of enforcement is slated for August 2019.

Now that we’re clear on the gist of this decision, let’s dive into the impact it will have on the industry.

3/ Insider’s Perspective — discussion with Taoufik Zagdoud, CALLR’s CEO

Analyzing new regulations is always exciting, at least to me. However, if you wonder about the potential impact of a given law, the industry experts are your best resource. I am lucky enough to sit next to one, Taoufik Zagdoug, CEO of CALLR, a French smart voice services provider active since 2009. We discussed ARCEP Decision 0881 and its impacts over the industry:

What was your first reaction after reading ARCEP Decision 0881?


Taoufik Zagdoud is the CEO and co-founder of CALLR

Taoufik Zagdoud: As a VoIP carrier, we’re in an interesting position. Historically, legacy carriers tended to oppose additional regulation, as they perceive it as a risk for their businesses. However our situation is more balanced: first, we recognize and understand that a robust legal framework is both needed and beneficial for the future of our industry.

Secondly, we take great care in nurturing our relationships with our customers so that we can share our expertise with them and help them to adapt to the new regulations. The best practices are known and documented; what we need now is to evangelize businesses on the necessity of preserving mobile as a highly qualitative and reactive communication medium. We frequently produce guidance ressources5 to keep pushing the adoption of best practices.

There are currently no reliable ways to filter traffic beforehand, that is without assuming that everybody is suspicious and/or guilty. We work the other way around: we monitor our network very closely and make it as easy as possible to report obnoxious activities, with a dedicated service:

Will it be sufficient to contain the spam and abuses proliferating on the networks?

TZ: My experience is enough to know that it won’t be enough. ARCEP decision 0881 is still a needed addition, as the text defines and regulates some concepts and technologies that were currently evolving in a legal grey area. It’s also forward-thinking, considering it paves the way for new use cases, with, for instance, the attribution a range of numbers for machine-to-machine communications.

Enough with the kudos, let’s talk for real. Neither the form and substance are sufficient there, and let me explain why:

  1. First, once again, authorities demonstrate a form of wishful thinking. Establishing a principle to ban a practice is one thing, but how do you enforce it? Usually, they tend to think that they can jump from principle to implementation with the right tech — and once again it’s not realistic there. Allow me to scratch the surface of this to illustrate my point: what is “automated traffic”? How do you automatically detect it? With the virtualization of telecoms, it’s simply impossible. There are so many ways to spoof one’s identity or disguise automated traffic as human-initiated that whatever the technical solution is, I know for a fact that it will be highly game-able and indeed gamed.

  2. Reading through the decision, I feel like automation is depicted as an evil and despicable practice. There are many use cases providing real and even some time-critical utility, despite being automated: critical alerting systems, non-profits using automation to help homeless people find food6, or even information/digest chatbots.

  3. Finally, I feel like this whole discussion took place in a fancy Parisian office, where the realities of the profession were forgotten. Currently, most of the call centers serving “developed countries” are set up in neighboring countries which have low wages (Tunisia -> Europe | India/Pakistan/Philippines -> the US, and so on). These call centers are serving major customers such as banks, insurances, or telecom providers. This is why changing telecom regulation have a real, practical impact on the job market of these countries. Let’s say we live in a magical world where we have the tech to block automated phone calls altogether — what do you think will happen? The way I see it, there will be a bigger need for human agents, which will drive the call centers to find countries where they can push their overhead (per agent) even lower. Do we really wanna enter this race to the bottom?

Global Voice API for smarter interactions


How is the industry reacting to the automated calls and texts crackdown?

TZ: Among my peers and within the industry, I hear many concerns regarding the future of automated call campaigns and contact center activities. The crackdown on predictive dialing, if real, would have a huge impact. Some go as far as saying that the enforcement of the bill, in August, will jeopardize the future of the whole industry.

I do not share this view. To me, the current situation feels like the pre-GDPR panic last year; we see massive expectations and worries about what will happen. I’m convinced that at the end of the day the impact will be way more moderate than expected. At least with GDPR, that’s what happened: first, considering all the exceptions, many businesses naturally found loopholes to avoid having to comply altogether. Else, even those who cannot find a suitable exemption are not too worried.

