Telephone Call Centers (U.S.) — NAICS 56142

A Histometrics rollup primer for a general investing audience — relevant to both public-market and private investors. This level synthesizes two child industries; figures are reported facts unless the wording marks them as projections or judgments.

1. Overview

The North American Industry Classification System (NAICS) is the U.S. government's standard scheme for grouping businesses. Code 56142, Telephone Call Centers, is the umbrella over two very different child industries: firms that answer phones and take messages for small businesses (561421, Telephone Answering Services), and firms that run full outsourced contact centers — customer service, technical support, and telemarketing — for large brands (561422, Telemarketing Bureaus and Other Contact Centers).

The two things this level has in common: both are labor-intensive services sold to other businesses, priced off the human agent-hour, and both sit squarely in the path of the same technology shock — generative artificial intelligence (AI), software that can hold a human-like phone conversation and so attacks the industry's core product. Almost everything else about them differs, and that contrast is the point of this primer. The level spans a two-person answering service that books appointments for a plumber all the way up to a 400,000-seat global outsourcer running an airline's support line.

At the U.S. level, telephone call centers are a roughly $28 billion industry (2022 receipts), employing about 386,000 people across some 4,100 establishments [1]. One child — contact centers (561422) — is about ten times the size of the other and carries essentially all of the public-market exposure. The answering-services child (561421) is a small, almost entirely private cottage industry. An investor's route into "telephone call centers" therefore depends heavily on which child you mean.

2. What's inside — the two children and how they differ

The level splits cleanly into a big industry and a small one that do related but distinct work. The single most useful thing to understand here is the contrast:

Reading the table. Notice the mismatch between establishments and revenue. Answering services are 25% of the storefronts but only 9% of the revenue — a swarm of tiny shops. Contact centers are 75% of the storefronts but 91% of the revenue — the same physical count contains far bigger operations. Average receipts per firm are roughly $2.8 million in answering versus $10.7 million in contact centers [3][4]. Both children are, by federal concentration math, highly fragmented (see §3), so neither is a "few giants own it all" market at home — but the type of owner is completely different: mom-and-pop and search-fund on one side, Nasdaq-listed multinationals and mega-PE on the other.

Where the code boundaries sit. Answering (561421) is simple message-taking and reception; the moment a provider runs large inbound/outbound campaigns or full customer service, it belongs in contact centers (561422) [2]. Both children exclude a company's own in-house ("captive") service center — that is classified in the parent company's industry, not here — and both exclude the software vendors (dialers, routing, cloud contact-center platforms) that arm these centers, which sit in software/data-processing codes [2]. Those exclusions matter for sizing (see §3).

3. Size (this level's rollup figures)

U.S. federal statistics for the combined level, NAICS 56142. Receipts, firm counts, and concentration come from the 2022 Economic Census; establishment, employment, and payroll counts from County Business Patterns (CBP) 2023 — two programs from different years, so do not combine them to compute a margin [1].

The rollup checks out. The two children add up to this level almost exactly: establishments (1,035 + 3,106 = 4,141) and employment (41,582 + 344,605 = 386,187) match the level file to the unit, and receipts (~$2.6B + ~$25.6B) and payroll (~$1.6B + ~$13.4B) reconcile to the level totals [1][3][4]. Contact centers (561422) supply roughly 91% of the money and 89% of the jobs; answering (561421) supplies the rest. In other words, when someone talks about "the U.S. call-center industry" as an investment, they are overwhelmingly talking about 561422.

