Key takeaways
- The EY / SP2C Barometer 2026 confirms a rupture: in 2025, the French outsourced customer relations market declined for the first time since 2021 (-1.4 %, to €3.5 bn). EY and the SP2C read this not as a cyclical accident but as a structural inflection: the shift "from volume to value", driven by AI.
- This thesis does not come from an isolated consultancy: it is voiced word for word by the Top 10 executives interviewed. Konecta talks about speech analytics to analyse "every exchange, from the reason for contact to the tone of voice"; Wisecom describes the contact center as a "sensor of market truth".
- Yet the study runs into a stubborn number: in 2025, 90.4 % of value is still generated by human advisors, 8.7 % by AI agents, 0.9 % by hybrid agents. The shift is announced; it has not yet happened.
- The missing link between the slogan and the proof is the measurement of every interaction. Without it, "moving from volume to value" remains a talking point: you cannot steer a value you only observe on 2 to 5 % of exchanges.
- AI-powered conversational analysis is precisely that measurement layer: it makes observable, across 100 % of interactions, what the study describes as the new terrain of performance (resolution, compliance, empathy, weak signals), in service of the augmented advisor, never of surveillance.
"From volume to value": the thesis of the 2026 Barometer
Each edition of the SP2C Barometer, produced with EY, maps the economic, social and territorial impacts of outsourced contact centers in France. The 2026 edition, subtitled "AI and Outsourced Customer Relations: Towards Augmented Performance", marks a shift in tone.
The central message is clear. After near-continuous growth since 2021, the French market fell -1.4 % in 2025, to €3.5 billion (qualified market at €3.23 bn, -1.1 %). In his foreword, Laurent Vagneur (Partner, EY Consulting) refuses to read this as a mere air pocket: "The major transformation lies precisely in a shift from volume to value, and, consequently, a change in the market's economic model."
In other words: outsourced customer relations is ceasing to be paid by the minute handled, to be remunerated instead on the impact produced, the quality of resolution, the contribution to growth. This is a displacement of the sector's economic center of gravity, and AI is the stated engine.
This reading is consistent with what Raisetalk has observed in the field for two years. Yet it raises a simple question, which the study poses without fully answering: if value shifts towards what happens inside the interaction, how is that value measured?
The thesis is carried by the players themselves, not only by EY
What gives this barometer its weight is that the thesis did not come down from a consulting firm: it rises from the field, voiced by the leaders of the market's largest outsourcers. And when you reread their testimonials, they all describe, in their own words, the same mechanism: leveraging what is said in conversations to create value.
| Executive | Company | What they describe |
|---|---|---|
| Paola Fabioni | Wisecom | "The contact center becomes a sensor of market truth. Every interaction is strategic data: expectations, irritants, weak signals." |
| Pierre-Étienne Genain | Konecta | "Thanks to Speech Analytics, we analyse every exchange, from the reason for contact to the tone of voice, in order to support advisors and provide concrete business insights to our clients." |
| David Marino | Oscaro.com | "Speech analytics will play a central role in analysing customers' real questions, detecting barriers to conversion and identifying products that need more explanation or pedagogy." |
| Karim Madich | Iteractii | "We no longer steer flows, but client assets, as a business partner committed alongside brands on their ROI." |
The convergence is striking. A "sensor of truth", the analysis of "every exchange", the detection of "barriers to conversion", the steering of "client assets" rather than flows: these are not four different visions, but one and the same idea. The raw material of tomorrow's value is the conversation, once structured and analysed.
For a conversational analysis provider, reading this thesis under the pen of Top 10 executives, stamped by EY and the SP2C, is worth more than any sales pitch. The category no longer needs evangelising: it is named, claimed, and set out as the sector's next construction site.
The number that reminds us the shift has not yet happened
There is one place in the barometer, however, where the discourse collides with reality. It is the breakdown of value by AI usage, measured in 2025 among SP2C members.
| Who generates value in 2025 | Share of revenue |
|---|---|
| Human advisors (no AI usage) | 90.4 % |
| AI agents (chatbots, voicebots, automation) | 8.7 % |
| Hybrid agents (AI-assisted advisors, copilots) | 0.9 % |
Nine euros out of ten are still generated by human advisors. Hybrid agents, precisely the form of AI that every executive calls for, the AI that augments the advisor instead of replacing them, account for less than 1 %.
The study is lucid on this point: "AI-based solutions retain a still limited contribution to value creation." The shift from volume to value is therefore, for now, more a trajectory than a state of fact. The sector is in its very first meters.