So far, about a year after its promulgation, the enforcement of GDPR has been abysmal. By January, the estimates gave more than 100 000 violations, with 59 000 data breaches reported by companies… for a total of 90 fines 7. Indeed, since GDPR was not enough to fight against phone harassment, the ARCEP included additional restrictions in decision 0881: it’s no longer possible to use a mobile CLI (stating with 06 or 07) for automated campaigns. The situation seems to be the same with other regulations, such as the US Robocall: the fines are simply not collected — The FCC Has Fined Robocallers $208 Million. It’s Collected $6,790.

The main threat might lie with the activities of consumer groups. The fear of generating a consumer group-initiated public backlash might have even been enough to nudge business towards sanitizing their practices.

La Quadrature du Net initiated a legal procedure against Facebook, Google, Apple, Amazon, and LinkedIn the very day the law was instated, May 28th, 2018. More than 12 000 citizens supported it. It is already producing results, with Google being sentenced by the CNIL (French data protection agency) a 50 million € fine8. The four other lawsuits were forwarded to the regulatory bodies of other European Countries.

GDPR is not without merits, and I think its most significant one was to re-empower the consumers in a way. It reminded them that they have rights over their data, and they can make them acknowledged.

quadrature du net poster - robocall, telco automation, arcep decision 0881

ℹ️ La Quadrature du Net is a French NGO fighting to preserve users privacy on the Internet

How do you approach regulation while designing new CALLR services? How do you account for all the different legal frameworks that you potentially have to comply with?

TZ: We have offices both in France and in the US, so we follow regulations coming both from the ARCEP and the FCC very closely. We serve the whole world anyway (points of presence in Asia and Africa), so we thrive to keep up to date with major changes coming from any country. For instance, while we were developing SENDR, our voice/text campaign management tool, we drew inspiration from FCC/FTC regulations. Overall, we try to raise our compliance level, quite often even above the law. For instance, SENDR handles STOP messages at the platform level, so we offer this feature in any country we serve, including those where it’s not a legal requirement.

Overall, SMS traffic tends to be more and better regulated than voice. Many countries have very restrictive legal framework surrounding text messages, including the United Arab Emirates — shortcode usage and the content of the campaign must be validated beforehand — or China where routes are cut if spam is detected on them. As a rule of thumb, I observe that countries, where the authorities react swiftly and with force (hefty fines/termination/ban) to abuses directly against the telecom operator, the practices evolve quickly, as the operators themselves eventually filter the undesired traffic.

The reason for the disparity between voice and text regulations (and their enforcement) lies in the very nature of the traffic. Text messages are easier to analyze automatically, which eases the spam detection process. Besides, the STOP system is very common and makes it even easier to detect abusive traffic (-> companies who still contact their prospects despite receiving a STOP). These are the main reasons I think making flagging systems easier to implement and maintain for text traffic.

France is not alone in this. What are the approaches used by other countries tackling the same issue? Are you able to see a pattern of what tends to be effective and what fails?

TZ: Over my ten years of professional experience in the field, I’ve seen mostly two approaches providing significant results.

The first one is the one currently implemented in the US and revolves around beefing up both the laws and the fines. The legal framework surrounding automated calling (robocall) was tightened over the last years, while the enforcement has been dramatically reinforced and the violation fines severely increased. The combined impact of all three actions was enough to make a significant dent in abusive and misleading calling practices. The FTC beefed up its stance against robocalls9 since the arrival of Ajit Pai who made it the focus of its tenure.

The second approach is more straightforward, more natural to set up and somehow more direct. Automated phone calls and texts are proliferating because they are cheap. Some countries significantly increased their voice and SMS price to make it much less attractive. Rates can be improved even further for inbound international calls to limit contact centers ability to call in the country. This approach proved successful in Europe, particularly for Switzerland, Belgium, and Italy.

Finally, other more radical approaches are worth considering. In the United Arab Emirates, for instance, telecom communications are heavily regulated. Marketing campaigns have to be declared beforehand to the relevant authorities (Telecommunications Regulatory Authority — TRA). As soon as any violation is observed, the operator is directly held responsible for allowing unlawful traffic to happen on its network and faces fines. It forces the operator to be particularly careful with the traffic they handle and it that sense works — they were able to significantly limit abuses. Emirati operators are now providing comprehensive resources to their clients (businesses) to help them comply10.

Global Voice API for smarter interactions


References & further reading

Want to dive deeper? Here are the few other sources available in English covering either the ARCEP decision or the problem of automation and robocalls:

Do you read French? Here are a couple of interesting documents if you want to know more:

Also, here are the sources (footnotes) referenced in the article: