What Teams Actually Pay After Switching
Observability spend is becoming a board-level line item. At mid-scale organizations running 30TB/month of telemetry, monitoring costs routinely exceed $20,000/month – and that figure only captures the invoice. Cloud egress fees, per-seat charges, and custom metrics overages push the real number higher than most finance teams realize.
The problem is not that teams chose the wrong platform. It is that pricing models designed for low-volume pilots behave very differently at production scale. Per-user fees that looked manageable with five engineers become a $7,000/month line item at twenty. Ingest pricing that seemed competitive at 1TB/month generates six-figure annual bills at 30TB. And the switching cost – dashboards, alerts, and queries locked in proprietary formats – keeps teams paying more than they should for longer than they want.
This guide evaluates seven New Relic alternatives through a cost lens: what teams actually pay at 30TB/month, where the hidden charges accumulate, and which pricing models stay predictable as you grow. For teams where data residency is also a factor, self-hosted alternatives eliminate cloud egress fees entirely – a $3,000/month saving that never appears on a SaaS invoice.
All cost estimates in this guide assume a mid-scale reference scenario: 30TB/month ingestion (~20TB logs, 7TB traces, 3TB metrics), 100 hosts, 20 full-platform users, 500,000 active metric series, and 30-day retention across all signal types. Only core observability is included – no security, profiling, or synthetics add-ons. Actual costs vary by retention settings, negotiated discounts, and feature mix.
What New Relic Actually Costs at Your Team Size
Before evaluating alternatives, it helps to see what New Relic actually costs across different team sizes – not list prices, but the combined bill when data ingest, user seats, and retention add up.
| Team Profile | Data / Users | New Relic /mo | CubeAPM /mo | Annual Saving | Saving % |
|---|---|---|---|---|---|
| Small team | 500 GB, 3 users | ~$458 | ~$75 | ~$4,596/yr | ~84% |
| Growing team | 5 TB, 10 users | ~$4,955 | ~$750 | ~$50,460/yr | ~85% |
| Mid-market | 30 TB, 50 users | ~$24,745 | ~$4,500 | ~$242,940/yr | ~82% |
| Enterprise | 200 TB, 150 users | ~$97,750 | ~$30,000 | ~$813,000/yr | ~69% |
New Relic costs: Standard plan, Original data ($0.40/GB beyond 100GB free), full platform users at $99 to $349 per user per month for full platform access. CubeAPM: $0.15/GB, no user fees. Enterprise pricing may include negotiated discounts not reflected here.
If you want to model your current New Relic bill before committing to a switch, the New Relic pricing calculator breaks down every cost dimension: data ingest, user seats, synthetics, and cloud egress fees most teams overlook.
The Real Cost of Staying on New Relic
The Seat Tax: What Collaboration Actually Costs
New Relic’s Standard plan charges $99/month per full platform user (maximum five). Pro costs $349/user/year. At 20 users, the seat bill alone is $6,980/month – before a single byte of data is ingested. The real cost is not just the invoice line. When engineering leads ration access to control spend, junior developers, support engineers, and product managers lose visibility. Incident response slows. Knowledge concentrates in a few hands. The organizational cost of limited access compounds faster than the billing cost.
CCU Pricing: When Your Bill Spikes During Incidents
New Relic’s Core Compute model removes seat limits but introduces Compute Capacity Units (CCUs) – a billing dimension that charges for queries, alerts, and data processing. The cost problem: CCU consumption spikes during incidents, precisely when engineers are running the most queries and dashboards are refreshing fastest. A platform whose costs increase when you need it most creates an adversarial relationship between operational urgency and budget discipline.
NRQL Lock-In: The Switching Cost Nobody Budgets For
Every dashboard, alert, and custom query your team has built in New Relic uses NRQL – a proprietary query language. When you evaluate alternatives, the sticker price delta is only part of the equation. The migration cost – engineering hours to rebuild your entire query library, re-validate alert thresholds, and retrain teams on new syntax – often exceeds a full quarter of observability spend. Platforms built on OpenTelemetry and SQL-compatible query engines avoid this compounding lock-in from day one.
