Data Quality Plane · Profiling & Anomaly Detection
Automated data quality monitoring that uses machine learning to detect anomalies.
Anomalo monitors data quality in warehouses and lakes, and the reason teams look at it is that it finds problems nobody thought to write a rule for.
Its approach is published as four pillars: observability on table metadata, validation rules you write, key metrics you care about, and unsupervised machine learning over the data itself.
The machine learning pillar is the differentiator. It learns a representation of the typical data in a table and flags when incoming data is meaningfully different, covering distributions, categorical values and even relationships between columns.
Crucially it learns a threshold per table, accounting for how much background noise each column carries, which is how it separates real issues from normal variation rather than applying one global rule.
Two models run on every monitored table daily with no configuration at all: one watching for spikes in missing values, one for anomalous records. That zero-configuration coverage is the practical selling point.
Alerts carry a severity score, root-cause analysis pointing at the likely source, and downloadable samples of good and bad rows, with automatic upstream and downstream lineage.
Feedback is part of the design. Users confirm or dismiss findings, thresholds are one click from the alert, similar issues cluster into a single alert, and secondary checks suppress false positives.
It can run in your own network. In-VPC deployment keeps row-level data, metrics and metadata inside your environment, with SaaS as the alternative, plus a Snowflake native app and an Azure marketplace listing.
In April 2026 it repositioned as an agentic platform of nine agents, five shipping and four coming, under the banner of self-driving data, with a separate self-serve product, Anomalo Analyst, launched later that year.
One caution when evaluating: its documentation is private to customers, so everything published is marketing-level rather than reference-level, and no pricing appears anywhere.
The same headings are used for every profiling & anomaly detection entry, so two tools can be read side by side.
Price on requestQuote only; 30-day trial at no charge
Quote-only. There is no pricing page and no pricing unit published, whether per table, per row, per check or per seat; the subscription agreement refers throughout to fees set out in an order. What is published: fees in US dollars, one-year terms auto-renewing with 60 days' notice, and a trial period of 30 days at no charge with limits set at Anomalo's discretion. A 2022 post mentions a free trial capped at five tables, which is not restated on any current page. Anomalo Analyst has a self-serve free start with no published tier limits.
Anomalo was founded in 2018 by Elliot Shmukler, its chief executive, and Jeremy Stanley, who had worked together on data quality problems at Instacart. It is fully remote across the United States and EMEA and publishes no headquarters. It is independent and venture-backed, with a $5.95m seed in December 2020, a $33m Series B in January 2024 led by SignalFire with Databricks Ventures taking it to $72m total, a $10m extension in November 2024 from Smith Point Capital, and an undisclosed strategic investment from Snowflake Ventures in March 2025. Named customers include Block, Discover, Atlassian, Notion and Equifax; no total is published.
Founded 2018 · anomalo.com
Data Quality Plane · Profiling & Anomaly Detection
Python library that generates exploratory data profiling reports from DataFrames. Renamed in April 2026 from ydata-profiling to fg-data-profiling, imported as data_profiling; the old package no longer receives updates.
Drafted with AI assistance and checked against the vendor’s own documentation.