What the numbers reveal:

• Labor-heavy, low-value-added. Average pay is about $38,900 [1] — consistent with roughly $18–19/hour agents — and revenue per employee is only about $73,000. Comparing 2023 payroll to 2022 receipts is a rough cross-program indicator, not a margin, but at face value payroll is on the order of half of receipts: the signature of a business whose main input is wages.
• Almost everyone is a "small business." Average receipts per firm (~$8.5M) sit below the Small Business Administration (SBA) size standards for both children ($19.0M for answering, $25.5M for contact centers) [5]. Those thresholds are federal-contracting eligibility rules, not valuations, but they confirm most of the 3,300 firms qualify as small.
• Fragmentation, oddly, increases at the combined level. The level HHI (144.7) is lower than either child (172.4 and 165.2). That is arithmetic, not a market fact: pooling two industries makes every firm's share of the bigger pie smaller, so the squared-share index falls. The plain reading holds in every slice — CR4 of 20.4% means the four largest firms hold about a fifth of U.S. revenue, and U.S. antitrust guidance treats anything under 1,500 HHI as "unconcentrated." At home this is a long tail under a few large operators [1].

Undercount and scope caveats — the federal figure understates true activity in three ways. First, these are employer counts: CBP omits the self-employed, firms without an Employer Identification Number, and shops with no paid staff. That undercount bites hardest in the answering-services child, where solo "virtual receptionists" and micro-operators are common; the Census Nonemployer Statistics program covers them, but our data set has no 56142-specific nonemployer figure [1]. Second, captive in-house centers are excluded by definition, so brands' own service operations don't appear here at all. Third, this is a domestic slice — the U.S.-based contact-center giants run most of their agents offshore (Philippines, India, Latin America), so their global scale dwarfs the $25.6 billion U.S. line [4]. Private market-research firms publish much larger "outsourcing" and "virtual receptionist" market numbers, but they use broader, blended definitions and disagree with one another; anchor on the federal figure and treat vendor sizing as directional [4].

4. Investable universe (where value concentrates across the children)

The two children could not be more different for an investor, and it maps directly to their size.

Answering (561421) — private only. There is no publicly traded pure-play telephone answering company. The direct operators are all private: independent regional and family shops, a few PE-backed platforms rolling up the sector (Nexa, Moneypenny), a premium U.S. brand (Ruby), an employee-owned firm (MAP Communications, via an employee stock-ownership plan, or ESOP), franchise systems (AnswerConnect, Nexa), and venture-backed AI-native entrants (Smith.ai) [15][16][17][18]. Public exposure to this child exists only indirectly, through the much larger contact-center and software names below.

Contact centers (561422) — where the public exposure lives. Even here, the code does not map cleanly onto public-company reporting: listed operators are diversified and report "customer experience," "digital operations," or "software" rather than a clean 561422 line, and the single largest is not U.S.-listed. Figures are each company's most recent reported full year; treat them as thematic proxies, not pure plays.

TELUS Digital (TIXT), a CX/business-process-outsourcing (BPO) operator majority-owned by Canada's TELUS, has moved to go private [10]. Large private/PE-owned operators include Foundever (Sitel + SYKES merger, ~$4.3B revenue), Alorica, Sutherland, VXI Global Solutions (Bain Capital), Qualfon, iQor, and [24]7.ai [11][12][13].

The adjacent software layer — a different code, the same demand wave. Cloud "contact-center-as-a-service" (CCaaS) and AI-orchestration vendors sell the software that runs and increasingly automates these centers: Five9, NICE, RingCentral, Twilio, and privately held Genesys (majority-owned by Hellman & Friedman and Permira, with a $1.5B Salesforce/ServiceNow investment) and Talkdesk [14][29]. This is where you get exposure to AI as the disruptor rather than the disrupted — and it is also the enabling layer beneath the newest AI answering-service startups. For the answering child specifically, voice-AI infrastructure (Twilio, plus startups Bland, Vapi, Retell) is the relevant proxy [29].

Bottom line: value in the public market concentrates almost entirely in the contact-center child and the software layer around it. The answering child is reachable only through private markets.

5. How the money works

Strip away the scale difference and both children run the same economic engine: sell agent capacity, and profit is the spread between what you bill for an hour and what a fully-loaded agent costs, multiplied by how many billable hours you keep that agent producing.