Cloud Egress: The Cost That Never Appears on Your Observability Invoice
When you send telemetry to any external SaaS platform – New Relic, Datadog, or any cloud-hosted alternative – your cloud provider charges approximately $0.10/GB for data leaving your VPC. At 30TB/month, that is $3,000/month in AWS or GCP egress fees. Over three years, that is $108,000 that does not appear on a single observability invoice but lands squarely on your cloud bill. Self-hosted platforms running inside your VPC have zero data-out cost.
What Drives Observability Costs at Scale
Most observability tools can collect logs, metrics, and traces. The cost differences at scale come down to how billing models behave as data volumes grow, more users need access, and more features get activated.
- Pricing model and billing predictability: GB-based pricing scales linearly – double the data, double the cost. Host-based, seat-based, and multi-dimensional models (hosts + custom metrics + indexing + spans) compound unpredictably. The gap between the cheapest and most expensive platforms at 30TB/month is over 10x.
- The custom metrics trap: Some platforms charge $5-$8 per 100 custom metric timeseries per month. At 500,000 active series, this adds $25,000-$40,000/year in charges that teams rarely budget for upfront. OTel-native platforms that treat all metrics equally avoid this entirely.
- Deployment model and egress cost: SaaS platforms incur cloud egress fees (~$0.10/GB). Self-hosted platforms eliminate this. At 30TB/month, the deployment model alone creates a $3,000/month cost difference before any license fees.
- Retention pricing: Some platforms include retention in the base price. Others charge separately for ingestion, processing, and storage – each with its own rate card. Regulated teams needing 90-day retention can see costs double on platforms with per-day retention fees.
- OpenTelemetry support – native vs bolt-on: OTel-native platforms ingest OTLP data without transformation. Platforms that added OTel as a layer on proprietary agents sometimes bill OTel metrics as custom metrics – turning an open standard into a cost liability.
- Observability depth (MELT coverage): Metrics, Events, Logs, and Traces. Platforms that require separate products for each signal type multiply licensing costs. Unified platforms covering all four signals in one license offer simpler cost structures.
- Ease of migration from New Relic: Can existing instrumentation be reused? OpenTelemetry-based migration lets you switch backends with a configuration change rather than re-instrumentation – the fastest path to cost savings.
1. CubeAPM
Best for: Lowest TCO at scale with predictable pricing and zero egress cost
Overview
CubeAPM is an OpenTelemetry-native, full-stack observability platform built for teams that want unified visibility across metrics, events, logs, and traces without the cost sprawl common in SaaS observability tools. It runs inside your own AWS, GCP, or Azure VPC – traces, logs, and metrics never leave your infrastructure boundary. CubeAPM handles upgrades, patches, and platform operations; you provide the infrastructure.
Ranked in the top 10 APM platforms in G2’s Spring 2026 APM Grid Report. Used by Policybazaar (insurance), Delhivery ($3.5B logistics), Mamaearth ($1.2B), world’s largest bus aggregator – redBus (NASDAQ-listed MakeMyTrip), Ola, and Practo (healthcare). SOC 2 Type II and ISO 27001 certified.
Key Features
- Full MELT observability: Metrics, events, logs, and traces in one platform with a single investigation workflow
- OpenTelemetry-native: Built from the ground up on OTel. No proprietary agent required. Compatible with existing Datadog, New Relic, Elastic, and Prometheus agents for incremental migration
- Self-hosted, vendor-managed: Deploys in your VPC. Zero cloud egress cost (saves ~$3,000/month at 30TB vs any external SaaS)
- AI-based Smart Sampling: Reduces low-value telemetry volume while preserving high-value traces
- Unlimited retention: Included in pricing – no separate retention charges
- 800+ integrations: Kubernetes, synthetic monitoring, RUM, and error tracking included
Pricing
Flat $0.15/GB ingested. No per-user fees. No per-host charges. Unlimited users and unlimited data retention included. The pricing model has exactly one variable: data volume.
At 30TB/month: ~$5,100/month all-in ($4,500 license + ~$600 infra)
Delhivery: 75% cost reduction. Mamaearth: ~70% savings, migrated in under an hour. redBus: 4x faster dashboards, 50% faster MTTR.
Pros
- 70-75% lower cost than enterprise APM at scale
- Complete data ownership – telemetry never leaves your VPC
- Predictable pricing with no hidden billing dimensions
- Zero cloud egress cost
- Fast migration – multiple customers report under-an-hour setup
Cons
- Requires self-hosted deployment in cloud or on-prem; may not suit teams looking for a SaaS-only model.
- AI/ML anomaly detection growing but not as mature as Dynatrace Davis AI
2. Datadog
Best for: Broad SaaS ecosystem coverage with the budget to absorb multi-dimensional billing
Overview
Datadog is the largest commercial observability platform and New Relic’s most direct competitor. Its integration catalog (900+) and feature breadth are unmatched – APM, logs, security, RUM, synthetics, and network monitoring under one roof. The trade-off is cost: host-based pricing compounds quickly at scale, and custom metrics charges are a persistent source of bill shock.
Key Features
- Unified observability: metrics, logs, APM, RUM, synthetics, security, database monitoring
- 900+ integrations – largest ecosystem in the category
- Kubernetes Explorer with pod, deployment, and resource visibility
- Synthetic monitoring across browser, API, and network-level tests
- OpenTelemetry support via OTel Collector and Datadog Agent
Pricing
Multi-dimensional billing: hosts + custom metrics + log ingestion ($0.10/GB) + log indexing (~$1.70/million events) + APM spans + RUM sessions. OTel metrics are often billed as custom metrics. This is the most complex pricing model in this comparison – modeling your actual cost requires input across six or more billing dimensions.
At 30TB/month: ~$30,000-$45,000+/month
Breakdown (30% logs indexed): 100 hosts ~$2,400 + log ingest 20TB ~$2,000 + log indexing ~$30,000 + APM spans ~$3,000-5,000 + custom metrics ~$5,000+. Log indexing is the dominant cost driver. Teams switching from New Relic’s per-GB model are often surprised by how log indexing costs compound – platforms with flat per-GB pricing avoid this entirely.
Pros
- Best-in-class integration ecosystem and product breadth
- Strong Kubernetes, container, and cloud-native monitoring
- Watchdog AI proactively surfaces anomalies
- Mature CI/CD and deployment tracking
Cons
- Billing complexity: host fees + custom metric overages + log indexing combine unpredictably
- OTel metrics billed as custom metrics – adds cost for teams adopting open standards
- SaaS-only – not suitable for teams with data residency requirements; self-hosted platforms are worth evaluating for those use cases
- Total cost at scale significantly exceeds alternatives with simpler pricing models
3. Dynatrace
Best for: Large enterprises where AI automation justifies the premium
Overview
Dynatrace differentiates with its Davis AI engine, which automatically maps service dependencies and performs causal root-cause analysis. Gartner ranks Dynatrace highest in “Ability to Execute” among observability vendors. The platform targets large enterprises with complex, fast-moving microservice estates.
Key Features
- Davis AI: Automatic baselining, anomaly detection, and probable-cause analysis
- Full-stack monitoring via OneAgent with automatic service discovery
- OpenTelemetry support via OTLP API, OTel Collector, and Dynatrace Collector
- Dedicated Kubernetes observability with flexible deployment via Dynatrace Operator
- Log management with separate ingest, processing, and retention pricing
Pricing
Usage-based with separate rate-card units. Full-Stack Monitoring at $0.01/memory-GiB-hour, Log Management ingest/process at $0.20/GiB, retain at $0.0007/GiB-day. The memory-GiB-hour model means costs scale with both host count and memory allocation – making it harder to forecast than flat per-GB pricing.
At 30TB/month: ~$20,000-$35,000+/month
Breakdown: 100 hosts x $0.08/hr x 8 GiB x 730 hrs ~$4,700 + log ingest 20TB x $0.20/GiB ~$4,100 + log retention ~$430 + traces/metrics/APM + commitment overhead.
Pros
- Best automated root cause analysis in the market
- Automatic full-topology discovery – minimal manual configuration
- Managed deployment option for data residency (Dynatrace Managed)
- Strong compliance and enterprise security features
Cons
- Proprietary OneAgent creates its own vendor lock-in
- Memory-GiB-hour pricing is harder to estimate than flat per-GB models
- Log retention billed separately from log ingestion
- Davis AI requires a baselining period – new deployments do not get full value immediately
4. Grafana Cloud (LGTM Stack)
Best for: OTel-first teams that want cost transparency with open-source foundations
Overview
Grafana Labs assembled the LGTM stack – Loki (logs), Grafana (dashboards), Tempo (traces), Mimir (metrics) – into a coherent observability platform. Grafana Cloud is the managed version. Paired with Grafana Alloy (an OTel Collector distribution), it provides dedicated OTLP endpoints that auto-route signals to the right backend. Best for teams already running Prometheus who want to extend to full observability without a forklift migration.
Key Features
- LGTM stack: Mimir for metrics, Loki for logs, Tempo for traces
- Grafana Alloy: OTel Collector distribution with built-in Prometheus pipelines
- Strongest dashboarding and visualization across multiple telemetry sources
- k6 performance testing integrated into the observability ecosystem
- Cost attribution features for metrics, logs, and traces
Pricing
Usage-based across telemetry types. Logs: $0.05/GB process + $0.40/GB write + $0.10/GB retain. Traces: same structure. Metrics: $6.50/1k active series. Platform fee: $19/month. The per-signal pricing means teams can optimize costs by reducing individual signal volumes – Adaptive Metrics and Adaptive Logs features help here.
At 30TB/month (managed cloud): ~$15,000-$20,000+/month
Breakdown: 20TB logs ~$11,000 + 7TB traces ~$3,500 + 500K metric series ~$4,000 + base. Adaptive Metrics/Logs features can reduce this materially.
Pros
- Fully OTel-native – no custom metrics penalty
- Adaptive Metrics/Logs actively help reduce billing
- Strong open-source community; highly customizable
- Self-hosted path available for cost-driven teams with operational capacity
Cons
- No native APM out-of-the-box – requires significant configuration
- Self-hosting at scale requires dedicated SRE expertise
- Usage-based pricing still grows with volume on managed cloud
- LGTM stack has a steep learning curve for teams new to Grafana
5. Elastic APM
Best for: Teams already on the ELK stack who want to add APM without a new line item
Overview
Elastic APM is the distributed tracing and application monitoring component of the Elastic Stack. For teams already indexing logs in Elasticsearch and visualizing in Kibana, adding APM is natural. It provides distributed tracing, service maps, error tracking, and MELT correlation. Note: Elastic APM’s OSS version reached end-of-service in September 2025 – evaluate managed options carefully for new deployments.
Key Features
- Native Elasticsearch integration: APM data correlates directly with log indices
- OpenTelemetry compatible across serverless, self-managed, and hybrid deployments
- Machine learning-based anomaly detection via Elastic ML
- RUM via JavaScript agent for frontend experience monitoring
- Available self-hosted (SSPL license) or Elastic Cloud
Pricing
Self-hosted is free; you cover infrastructure. Elastic Cloud: consumption-based. Serverless Observability: Logs Essentials from $0.07/GB ingested + $0.017/GB retained/month. The cost advantage here is strongest for teams that already operate Elasticsearch clusters – adding APM requires minimal incremental spend. For greenfield deployments, the operational cost of running Elasticsearch at scale should be factored in.
At 30TB/month (Elastic Cloud): ~$8,000-$15,000/month
Pros
- Zero incremental cost if already running Elastic for logs
- Strong log + trace correlation in the same query interface
- Self-hosted option keeps data on your infrastructure
- ML-based anomaly detection included
Cons
- Significant operational overhead to run self-hosted at scale
- KQL (Kibana Query Language) is less developer-friendly than SQL
- 2021 SSPL licensing change – review for open-source compliance
- APM experience is less polished than purpose-built APM tools
6. Splunk Observability Cloud
Best for: Enterprises with existing Splunk investments where the marginal cost is justifiable
Overview
Splunk Observability Cloud is known for real-time, full-fidelity monitoring across infrastructure, applications, and user interfaces. Its OpenTelemetry collection path uses the Splunk Distribution of the OTel Collector. Strong for enterprises that want mature SaaS monitoring with deep infrastructure and APM workflows alongside Splunk’s security portfolio.
Key Features
- Full-fidelity distributed tracing (no default sampling)
- AI-powered alerting with built-in noise reduction
- Deep Splunk SIEM and log analytics integration
- Real-time stream processing for telemetry
- Uses Splunk Distribution of the OpenTelemetry Collector
Pricing
Modular with separate packaging for infrastructure, APM, RUM, and synthetics. Infrastructure Monitoring from $15/host/month. App & Infra from $60/host/month. End-to-End from $75/host/month. At full-stack coverage, Splunk is consistently the most expensive option in this comparison – the cost is justifiable primarily when unified IT and security observability within an existing Splunk deployment offsets the premium.
