General Information
What AnomalyGuard is, who it is for, and what problem it solves.
What is AnomalyGuard?
AnomalyGuard watches time-series metrics in your databases – sales, occupancy, volumes, margins, and similar KPIs – and automatically flags meaningful changes: spikes, drops, gaps, trend shifts, and more. Results are ranked and delivered on Home, through role-based filters, or via API / CLI.
How is it different from dashboards and BI tools?
Dashboards show what happened. AnomalyGuard tells you what changed and deserves attention – so teams stop hunting exceptions and start deciding.
Why does anomaly detection matter?
Late signals cost money. Catching metric shifts early reduces incident impact, protects revenue, and surfaces opportunities hidden in noisy data.
Who is it for?
Data / BI teams with dashboards but weak alerting, CTOs who do not want to own custom anomaly ML, and ops / product teams where metrics hit P&L.
Do I need a data science team to run it?
No. Analysts and admins familiar with your warehouse can connect sources, define data views, and tune detectors. Advanced users can also manage definitions via YAML, CLI, and CI.
Can it be used beyond finance?
Yes. Any daily (or periodic) business metric works – retail, hospitality, logistics, manufacturing volumes, digital funnels, and more.
Product & Scalability
Capabilities, integrations, and how AnomalyGuard fits your stack.
What is a data view?
A data view is one monitoring definition: source query, business dimensions, detectors, and jobs. Example: “sales by country and channel.” Plans are licensed by how many data views you create – not by seats.
Which databases can I connect?
Common warehouse and operational databases such as PostgreSQL, Microsoft SQL Server, Snowflake, Oracle, and more. AnomalyGuard reads what you configure; it does not require migrating your warehouse.
What do I need to run it?
AnomalyGuard (Azure Marketplace managed app or Docker) plus an external PostgreSQL database for its own storage, and Microsoft Entra ID for sign-in.
Too many alerts / too much noise?
Tune detectors and sensitivity, use aggregation where it fits, and deliver role-based filters so each audience sees ranked priorities – not every raw fluctuation.
Do you offer API and CLI?
Yes. REST API and the anomalyguard CLI support automation, CI/CD for data-view definitions, and AI agents that pull anomalies and summaries without opening the UI.
How do role-based filters work?
One data view can power many saved filters – company-wide, regional, product, or team lenses – so HQ and local ops share detection but not the same noise.
Is it scalable?
Yes. Deploy on Azure Container Apps with configurable worker capacity, or run Docker in your environment. Scale compute to your load; grow product usage by adding data views.
Cloud, on-prem, or hybrid?
Azure Marketplace for managed deploy in your subscription, or Docker on your infrastructure. Your source data stays where you already keep it.
Can we manage access with our identity provider?
Yes – Microsoft Entra ID sign-in, with permissions for who can administer vs. who can consume specific data views and results.
Pricing & Licensing
How plans work, what you pay for, and what Azure bills separately.
How is AnomalyGuard licensed?
By the number of data views you can create. Users, filters, comments, and API keys are not seat-licensed – the whole organization can consume results under the same plan.
What are the public plans?
Start (1 data view), Small Team (5), Growing Team (20), Enterprise (no data-view cap). See current list prices on the Pricing page.
Is there a free Demo?
Yes – a private Demo plan (time-limited, up to 5 data views) is available for approved evaluations. Contact us to request access.
Do more users cost more?
No. Unlimited users on every plan. You upgrade when you need more monitoring definitions (data views), not when more colleagues open Home.
Is Azure infrastructure included in the software price?
No. Marketplace software pricing is separate from Azure usage (Container Apps, PostgreSQL, networking, logs). You pay Microsoft for the resources you run.
Can we buy annually?
Yes – annual options are available (typically about two months free vs. monthly). Enterprise and larger deals can also use private offers.
How do we upgrade?
Change Marketplace plan / offer when you need a higher data-view limit. The in-app license cap follows the plan you deploy.
Business Value & Rollout
ROI, timeline, and how teams typically go live.
What ROI should we expect?
Faster detection of revenue and ops issues, less analyst time spent on exception hunting, and fewer “missed it until the weekly pack” surprises. Highest impact in domains where delayed signals are expensive.
How long to first value?
Often days to a few weeks: deploy, connect one high-value metric, tune detectors, pin a filter on Home. Broader multi-domain rollouts typically land in 1–4 weeks depending on data readiness.
Can we roll out gradually?
Yes – start with one data view (Start plan), expand to a domain team (Small Team), then cross-department coverage (Growing / Enterprise).
How heavy is it on IT?
Lightweight compared with building anomaly ML in-house. You provide Azure (or Docker host), PostgreSQL, and Entra ID app registration; AnomalyGuard handles detection and delivery.
Do you replace our BI tool?
No. AnomalyGuard complements Power BI / Looker / Tableau – it watches metrics and prioritizes change; your BI stack remains for exploration and reporting.
Can partners or your team help implement?
Yes – we and certified partners can support onboarding, first data views, and detector tuning. Reach out via Contact.
Where can I learn more?
Product docs at doc.anomalyguard.com, plus How it works, Features, and Benefits.
Data Security & Privacy
Where data lives, what leaves your environment, and how access is controlled.
Where does my data live?
In your infrastructure. Source databases stay yours. AnomalyGuard runs in your Azure subscription (Marketplace) or on your hosts (Docker), with its own PostgreSQL that you control.
Do you send data to external AI / LLM vendors?
No. AnomalyGuard does not send your data to third-party AI platforms. Optional summarization can use an LLM provider you choose and configure – under your control.
Is processing local to our environment?
Yes – detection runs where you deploy. There is no requirement to copy your warehouse into a vendor SaaS data lake.
Can we run fully on-premises?
Yes via Docker, with your PostgreSQL and Entra ID (or equivalent setup per docs). Ideal when cloud egress of operational data is restricted.
How is access secured?
Microsoft Entra ID authentication, role and data-view permissions, and API keys for automation. You decide who sees which results.
Do you store our connection strings safely?
Connector secrets are stored encrypted in your AnomalyGuard database – not in plaintext configs handed around by email.
Still have questions?
Tell us about your metrics and stack – we will help you pick the right plan and first data view.