Billing. Answering services (561421) charge per minute of talk time or per call, packaged into monthly subscription plans that bundle a block of minutes — retail full-service answering runs roughly $0.75–$1.75 per minute, pooled "shared-agent" plans less [561421-primer][19]. Contact centers (561422) price per productive agent-hour — roughly $28–$45 onshore (U.S.), $14–$22 nearshore, $8–$14 offshore — plus per-minute, per-seat/month, and increasingly outcome-based (per resolved case or per sale) deals, often with service-level bonuses or penalties [20].

The cost base is labor. Wages, benefits, recruiting, training, supervision, and quality assurance run roughly 70–95% of a center's cost; telecom, cloud software, security, and insurance make up the rest [20]. The U.S. median customer-service wage was about $20.59/hour in May 2024, versus roughly $2–3/hour in the Philippines — which is why offshoring is the classic margin lever in the contact-center child [21]. The answering child, sold on a domestic "brand voice," has less room to arbitrage geography and instead squeezes margin through occupancy.

The margin engine — pooling and occupancy. Both children make money the same way: one receptionist idling between a small firm's occasional calls is expensive dead time, so operators pool many clients onto a shared roster and keep agents busy. The metrics that decide whether an owner makes money are operational and shared across both children: utilization/occupancy (billable hours ÷ paid hours), attrition and ramp time (turnover is chronically high, and every departure means re-hiring and re-training at the owner's expense), revenue per productive hour and program mix (complex, regulated work bills more than basic answering), and quality gates — average handle time (AHT), first-contact resolution (FCR), customer satisfaction (CSAT), and abandonment — that trigger bonuses or penalties. A provider in either child can grow revenue while destroying profit by staffing ahead of demand, tolerating turnover, or accepting low-priced contracts. Because labor dominates and much of the work is commoditized, operating margins are thin — typically mid-single-digit to low-double-digit earnings before interest, taxes, depreciation, and amortization (EBITDA). Standard factory "capacity utilization" and retail "same-store sales" lenses do not fit; revenue-per-productive-hour, occupancy, attrition, and contribution-margin-per-seat do.

Revenue quality. Recurring subscriptions and thousands of small, diversified clients give the better answering operators annuity-like revenue and low churn (the service is wired into the client's phone number) — which is why PE likes that child. Contact centers, by contrast, live on larger contracts with more concentration risk: TTEC's Engage segment reported 82% revenue retention in 2024 (down from 95% in 2023), with its five largest customers at 32% of revenue [6]. Those are company-specific figures, but they show why retention and customer concentration are the numbers to watch across the level.

6. Demand drivers

Both children ride overlapping but not identical demand.

• Total interaction volume. More customers, transactions, and digital services mean more calls, tickets, and chats. E-commerce and subscription growth are tailwinds — mostly for contact centers [561422-primer].
• Small-business formation and the "missed-call" problem. The answering child grows with the call-dependent trades (legal, medical/dental, home services). Industry estimates value a single missed call to a service business at $100–$200, and answering services exist to convert those lost calls into booked jobs [23].
• Corporate cost pressure. Outsourcing turns a brand's fixed service cost into a cheaper variable one; recessions and margin squeezes push more volume out the door to contact centers [561422-primer].
• 24/7 coverage and specialization. Both children sell around-the-clock and overflow coverage a client can't staff itself, and higher-value verticalized handling (legal intake, medical, technical support) commands better pricing and stickier relationships [561421-primer][561422-primer].
• Geographic arbitrage and reshoring. Wage gaps drive offshoring in the contact-center child; data-privacy, quality, and political concerns periodically pull some volume back onshore or nearshore [27].
• AI — the dominant swing factor for the whole level. AI both destroys demand (deflecting simple calls to bots) and creates it (cheap AI answering pulls in micro-businesses that never bought human service, and enterprises need help deploying, running, and supervising AI). Which effect wins, per client, is the central revenue question [22][23].