At 30TB/month: ~$35,000-$60,000+/month
Pros
- Full-fidelity traces – no sampling means no blind spots
- Best-in-class integration with Splunk Security for unified IT and security observability
- Mature enterprise compliance story
Cons
- Among the most expensive platforms in this comparison
- Modular pricing with separate packaging for each capability
- Best value only with existing Splunk investment – overkill otherwise
- SaaS-first – not suitable for teams requiring self-hosted backends
7. Sentry
Best for: Developer-first teams where error resolution speed matters more than infrastructure depth
Overview
Sentry is known for developer-first error monitoring that has grown into a broader application monitoring platform covering errors, tracing, logs, session replay, profiling, and cron monitoring. Best for developer-led teams that want fast issue triage without adopting a heavier infrastructure-first observability platform.
Key Features
- Session Replay: Video-like reproductions of user sessions for web and mobile – a meaningful advantage for frontend debugging not available in most observability platforms
- Error monitoring with stack traces, breadcrumbs, and context
- Distributed tracing and performance monitoring
- Profiling and cron monitoring
- Self-hosted option available
Pricing
Event + usage-based. Team plan from $26/month base. Logs: $0.50/GB (Team PAYG). Spans: from $0.0000020/span above 5M. Sentry’s cost profile is moderate at high volume but scales differently than infrastructure-first platforms – the primary cost driver is event volume rather than data ingest or host count.
At 30TB/month: ~$15,260/month
Pros
- Best-in-class developer experience for error triage
- Session Replay provides video-like debugging not found in traditional APM
- Self-hosted option for data control
- Strong frontend and mobile debugging capabilities
Cons
- Primarily error and debugging focused – not full infrastructure observability
- Teams needing deep infrastructure monitoring will need a complementary tool
- Pricing at high volume can approach traditional APM costs
- Less suited for infra-first or SRE-led observability workflows
Cost Comparison at 30TB/Month Ingestion
| Tool | Est. Cost @ 30TB/mo | Pricing Model | OTel Native | Data Residency | Self-Hosted |
|---|---|---|---|---|---|
| CubeAPM | ~$5,100/mo all-in | $0.15/GB ingestion-based | Native | Always (in-VPC) | Yes (vendor-managed) |
| Elastic APM | ~$8K-$15K | Deployment-based | Partial | If self-hosted | Yes |
| Sentry | ~$15K-$32K | Event + usage | Partial | If self-hosted | Yes |
| Grafana Cloud | ~$15K-$20K+ | Usage-based | Native | If self-hosted | Yes |
| New Relic (ref.) | ~$20K-$25K+ | Ingest + per-user | Partial | SaaS only | No |
| Dynatrace | ~$20K-$35K+ | GiB-hour + commit | Partial | Managed option | Managed |
| Datadog | ~$30K-$45K+ | Host + feature-based | Partial* | SaaS only | No |
| Splunk | ~$35K-$60K+ | Host + enterprise | Partial | Limited | Limited |
* OTel metrics in Datadog are often billed as custom metrics. New Relic included as reference. All estimates use the methodology assumptions above. Vendor discounts and EDP commitments can significantly reduce SaaS costs.
How to Migrate from New Relic to an OTel-Native Platform
Migration anxiety is real – you have built dashboards, tuned alerts, and accumulated operational knowledge in New Relic over months or years. The good news: switching to an OpenTelemetry-native platform is structurally less painful than previous APM migrations. Once your services emit OTLP data, you can point that data at any compatible backend with a configuration change rather than re-instrumentation.
| Week | Focus | Key Actions | Exit Criteria |
|---|---|---|---|
| 1 | Instrument inventory | List every service using NR agents. Map data volumes per service. Choose target platform. | Full inventory. Platform selected. |
| 2 | Parallel run | Deploy OTel Collector alongside NR agents on 1-2 non-critical services. Dual-write telemetry. Compare dashboards. | Traces/metrics parity confirmed on pilot services. |
| 3 | Dashboard migration | Recreate top 10 critical dashboards and all active alerts. Validate alert accuracy. Remove NR agents from pilot services. | Critical dashboards live. Alert parity verified. |
| 4 | Full cutover | Roll OTel agents to remaining services. Cancel NR agents service-by-service. Run NR in read-only mode for 2 weeks. | All services on new platform. NR agents decommissioned. |
Practical note: Run both platforms simultaneously for at least two weeks before cancelling New Relic. Teams consistently discover dashboards they forgot existed, alerts that were silently firing, or integrations that depended on NR’s API. Document what each dashboard is measuring, not the NRQL syntax, before migrating.