This figure is not a weakness of AI: it is the measure of the ground still to cover. A market where AI carries only 8.7 % of value, and where hybrid AI stands at 0.9 %, is not a market saturated with technology. It is a market where almost the entire challenge, the industrialisation of hybrid models, still lies ahead. And that challenge is first and foremost a challenge of measurement: you do not industrialise, nor remunerate, a value you cannot observe.
A sensor of truth, yes, but measured how?
Let us return to Wisecom's image: the contact center as a "sensor of market truth", each interaction as strategic data. The image is apt. It is also incomplete as long as the next link is left unspecified: a sensor is only worth its ability to record and render what it captures.
Yet this is exactly where most contact centers fall short. The conversation is the richest point of contact, the most revealing of customer expectations and irritants. It is also the only one that, historically, is barely measured. Dashboards overflow with AHT, answer rates, declarative NPS. But what was actually said in the exchange is analysed on only 2 to 5 % of interactions, listened to manually, with all the bias and slowness that implies.
The barometer says as much in its own way when it describes the sector's activities. In 2025, commercial assistance still accounts for 67 % of revenue and the telephone for 73 % of volumes. Value is overwhelmingly at play by voice, live, across millions of conversations. Claiming to steer that value while observing only a tiny fraction of it is flying blind.
The "sensor of truth" is only a sensor if you plug into it a measurement layer capable of processing 100 % of exchanges. Without that, the phrase remains a nice seminar metaphor.
Speech analytics: the category is validated, execution remains to be done
Two executives, at Konecta and at Oscaro, explicitly name the lever: speech analytics. This is a direct validation of the category, and it deserves to be taken seriously all the way through.
David Marino (Oscaro) describes its target use with remarkable precision: "analysing customers' real questions, detecting barriers to conversion and identifying products that need more explanation or pedagogy." This is not agent surveillance. It is market intelligence, product knowledge, commercial optimisation, extracted from the conversation.
He adds a projection that every customer relations manager should ponder: over time, "a customer service of excellence will have a direct impact on cancellation rate, repurchase rate, but also on brand visibility and preference in AI-driven environments." The conversation ceases to be a cost and becomes a measurable revenue lever.
Naming a lever, however, is not enough to activate it. Between "we do speech analytics" and "we analyse, structure and leverage 100 % of our conversations to steer resolution, compliance and conversion", there is a gulf of industrialisation. It is this gulf that most players still have to cross, and it largely explains why AI accounts for only 8.7 % of value.
Speech analytics does not mean "listening to a few calls with an AI". The real rupture is exhaustiveness and structuring: turning every exchange, across all channels, into comparable data (reason for contact, resolution, adherence to mandatory steps, tone, objections, opportunities). It is this structured material, and it alone, that makes it possible to move from descriptive reporting to steering value per interaction. To go deeper, see our article on the untapped information in your calls.
The missing measurement between promise and reality
The barometer describes a sector repositioning itself "less oriented towards volume and production than towards quality of resolution, customer satisfaction and direct contribution to growth". Translate each of these words into an operational question, and the role of measurement becomes obvious.
| What the new model promises (EY / SP2C) | The steering question it raises | What conversational analysis makes measurable |
|---|---|---|
| Quality of resolution | Was the customer's problem really solved, or will they call back? | Resolution score per exchange, detection of repeat and callback patterns |
| Contribution to growth | Are commercial opportunities detected and converted? | Spotting intent signals, objections, effective closing techniques |
| Satisfaction and loyalty | Was the customer understood, listened to, reassured? | Tone, rephrasings, talk-time ratio, active listening, empathy |
| Compliance and trust | Are mandatory disclosures and the duty of advice respected? | Rate of adherence to regulatory steps across 100 % of exchanges |
Every line of this table is a component of the "value" the study calls on us to prioritise. And each one plays out in the conversation, so it remains invisible unless conversations are analysed. The value announced by the barometer and the material observed by conversational analysis are exactly the same thing.
This is also true on the compliance side, a point the SP2C's editorial places high among its 2025 priorities (AI Act, data protection, regulation of cold calling). In regulated segments, "value" includes not exposing the brand to legal risk on every call. Here again, only exhaustive measurement protects. See our dedicated article on sales compliance in regulated sectors.
The augmented advisor needs a measurement layer
The barometer's third section is unambiguous about who should benefit from AI: "Towards a social performance making humans the primary beneficiary of AI." The narrative thread of every testimonial is the same: augmentation, not substitution.
The social figures confirm that the sector is betting on its people despite the contraction (-3.6 % headcount, to 36,478 FTE in France). Permanent contracts rise to 84 %, advisor pay increases by +5 %, and above all training rebounds sharply: initial training at 94h (+7.3 %) and continuous training at 34h (+20 %). The role is recomposing around data and digital functions (Data Analyst, digital project specialist, bot trainer).