Demand is economically sensitive but less cyclical than sectors tied to physical production: volume tracks clients' own activity. Healthcare, emergency, government, and legal work tend to be more defensive than discretionary consumer support.

7. Regulation

The NAICS code is not a license; obligations depend on what the provider does with the calls and data. Across the level, the heaviest rules land on outbound calling and sensitive-data handling; a basic inbound answering or service operation carries less direct risk, though its clients (healthcare, finance) are heavily regulated.

• Telephone Consumer Protection Act (TCPA, 1991). Enforced by the Federal Communications Commission (FCC), it restricts autodialed and prerecorded "robocalls," robotexts, and caller-ID spoofing, and governs consent. It bites whenever any operator dials out on a client's behalf — appointment reminders, callbacks, sales — with willful violations up to $1,500 per call. Regulators increasingly treat AI-generated voices as "artificial" calls covered by these rules, which directly affects AI answering [24].
• FTC Telemarketing Sales Rule (TSR) and National Do Not Call Registry. The Federal Trade Commission's (FTC) rule governs outbound disclosures, calling times, abandoned/prerecorded calls, and Do Not Call scrubbing (generally every 31 days), with civil penalties cited as high as $53,088 per violation. The FY2025 registry held 258 million-plus active numbers [25]. A basic inbound answering service is not telemarketing, but an operator creates exposure the moment it dials out.
• STIR/SHAKEN caller-ID authentication (mandated under the TRACED Act to fight spoofing) applies to the calling infrastructure [26].
• Sector-specific data rules. The Health Insurance Portability and Accountability Act (HIPAA) makes an operator handling protected health information a "business associate" that must sign an agreement and use compliant secure messaging — a real cost but also a moat for the medical vertical. The Payment Card Industry Data Security Standard (PCI DSS) applies whenever card data is taken; call recordings and agent notes add exposure [28].
• Labor and the political wildcard. The Fair Labor Standards Act (FLSA) governs wage, overtime, and worker-classification questions [561422-primer]. Separately, the Keep Call Centers in America Act, reintroduced in the Senate in mid-2025, would pressure firms to disclose and limit offshoring of U.S. customer-service work — a political risk that cuts against the contact-center child's core cost model [27].

The rules are in flux. In January 2025 the Eleventh Circuit (Insurance Marketing Coalition v. FCC) vacated the FCC's "one-to-one consent" rule, and the FCC has proposed rolling back several consent requirements [26]. Diligence should cover consent records, Do Not Call procedures, recordings, complaint history, state "mini-TCPA" and biometric-privacy laws, data-security controls, and — increasingly — rules requiring disclosure that a caller is an AI agent.

8. Consolidation

Both children are fragmented at home and both are consolidating, but along different logic.

Contact centers (561422) — scale game at the top. The domestic market is fragmented (HHI 165) [4], yet the top end has consolidated into a global oligopoly. The landmark deal was Concentrix's ~$4.8 billion acquisition of France's Webhelp (2023), creating a ~$9–10 billion revenue leader, which followed Sitel's 2021 merger with SYKES to form Foundever [11]. The logic: labor-arbitrage BPO is a scale game — the biggest buyers of real estate, telecom, and AI tooling win on cost, and only the largest can fund the AI reinvention clients now demand.

Answering (561421) — PE roll-up of a cottage industry. Recurring subscription revenue plus extreme fragmentation is a textbook consolidation setup. Nexa and Moneypenny have grown by acquiring competitors; MAP and AnswerNet openly solicit answering assets; Ruby is PE-backed; and franchise systems add a growth channel [15][16][17][18]. A roll-up creates value by standardizing technology, training, compliance, and sales — and can destroy it through customer churn, incompatible systems, and lost founder relationships.

The AI wildcard cuts across both. Voice-AI infrastructure has slashed the cost of building an "AI receptionist," unleashing low-cost entrants and forcing incumbents to bolt on AI; the emerging consensus in both children is a hybrid model — AI for speed, scale, and simple calls; humans for nuance, complex intake, and escalation [10][23]. Consolidation can as easily destroy value as create it, through client loss, integration failure, excess debt, or acquired labor and litigation liabilities.