Which New Relic Alternative Fits Your Cost Profile?
- Choose CubeAPM if your priority is lowest TCO with predictable billing. $0.15/GB, unlimited users, zero egress cost – the simplest cost model in this comparison.
- Choose Datadog if you can absorb higher costs for the broadest SaaS ecosystem. Model custom metrics and log indexing costs carefully – they are the primary bill drivers at scale.
- Choose Dynatrace if your organization values AI automation enough to justify the premium. Be prepared for annual commitments and memory-based pricing complexity.
- Choose Grafana Cloud if you want usage-based pricing with cost optimization levers. Adaptive Metrics and Adaptive Logs give you active controls to reduce spend.
- Choose Elastic APM if you already run ELK and want to add APM without adding another vendor invoice. The cost advantage disappears for greenfield deployments.
- Choose Splunk if your organization already has a Splunk investment and the marginal cost of adding observability is lower than onboarding a new platform.
- Choose Sentry if your team spends more time debugging application errors than infrastructure issues, and you want developer-first tooling at moderate cost.
When New Relic Is Still the Better Choice
New Relic is still the right choice for teams where the cost of switching exceeds the cost of staying.
- If New Relic’s 100GB free tier is suitable for your use case.
- You want one vendor covering APM, infrastructure, browser, synthetics, Kubernetes, and incident workflows in one place
- You are already heavily invested in the New Relic ecosystem (dashboards, alerts, NRQL queries) and migration cost exceeds the pricing delta
- You need native OTLP ingest inside a commercial SaaS platform without moving to a self-managed stack
- You need mature synthetic monitoring with scripted browser/API tests and private locations
- You are comfortable actively managing telemetry volume, CCU consumption, and ingest governance
- You want AI-assisted observability (New Relic AI, alert-coverage analysis) built natively into the platform
Final Thoughts
Observability cost is not just the invoice from your monitoring vendor. It is the combined total of license fees, cloud egress charges, engineering time spent managing billing complexity, and the organizational cost of rationed access. Most teams underestimate the gap between their quoted rate and their actual all-in spend by 30-50%.
The practical decision comes down to which cost structure matches how your organization actually operates. If budget predictability matters most, platforms with single-variable pricing – one rate per gigabyte, no per-host or per-seat multipliers – eliminate the billing surprises that erode trust between engineering and finance. Else if compliance or data residency is a hard requirement, self-hosted platforms structurally eliminate egress costs and keep the cost model simpler. If enterprise automation is the priority, the AI-led platforms justify their premium for complex environments where reduced MTTR offsets higher licensing costs.
Run the numbers at your actual data volume, user count, and retention requirements before committing. The cost difference between the cheapest and most expensive platforms in this comparison exceeds 10x at the same scale – and that gap only widens as you grow.
Frequently Asked Questions
1. What is the cheapest New Relic alternative in 2026?
For teams beyond New Relic’s free tier (1 user + 100GB/month), platforms with flat per-GB pricing and no per-user fees offer the lowest TCO at most team sizes. The cost advantage is most pronounced at mid-market scale (30TB/month), where single-variable pricing models cost 70-80% less than multi-dimensional billing. Cloud egress savings add another $3,000/month for self-hosted options.
2. Why is my New Relic bill higher than the listed price?
New Relic’s pricing has multiple compounding dimensions: data ingest ($0.40/GB), full platform user seats ($99/user/month on Standard), data retention upgrades, synthetics checks, and – often overlooked – cloud egress fees from your cloud provider. The listed per-GB rate is only one component. Teams consistently report that their actual bill is 40-60% higher than what they initially modeled.
3. How much can I save by switching from New Relic?
Savings depend on team size and data volume. At 30TB/month with 50 users, the difference between New Relic and the lowest-cost alternative in this comparison is approximately $20,000/month – over $240,000 annually. Smaller teams (5TB, 10 users) see annual savings around $50,000. Enterprise teams