← Back to blog
    FIELD NOTES · RESEARCH

    How big is the contact centre testing market? A bottom-up estimate

    Researched and prepared by AI for Nopaque9 min read
    ~$135MActual vendor revenue, 2026
    $400–900MAddressable market
    ~22%Penetration
    ~60%Held by one vendor

    Why this estimate exists

    Contact centre testing is not a category that Gartner, Forrester or IDC size. It sits below the threshold at which analyst firms publish dedicated market forecasts, so anyone asking "how big is this market" is left with vendor marketing claims or general-purpose market-research reports that measure something else entirely.

    This model is built the only way a market this size can be built honestly: bottom-up, from what the vendors in it actually bill. Where statutory accounts exist we have used them. Where they do not, we say so and show the reasoning.

    What the vendors actually bill

    Estimated 2026 revenue attributable to contact centre testing and assurance. Where a vendor sells other things — network monitoring, systems integration, UC diagnostics — only the testing portion is counted.

    VendorEst. 2026 revenueConfidencePrincipal basis
    Cyaraincl. Spearline, Botium, CentraCX, QBox$72–88MModerateInc. 5000 company-substantiated 48% three-year growth (CY2022–25); three independent estimators cluster $65–75M; revenue-per-employee cross-check at $181k
    Hammer / VistaCXInfovista$16–20MLow–modSeven2 discloses Infovista group at €106m; Empirix was ~30% of combined headcount at acquisition; Hammer ~45–55% of Empirix
    Klearcom$8–9.5MMediumDeloitte Fast 50 methodology sets a verified >€1m floor; company-stated 79% growth for 2025; Synova PE investment below €50m
    Operataobservability-adjacent$5–8MLow–modA$11m Series A at A$100m+ post-money, February 2026; published per-agent-minute pricing
    Voice AI testing cohortCoval, Cekura, Hamming, Bluejay, Roark, Evalion, SuperBryn$4.5–13MLowPublished pricing against disclosed customer counts; Coval's ~60 enterprises at $500–4,500/mo
    PumpCX$2.5–5MLowNo filings possible — trades through a discretionary trust
    Nectarassurance slice only$2–3.7MLowSingle modelled total (~$17m); testing share estimated at 12–22%
    Bespoken$1.5–3MLow–modNo fresh capital since a small 2022 bridge; headcount-derived
    PointelSTAMP product only$1.3–3.2MLowSystems integrator; proprietary IP typically 5–12% of revenue
    Occam$1.5–2.4MMediumFiled UK accounts. Trade debtors £171,088 (FY25), retained profit £467,359, five employees
    CloudCX$0.2–0.6MMediumFiled UK accounts. Micro-entity, net assets £83,834, four employees
    Long tailnumber monitoring, regional players$2–5MLowFragmented; mostly sub-$1m vendors
    Total$117–161MCentral case ~$135M

    Two observations follow immediately. First, one vendor is roughly 60% of the market — this is a category with a single scaled player and a long tail of sub-$10m businesses. Second, the hardest numbers in the table are the smallest ones. Occam and CloudCX are UK-registered and file public accounts; everything above them is estimated, because Irish small-company filings carry no profit and loss account, Australian proprietary companies below the "large" threshold file nothing, and US private companies file nothing at all.

    The $6.5 billion figure is not real

    What the claim is. Cyara's January 2022 funding announcement states that it "has forged an early lead in the contact center testing market, estimated by Forrester to reach US $6.5 billion by 2024." The figure has since circulated widely as though it were independent analyst research.

    What it actually is. The footnote on Cyara's own site reads: "Forrester Market Insight Study on CFTA for Contact Centers: Market Insight Study, a commissioned study conducted by Forrester Consulting on behalf of Cyara, November 24, 2021." It is paid custom advisory work, scoped and funded by the vendor, and it has never been published — there is no document number, no abstract, no public landing page, and no third party has ever independently cited it.

    Why it cannot be true. IDC measures the entire worldwide automated software quality market — all test automation, quality management, performance testing and code analysis, across every industry and every use case — at $6.3 billion in 2025. The claim is that contact centre testing alone would exceed the whole global software testing tools market, one year earlier. It also implies enterprises spend 64 pence on testing for every pound spent on the contact centre platform itself.

    Forrester's genuine published position on the category is the opposite. Its Wave: Continuous Functional Test Automation Suites, Q2 2020 scored Cyara as a Contender and noted that it "provides a CFTA suite narrowly focused on improving CX in contact centers" — a niche within a larger market, with no dollar sizing attached. Cyara itself stopped citing the $6.5B figure after September 2022.

    How big the market could be

    Two independent methods, deliberately chosen so that they fail differently.