9. Risks

• AI substitution and price deflation — the defining, level-wide risk. In the answering child, commodity message-taking is drifting toward AI's roughly $0.10-per-call economics, threatening human-only shops [23]. In the contact-center child, live deployments already replace agents at scale (fintech Klarna publicized replacing the work of ~700 agents with a chatbot), and public markets have repriced the sector severely — parts of the press have called it "uninvestible," short interest has run into double digits, and Concentrix has booked large goodwill impairments [22].
• Thin margins and customer concentration. Low profitability leaves little cushion; losing or repricing one large program can strand facilities and trained agents (see the TTEC disclosure, §5) [6].
• Labor. Wage inflation, chronic turnover, and the cost of 24/7 staffing squeeze the human-heavy model; the U.S. Bureau of Labor Statistics (BLS) projects domestic customer-service-rep employment to fall ~5% and telemarketers ~22% between 2024 and 2034 [21].
• Regulatory and litigation risk. TCPA class actions, Do Not Call and consent failures, HIPAA/PCI breaches, and new AI-voice disclosure rules [24][25][28].
• Cybersecurity and outages. Operators hold sensitive personal, medical, financial, and business data and depend on telecom, cloud, and client systems.
• Currency, geopolitics, and reshoring politics. Offshore delivery exposes the contact-center child to foreign-exchange swings, delivery-country risk, and legislation such as the Keep Call Centers in America Act [27].
• Cyclicality and end-market concentration. Volume tracks client budgets — SMB verticals (home services, real estate) in answering; financial services, telecom, retail, and travel in contact centers.
• Reputational fragility. The service is the client's first impression; a bad interaction directly damages the customer's brand.
• Data-label ambiguity. Public statistics and third-party forecasts routinely blur 561421, 561422, broader BPO, and CCaaS software — read labels carefully.

10. How to invest and the outlook

Public routes — contact centers only. There is no U.S.-only listed pure-play and no listed answering-services company at all. The available proxies are the contact-center operators — Concentrix (CNXC), TTEC, ibex (IBEX), TaskUs (TASK), Teleperformance (TEP/TLPFY), TELUS Digital (TIXT) — which trade at depressed valuations after the AI sell-off; these are contrarian, higher-beta bets that AI augments rather than eliminates the outsourced agent [4][5][6][7][8][10][22]. The adjacent software vendors (Five9, NICE, RingCentral, Twilio) offer the opposite tilt: exposure to AI as the disruptor [29]. Separate service firms from software; examine organic-vs-acquired growth, client concentration and renewals, onshore-vs-offshore delivery, revenue per seat, utilization, attrition, leverage, and AI economics. For valuation, enterprise-value-to-EBITDA and free-cash-flow yield are the workable lenses — not regulated-utility rate base or real-estate-trust metrics, which do not apply here.

Private routes — where the answering child, and most of the level, actually trades.

• Buy or operate an answering service. SBA-friendly targets: recurring revenue, low capital intensity, almost universally "small." Prize low churn, vertical specialization, low owner-dependence, compliant medical/payment handling, and measurable agent productivity [561421-primer].
• Back a roll-up. Both children draw PE — a scale play in contact centers, an early-stage fragmentation play in answering [11][15][16].
• PE platforms. Most of the contact-center child is private (Foundever, Alorica, VXI, iQor); favor recurring or regulated work, diversified clients, documented compliance, and a credible automation plan [11][12][13].
• Venture and franchise. Fund an AI-native receptionist or the voice-AI layer beneath it; or enter answering through a franchise (AnswerConnect, Nexa) [18][29].