But "augmenting" an advisor implies knowing where they need it. That is the blind spot raised by Angélique Gérard (STEM Academy) with a phrase that should alert any operations director: "AI without training is a full-scale stress test... for the teams." And further on: "Tomorrow's advisor will not be replaced by AI. They will be outrun by those who know how to use it."
Deploying AI without measuring what happens inside the interactions is exactly the "adaptation fatigue" she describes: higher expectations, new tools, but no structured skills development. The measurement layer changes the game because it makes coaching factual, personalised and fair: instead of feedback based on three randomly monitored calls, it delivers support grounded in trends observed across the entirety of an advisor's exchanges.
Measurement in service of the advisor, never against them. Analysing 100 % of interactions is not about surveillance: it is about identifying what sets the best advisors apart, making it teachable, and giving everyone feedback grounded in facts rather than impressions. It is the condition for AI to "reveal talent" rather than exhaust it, to borrow STEM Academy's words. See our article on data-driven agent coaching.
This requirement echoes, moreover, a concern voiced by Martin Dufourcq (Armatis) in the study: the risk that AI "levels experiences" and standardises the relationship. The answer is not to measure less, but to measure better: to preserve the singularity of a brand positioning, you must be able to observe what, in the conversations, creates that singularity.
What the EY / SP2C Barometer 2026 means for your contact center
Beyond the thesis, the study delivers a precise snapshot of the market. Here are the most actionable signals, and how a conversational analysis practitioner reads them.
| Finding of the 2026 barometer | What it implies | The role of conversation measurement |
|---|---|---|
| Market in decline (-1.4 %), budgets under pressure | Every euro invested must prove its ROI | Tie every analysis to a business KPI (resolution, conversion, compliance, churn), not to an isolated quality dashboard |
| AI at 8.7 % of value, hybrid at 0.9 % | The challenge is not to adopt AI, but to industrialise it | Move from 2-5 % of calls monitored to 100 % analysed, the precondition of any industrialisation |
| Telephone at 73 %, commercial assistance at 67 % | Value plays out by voice, live | Analyse voice at scale, where revenue is actually concentrated |
| Contrasting sectors: Banking (+4 %), Insurance (+6 %) rise; Telecom and Energy erode | Reorient effort towards buoyant and regulated segments | On these segments, compliance and the duty of advice are an integral part of "value" |
| Training sharply up (+20 % continuous) | Coaching becomes a central investment | Feed training with insights drawn from real interactions, not theoretical examples |
| Sovereignty cited as a purchasing criterion (Iteractii) | Regulated players demand guarantees | Favour analysis hosted in Europe, with no reuse of data for training |
Where to start
The study values the small-steps strategy, targeted initiatives that prove their value before scaling. For a contact center, this translates concretely into:
- Choose a pilot scope: a campaign, a team, a call reason with high stakes (resolution, conversion or compliance).
- Analyse one month of conversations on that scope to establish a baseline: scores by dimension, gap between the best advisors and the rest, recurring callback reasons.
- Isolate the 2 or 3 levers that weigh most on the target business KPI.
- Launch targeted coaching, backed by real examples drawn from the analysed exchanges.
- Measure progress month after month, advisor by advisor, and quantify the gain.
That is how "from volume to value" stops being a slogan and becomes a measurable trajectory. To go further on method, see our complete guide to quality monitoring in call centers and our overview of AI-powered QM.
Make the shift measurable
- Try it for free: app.raisetalk.com/try
- Contact us: www.raisetalk.com/contact
The EY / SP2C Barometer 2026 makes the right diagnosis: the outsourced customer relations market is shifting from volume to value, and AI is the engine. It nonetheless leaves one question open, that of measurement. You do not steer a value you do not observe; you do not industrialise hybrid models without data on what truly happens inside the interactions; you do not augment an advisor without knowing where to augment them. AI-powered conversational analysis is that measurement layer: it turns the "sensor of truth" described by the sector's leaders into a real steering instrument, across 100 % of exchanges, continuously, in service of humans and never of their surveillance. The slogan is right. It was missing the proof. The proof is measurement.
Source: SP2C and EY (Ernst & Young Advisory), 2026 Barometer of the economic, social and territorial impacts of outsourced contact centers in France: "AI and Outsourced Customer Relations: Towards Augmented Performance", 2026 Edition. Figures and quotes drawn from the study. See also our analysis of the KPMG CEE Barometer 2025-2026.