A vague RFP produces vague proposals. Three bids that can't be compared, a vendor selection cycle that drags on for months, and a partner who doesn't actually do what you needed. A precise RFP, one that specifies what evidence you require, how you will score responses, and what conditions will lead to disqualification, produces apples-to-apples bids in two to three weeks and a partner who scales with you. This guide covers the eight sections that every contact centre RFP should contain, with the common mistakes in each and what a strong answer looks like.
Quick answer: a complete contact centre RFP covers eight areas: scope of service, SLA commitments, staffing and training, technology requirements, compliance and security, pricing structure, transition plan, and governance. Each needs specific, verifiable requirements rather than open-ended questions, or you'll receive proposals that can't be compared. The full breakdown, with example wording for each section, follows below.
This is Armatis's complete RFP guide, part of a broader series on evaluating and selecting an outsourcing partner. If you haven't yet defined your evaluation criteria, start with the 25 criteria for evaluating a provider. Once proposals come in, the guide to comparing proposals shows how to normalise them for a fair comparison. And if you're still deciding whether to outsource at all, the readiness guide is the place to start.
The structure below applies whether you're outsourcing for the first time or replacing an incumbent provider. It's designed for European companies operating under GDPR, though the operational logic applies globally.
An RFP (Request for Proposal) exists to do three things at once: force every provider to answer the same questions so their proposals can be compared side by side, create a written record that protects you if the engagement is later challenged internally or by procurement, and give the provider enough structured information to price accurately instead of guessing. Skip the RFP and you get proposals that are impossible to compare, because each provider will have interpreted your needs differently.
Not every situation calls for a full RFP. Three formats exist, and picking the wrong one wastes time on both sides.
The table below maps your situation to the right format, so you don't build a 25-page RFP when an RFQ would do, or a two-page RFQ when the complexity of your requirements actually needs a full RFP.
Copy this structure directly as the skeleton of your document. It maps to the eight sections detailed below, in the order buyers and providers generally expect to see them.
Sections 1, 10, and 11 sit outside the eight core areas below. Section 1 is a short paragraph: who you are, why you're evaluating outsourcing, and what success looks like in one sentence. Section 10 is where the ten questions to ask a BPO provider become useful in a different form: turn them into a scoring grid (1 to 5 per question) so every proposal is evaluated against the same criteria instead of on gut feeling. Section 11 is purely logistical: submission deadline, format, and the name of your single point of contact for questions.
This section defines exactly what you're outsourcing. It sounds obvious, but ambiguity here is the root cause of most post-signature disputes. Specify the channels (voice, email, chat, social, messaging), the interaction types (customer service, technical support, sales, complaints, back-office), the hours and days of coverage, the volume forecasts by month with seasonal peaks flagged, and the languages required.
Be explicit about what is excluded. If you're keeping Tier 3 escalations in-house, say so. If social media moderation is out of scope, say so. The clearer the boundary, the more accurate the pricing and the fewer surprises at go-live.
Common mistake: describing the scope in terms of headcount ("we need 20 agents") rather than outcomes ("we need to handle 15,000 contacts per month at 85%+ CSAT across three channels"). Headcount is the provider's problem to solve. Outcomes are yours to define.
What a strong response looks like: the provider maps your scope to a delivery model (dedicated, pooled, or hybrid), proposes a staffing plan with volume assumptions documented, and flags any risks in the scope as defined, rather than accepting everything without question.
Example wording to include in this section: "The provider will handle customer service via voice, email, and chat, covering order tracking, returns, and account changes, in French and English, from 8am to 8pm Monday to Saturday. Estimated monthly volume: 12,000 contacts, with a seasonal peak of 30,000 contacts in November and December. Tier 3 technical escalations remain managed in-house."
Service Level Agreements translate expectations into measurable, contractual obligations. They need to cover four dimensions: speed (service level, response time by channel), quality (CSAT floor, QA score, NPS), efficiency (AHT targets, FCR), and availability (abandonment rate cap, schedule adherence).
Each SLA should specify how it is measured (what tool, what sample, what period), what happens when it is missed (financial penalties, corrective action plan, escalation), and what happens when it is consistently exceeded (gain-sharing, scope extension).
Common mistake: setting SLAs without understanding their cost implications. An 80/20 service level (80% of calls answered within 20 seconds) requires significantly more staffing than a 70/30 target. If you demand the former but budget for the latter, every proposal you receive will contain a hidden quality compromise somewhere.
What a strong response looks like: the provider proposes SLAs that are achievable given the volume and budget, explains the trade-offs between tighter SLAs and cost, and offers a ramp-up period with progressive targets rather than contractualising full performance from day one.
Example wording to include in this section: "The provider must guarantee an 80% service level within 20 seconds on voice, a CSAT of 85% or above measured monthly via post-interaction survey, an average handling time under 6 minutes, and an abandonment rate below 5%. Targets apply from month 3 onward, with progressive targets during the ramp-up period."
This section defines how the team is built, trained, and retained. Specify whether you require a dedicated team (exclusive to your account) or accept a pooled model (shared advisors across clients). Define the minimum qualifications: language proficiency levels, sector experience, technical skills.