Outlook (forward-looking judgment; the federal data contain no growth forecast). The near term across the whole level is defined by the AI adoption curve, which cuts both ways — expanding the market by making service affordable for buyers who never bought it, while compressing price on commodity call-handling. The plausible path is bifurcation in both children: undifferentiated, human-only, low-complexity operators face structural pressure as automation absorbs simple interactions, while firms that reposition around complex, regulated, high-value, and emotionally sensitive work with humans in the loop — legal and medical intake, technical support, premium brand-voice service — can defend and even grow revenue per relationship. Total interaction volume is still rising; the open question for the level is how much of it a human will ever touch, and how much of that human-plus-AI work the winners capture. For investors, the largest child is in the middle of a violent, unsettled re-rating; the smaller child is a quieter, still-early private roll-up. Both are the same bet, sized differently: that the durable value migrates to specialization, compliance, brand experience, and the AI-and-infrastructure layer — and away from the plain human agent-hour that both industries were built on.

---

Sources

Drawn from the two child primers (561421, 561422) and our ground-truth federal statistics for this level.

1. U.S. Census Bureau. 2022 Economic Census (Concentration of Largest Firms) and County Business Patterns 2023 — NAICS 56142, Telephone Call Centers (receipts, firms, establishments, employment, annual and first-quarter payroll, CR4/CR8/CR20/CR50, HHI). 2022–2025. https://data.census.gov/
2. U.S. Census Bureau. 2022 NAICS Manual — codes 56142, 561421 (Telephone Answering Services), 561422 (Telemarketing Bureaus and Other Contact Centers); definitions and exclusions (captive centers; software vendors NAICS 513210/518210; paging 517). 2022. https://www.census.gov/naics/
3. U.S. Census Bureau. 2022 Economic Census and County Business Patterns 2023 — NAICS 561421, Telephone Answering Services (receipts ~$2.59B; 934 firms; 1,035 establishments; 41,582 employees; HHI 172.4). 2022–2025. https://data.census.gov/
4. U.S. Census Bureau. 2022 Economic Census and County Business Patterns 2023 — NAICS 561422, Telemarketing Bureaus and Other Contact Centers (receipts $25.6B; 2,380 firms; 3,106 establishments; 344,605 employees; annual payroll $13.4B; HHI 165.2). 2022–2025. https://data.census.gov/
5. U.S. Small Business Administration. Table of Small Business Size Standards (NAICS 561421 = $19.0M; 561422 = $25.5M average annual receipts). 2023. https://www.sba.gov/document/support-table-size-standards
6. U.S. Securities and Exchange Commission / TTEC Holdings, Inc. Form 10-K and FY2024 results (revenue $2.208B; net loss ~$321M; Engage segment 82% revenue retention 2024 vs. 95% 2023; five largest customers 32% of revenue). 2025. https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=1013880&type=10-K
7. TaskUs, Inc. 2025 Form 10-K; founder/Blackstone take-private (~$1.0B revenue; digital CX, trust and safety, AI). 2026. https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0001829864&type=10-K
8. IBEX Limited. FY2024 Results (revenue ~$525M). 2024. https://investors.ibex.co/
9. Conduent Incorporated. 2025 Form 10-K. 2026. https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0001677703&type=10-K
10. TELUS / TELUS Digital (TIXT). Investor information; proposal to take TELUS Digital private. 2025. https://www.telusdigital.com/investors
11. Everest Group / Foundever / Concentrix. Concentrix acquires Webhelp (~$4.8B, 2023); Foundever (Sitel + SYKES) formation and ~$4.3B scale. 2023–2025. https://www.everestgrp.com/uncategorized/concentrix-acquires-webhelp.html
12. PitchBook / IBISWorld. Alorica Inc. company profile (private; revenue and headcount estimates). 2026. https://pitchbook.com/profiles/company/42917-68