    Method 1 — attach rate to platform spend

    Metrigy measures CCaaS at $8.4bn (2025, +16.4%) and on-premises contact centre platforms at $1.7bn, giving roughly $10.1bn of contact centre platform software spend. Economy-wide, testing tools run at about 0.5% of software spend (IDC's $6.3bn against Gartner's $1,271bn). Voice carries an unusually high test burden — real carriers, DTMF, latency, audio quality, in-country routing — so a 2–5% attach is defensible where 0.5% is the norm.

    $10.1bn × 2–5% = $200–505M

    Method 2 — addressable accounts × contract value

    Genesys reports 8,000+ organisations, Five9 3,000+, NICE 25,000 across all segments, Amazon Connect "tens of thousands". Netting overlaps and adding large on-premises estates gives roughly 50,000 organisations on a managed contact centre platform. Perhaps 15–25% have the scale, multi-country footprint or regulatory exposure to justify dedicated testing tooling, at a blended $40–80k contract value.

    7,500–12,500 accounts × $40–80k = $300M–1.0B

    Converging

    The two methods overlap across $400–900 million, which we take as the addressable market, with a central case around $600m. Current spend of ~$135m puts penetration at roughly 22%.

    That is meaningfully more penetrated than the category's own marketing suggests, and the reason is worth stating plainly: a large share of contact centres that could buy testing tooling have concluded that manual regression testing at release gates is adequate for a deterministic IVR that changes twice a year. They are not an untapped pipeline. They are a considered "no".

    What changes the denominator

    Two forces are pulling the market in opposite directions, and they act on different buyers.

    Downward: platform vendors are bundling the basics

    Amazon Connect shipped native testing and simulation in preview on 30 November 2025, extended it to chat in March 2026, and publishes no separate price for it — it appears to be bundled. It tests contact flows: configure a caller number and call reason, assert an outcome, run tests in parallel. That is precisely the deterministic flow-regression work that has historically been sold as a product.

    It is narrower than it first looks — English only, single platform, and it explicitly mocks out Lex and Lambda rather than testing them, so it does not touch AI agent behaviour. But on the highest-volume contact centre platform in the world, the price anchor for basic functional testing is now zero. Cross-platform testing, in-country PSTN validation and carrier-path verification remain defensible; single-platform flow testing does not.

    Upward: AI agents create a genuinely new buyer

    The adoption base is real. Genesys reports more than 70% of Genesys Cloud customers using its AI capabilities and AI conversations up 120% year on year; NICE states AI was included in 100% of enterprise CXone deals in a quarter; Five9's AI revenue reached a $150m+ run rate. Sierra AI reached $150m ARR by February 2026, Parloa exceeded $50m, and Vapi alone processes between one and five million calls a day.

    Testing those agents is not the same product. Deterministic testing asserts that a call reached the right queue. Agent testing has to assess hallucination, bias, compliance, tone and goal completion across non-deterministic outputs — which is why Cyara had to build new Compliance and Bias modules rather than extend the existing ones, and why an entirely separate vendor cohort has emerged. Crucially, some of that cohort's customers are voice platform companies rather than contact centres, which means new budget from buyers the incumbents have no relationship with.

    Where this lands by 2030

    Model the two pools separately, because they have different buyers and opposite trajectories:

    • Legacy flow and path testing — flat to declining as platform bundling spreads beyond AWS. Call it $120–180m.
    • AI agent behavioural testing — genuinely new, currently in the high single-digit millions, growing fast off a small base. At a mature 2–4% attach to agent vendor revenue, $150–400m.

    Combined: roughly $270–580 million by 2030, central case near $400m. Widely-quoted projections above that figure are extrapolating the growth of the AI agent market itself, not the assurance layer attached to it.

    Where this model is weakest

    Stated plainly, so that anyone using these figures knows what they are relying on:

    • No primary revenue figure exists for the largest vendor. Cyara has never published an absolute revenue number, and no regulator requires it to. The $72–88m rests on a company-substantiated growth rate applied to a level derived from third-party estimators.
    • The Hammer split is judgement. No source anywhere separates Infovista's contact centre testing revenue from its network testing revenue. The 45–55% assumption inside Empirix has no supporting citation.
    • PumpCX is unverifiable in principle. It trades through a discretionary trust, so no filing will ever exist, and its headline market-share claim does not appear in its own press releases.
    • Entity-level filings understate groups. Occam and CloudCX both have activity outside the UK entity whose accounts we cite. Those are floors, not totals.
    • The addressable market range is wide for a reason. No published source gives a global distribution of contact centres by seat band. ContactBabel publishes one for the UK, behind a paywall, and we have not bought it.

    We would rather publish a defensible range with the soft spots labelled than a single confident number that cannot be interrogated. If you have primary data that moves any of these figures — particularly filed accounts we have missed — we would like to see it.

    Sources

    Figures are estimates unless attributed to a filing or a company statement. This page is reviewed annually. Researched and prepared by AI for Nopaque.

    Want us to explore your IVR?

    TotalPath runs the same kind of test against your stack. Real audio. Real findings.