Ask the provider to detail their recruitment process, their training programme (duration, content, certification criteria), their QA methodology, and their advisor turnover rate on comparable accounts. According to SQM Group, replacing a single contact centre advisor costs over 19,000 euros when all factors are included. A provider who doesn't track this number, or won't share it, is either not measuring or not comfortable with the answer.
Common mistake: specifying headcount without specifying the knowledge transfer and training methodology. Twenty agents hired in week one and briefed for two days will produce worse results than twelve agents hired in week one, trained for two weeks, and certified before going live.
What a strong response looks like: the provider presents a structured onboarding timeline with milestones, describes how they build and maintain your knowledge base, and shares their approach to advisor career development and retention.
Example wording to include in this section: "Advisors must hold native or C1-level proficiency in French and B2-level in English, minimum 6 months of customer service experience, and complete a certification test before handling live interactions. The provider will disclose its annual advisor turnover rate on comparable accounts as part of its proposal."
Specify the platforms and tools the provider must integrate with: your CRM (Salesforce, HubSpot, Dynamics), your ticketing system (Zendesk, Intercom, Freshdesk), your telephony infrastructure, and any proprietary systems. Clarify whether the provider is expected to operate within your stack, bring their own, or propose a hybrid.
In 2026, the AI dimension belongs in this section. Define your expectations: do you want the provider to deploy their own AI tools, integrate with yours, or both? What is the expected role of AI (triage, draft responses, full automation of certain interaction types)? How should AI performance be tracked alongside human performance?
Common mistake: treating technology as an afterthought. A provider who promises to integrate with your CRM but discovers at implementation that the integration requires custom development will add weeks to your timeline and cost to your project.
What a strong response looks like: the provider maps the integration architecture in their proposal, identifies potential friction points, and includes the integration timeline in the transition plan. For an overview of what omnichannel technology integration looks like in practice, see the guide to building an effective omnichannel strategy.
Example wording to include in this section: "The provider must operate within our existing Zendesk instance without requiring a parallel system. Any AI tooling proposed must integrate with Zendesk's ticketing workflow and report performance through a shared dashboard accessible to our team in real time."
A mid-sized European retailer once sent an RFP that specified scope, SLAs, and pricing in detail, but left compliance as a single line: "provider must be GDPR compliant." Three providers passed the shortlist on service quality. Only during contract negotiation did the retailer discover that one provider's proposed delivery model relied on a sub-processor outside the EU for after-hours coverage, an arrangement that would have required a separate legal review the retailer hadn't budgeted time for. The deal collapsed six weeks before the planned go-live, and the retailer restarted the search from scratch.
The lesson isn't that compliance is the only section that matters. It's that any section left vague becomes the section where the real risk hides, because providers will fill the gap with whatever is easiest for them, not whatever is safest for you.
For any outsourcing engagement involving European customer data, GDPR compliance is a mandatory requirement, not a differentiator. Your RFP should require specific evidence: ISO 27001 certification (covering the sites that will handle your data), a completed Data Processing Agreement (DPA) template, the list of sub-processors with their locations, and the provider's data breach notification procedure.
For regulated sectors, add sector-specific requirements: PCI DSS for payment data, EBA outsourcing guidelines for banking, NIS2 readiness for critical infrastructure operators. Ask the provider to describe how they handle regulatory audits, including your right to audit their operations, with what notice period and at whose cost.
Common mistake: accepting a checkbox ("yes, we are GDPR compliant") instead of evidence. A provider who cannot produce their sub-processor list on request, or whose ISO 27001 certificate doesn't cover the specific site handling your data, is not ready for a regulated engagement.
What a strong response looks like: the provider includes the DPA as an annex to their proposal, names every sub-processor, specifies data hosting locations, and describes their audit readiness process. For a deep dive into compliance requirements in regulated outsourcing, see the KYC outsourcing and GDPR guide.
Example wording to include in this section: "The provider must hold current ISO 27001 certification covering the specific site(s) handling our data, provide a completed DPA prior to contract signature, disclose all sub-processors with their physical locations, and grant an annual audit right with 30 days' notice at no additional cost."
Understanding how you'll be billed matters as much as understanding the amount. BPO providers generally use three models, often combined: per-hour per-agent (the most common for dedicated models), per-interaction or per-ticket (common for pooled models), and outcome-based (tied to CSAT, conversion, or resolution targets).
Your RFP should ask providers to break down their pricing into its components: agent labour cost, management and QA overhead, technology fees, training and ramp-up costs, and any variable fees for peak periods. According to Armatis's analysis, a pooled BPO provider typically reduces cost per contact by 15% to 35% compared to equivalent in-house management, but this only holds when governance is properly structured and SLAs are actively managed.
Common mistake: optimising only for price. The cheapest proposal usually produces the most expensive engagement, because the compromises show up later as quality issues, missed SLAs, and a costly mid-contract renegotiation. Ask providers to show the total cost of ownership over three years, not just the monthly rate.
What a strong response looks like: the provider presents a pricing model that aligns their incentives with yours (gain-sharing on quality improvements, penalties on SLA misses), transparently separates fixed from variable costs, and includes the cost of the transition phase. For a full treatment of what outsourcing really costs, see the guide to the real cost of outsourcing.