13. Bain Capital Private Equity / VXI Global Solutions. Bain Capital acquisition of VXI Global Solutions. 2022. https://vxi.com/news/bain-capital-completes-acquisition-of-vxi-global-solutions/
14. Genesys. Genesys Announces $1.5 Billion Investment by Salesforce and ServiceNow (Hellman & Friedman and Permira majority owners). 2025. https://www.genesys.com/company/newsroom/announcements/
15. BusinessWire / Nexa Receptionists. Nexa Receptionists Holdings — acquisition-led growth; brand portfolio. 2021–2024. https://www.nexa.com/
16. Updata Partners. Updata Partners Invests in Ruby Receptionists (majority investment, 2017). https://updata.com/updata-partners-invests-in-ruby-receptionists/
17. Moneypenny / ECI Partners. The Moneypenny Group Acquires Sunshine Communication Services (U.S. expansion; earlier VoiceNation acquisition). 2024. https://www.moneypenny.com/us/resources/news/
18. AnswerConnect; Smith.ai; MAP Communications (100% employee-owned ESOP); AnswerNet. Company/brand overviews; franchise channels; answering-asset acquisitions. 2024–2026. https://www.answerconnect.com/
19. Forbes Advisor. Best Answering Services / Best Virtual Receptionists — pricing and plan comparisons. 2025–2026. https://www.forbes.com/advisor/business/software/best-answering-services/
20. Callforce / Site Selection Group / Retell AI. Call-center outsourcing pricing by region and model; labor as 70–95% of cost. 2026. https://callforce.global/blog/call-center-outsourcing-cost/
21. U.S. Bureau of Labor Statistics. Occupational Outlook Handbook — Customer Service Representatives (median $20.59/hr, $42,830; −5% 2024–34) and Telemarketers (−22% 2024–34). 2025. https://www.bls.gov/ooh/office-and-administrative-support/customer-service-representatives.htm
22. Bloomberg / Nearshore Americas. Call-Center Stocks Fall on Worry AI Makes Them 'Uninvestible' (Klarna, Teleperformance short interest, Concentrix impairment). 2026. https://www.bloomberg.com/news/articles/2026-06-30/call-center-stocks-fall-on-worry-ai-is-makes-them-uninvestible
23. AI answering / virtual-receptionist industry reports (vendor and trade estimates — treat as low-confidence). Missed-call economics ($100–$200 per missed call); AI-vs-human per-call cost (~$0.10 vs ~$1.38–$1.60); hybrid model; verticalization. 2026. https://www.nextiva.com/blog/ai-answering-service.html
24. Federal Communications Commission. Telephone Consumer Protection Act (TCPA); Unwanted Calls and Texts (consent; up to $1,500 per willful violation; AI voices treated as "artificial" calls). 2026. https://www.fcc.gov/consumers/guides/stop-unwanted-robocalls-and-texts
25. Federal Trade Commission. Complying with the Telemarketing Sales Rule (31-day scrub; penalties up to $53,088/violation); National Do Not Call Registry Data Book FY2025 (258M+ registrations). 2025. https://www.ftc.gov/business-guidance/resources/complying-telemarketing-sales-rule
26. National Consumer Law Center / U.S. Court of Appeals for the Eleventh Circuit. TCPA/robocall developments 2024–2025 (STIR/SHAKEN, TRACED Act, FCC consent proposals); Insurance Marketing Coalition v. FCC (one-to-one consent vacated, Jan. 2025). 2025. https://library.nclc.org/
27. Site Selection Group. The Keep Call Centers in America Act; reshoring and nearshoring statistics. 2025. https://info.siteselectiongroup.com/blog/the-keep-call-centers-in-america-act-a-turning-point
28. U.S. Department of Health and Human Services (HIPAA — Business Associates) and PCI Security Standards Council (PCI DSS). Sensitive-data handling for call centers. 2019–2026. https://www.hhs.gov/hipaa/for-professionals/privacy/guidance/business-associates/index.html
29. Five9 / NICE / RingCentral / Twilio / Genesys / Talkdesk, and voice-AI infrastructure (Bland, Vapi, Retell). Contact-center-as-a-service and AI-orchestration vendors — the adjacent software layer. 2025–2026. https://www.twilio.com/