Example wording to include in this section: "Providers must present pricing broken down by: cost per FTE hour or per interaction, technology and platform fees, one-time transition costs, and any variable surcharge for peak-period staffing. Include a 3-year total cost of ownership projection based on the volume forecast in Section 2."
This section asks providers to describe how they will take over operations from your current setup (whether in-house or from another provider). A credible transition plan covers: timeline with milestones and go/no-go gates, knowledge transfer methodology, training and certification process, pilot phase perimeter, progressive ramp-up plan, and risk mitigation measures.
A realistic transition for a standard contact centre outsourcing engagement takes 8 to 12 weeks. More complex setups (multi-country, regulated sectors, large-scale technology migration) can take 16 weeks or more. Any provider who promises a materially shorter timeline should be asked to explain exactly what they are skipping.
Common mistake: treating the transition as a formality. The transition is where the foundations of the engagement are built. Rush it, and every subsequent month will be spent fixing what should have been set up correctly from the start.
What a strong response looks like: the provider presents a structured transition methodology with named roles, documented phases, and a pilot plan. They flag potential risks proactively rather than accepting everything without question. For a step-by-step breakdown, see the transition guide.
Example wording to include in this section: "Provide a detailed transition plan from contract signature to go-live, including named milestones, a knowledge transfer schedule, a pilot phase covering at least one full channel for a minimum of 2 weeks, and explicit go/no-go criteria before full-scale ramp-up."
This is the section most RFPs underweight, and the one that determines whether the engagement improves year over year or slowly degrades. Ask providers to describe their governance model: what meetings happen (operational reviews, steering committees, strategic reviews), at what frequency, with whom, and with what decision-making authority.
Ask how QA is structured: who evaluates interactions, how often, against what scorecard, and how findings are translated into training and process improvement. Ask how continuous improvement is driven: who identifies optimisation opportunities, how are they prioritised, and how is their impact measured?
Common mistake: assuming governance will take care of itself. Without a defined governance model, an outsourcing engagement drifts toward the provider's default operating mode, which is optimised for their efficiency, not your outcomes.
What a strong response looks like: the provider includes a governance charter in their proposal, names the roles involved, describes the QA methodology, and proposes a continuous improvement framework with documented success criteria.
Example wording to include in this section: "The provider will hold weekly operational reviews, a monthly steering committee with our management, and a quarterly strategic review. QA will cover a minimum of 5% of interactions per advisor per month, scored against a shared scorecard, with results reported alongside CSAT and AHT in a shared dashboard."
Two things that will improve the quality of every response you receive. First, share your current cost per contact or total annual spend. Not your exact rate cards, but enough context for providers to calibrate their proposals realistically. Second, send the RFP to four to six pre-qualified providers, not twenty. A provider who knows they are competing against five others will invest in their response. One who knows they are one of twenty will not.
For an overview of the European outsourcing landscape and the main provider categories, see the panorama of leading BPO providers in Europe.
This guide covers what goes into the RFP document itself. It's one part of a broader evaluation process, and each stage has its own guide:
Typically 15 to 30 pages. Shorter risks missing critical details that produce ambiguous proposals. Longer signals over-engineering and discourages quality responses from the best providers.
Four to six pre-qualified providers is the optimal range. Fewer limits your options; more dilutes the quality of responses because providers invest less effort when they know the competition is broad.
Share your current cost per contact or total annual spend, but not your exact rate cards. This gives providers enough context to calibrate realistic proposals without anchoring every response to your existing pricing.
Optimising only for price. The cheapest proposal typically produces the most expensive engagement over time, through missed SLAs, quality issues, and a costly mid-contract renegotiation. Always evaluate on total cost of ownership over three years.
Yes. In 2026, AI is a core component of contact centre operations, not an optional add-on. Your RFP should define expectations for AI's role (triage, response drafting, full automation), how AI performance will be measured, and how the provider integrates AI with human agents.
An RFI (Request for Information) is a lighter document used early to shortlist providers when you have five or more candidates. An RFP (Request for Proposal) is the full evaluation document, covering scope, SLAs, staffing, technology, compliance, pricing, transition, and governance, used once you've shortlisted providers and need comparable, binding proposals. An RFQ (Request for Quotation) is pricing-only, appropriate when the scope is already standardised and cost is the only remaining variable.
Yes, in the form of a scoring grid rather than open questions. Add them as a section near the end of the document where each proposal is rated 1 to 5 against a fixed set of criteria, so every provider is evaluated the same way instead of on gut feeling.
Armatis is a European specialist in customer relations and business process outsourcing (BPO), operating across multiple continents with thousands of employees serving companies of all sizes and sectors. The company designs and manages end-to-end customer service operations: multichannel contact centres, complaints handling, technical support, back-office and digitised processes. Backed by integrated technology infrastructure and the ability to adapt to any sectoral and regulatory context, Armatis helps its clients combine operational performance, quality of experience and cost control, wherever they need it.
Contact our teams to discuss your challenges and find out how we can support you
Join the leaders who trust our multilingual and technological expertise.