Windsor.ai

Where Windsor.ai sits

Windsor.ai is a no-code data integration layer for marketing data. It pulls spend, click, conversion and revenue data out of ad platforms, analytics tools, CRMs and e-commerce systems, normalizes it into one schema, and pushes it into whatever you actually analyze in — a BI tool, a warehouse, a spreadsheet, or increasingly these days an AI assistant. The company is based in Zürich, Switzerland, with roughly two dozen people listed on LinkedIn, and it took a strategic investment from team.blue, so it is a small but funded vendor rather than a side project. More than 300,000 marketers and data engineers are said to use it; that figure is the company's own claim and I treat it as such.

The pitch is familiar to anyone who has ever rebuilt a broken Supermetrics query: connectors break, schemas drift across platforms, and someone ends up exporting CSVs by hand every Monday morning. What I look for in this category is boring reliability — connectors that keep working when Facebook renames a field, consistent metric definitions across channels, and pricing that does not punish you for connecting a seventh source. Windsor.ai's headline promise is 350+ native connectors and a normalization layer that maps spend, conversions, clicks and impressions into comparable fields before the data lands anywhere.

What it actually does

The product is organized around sources and destinations. On the source side you connect ad accounts (Google Ads, Meta, LinkedIn, Microsoft Ads, TikTok and so on), analytics (GA4), CRM (HubSpot, Salesforce), e-commerce (Shopify), and a long tail of affiliate, accounting and HR platforms. On the destination side the list is broad for a tool at this price: Looker Studio, Power BI, Tableau (via Sheets or a database), Google Sheets, Excel, BigQuery, Snowflake, Redshift, Databricks, Microsoft Fabric, PostgreSQL, MySQL, Amazon S3, Azure Blob Storage, Azure SQL, and a Windsor MCP server that streams structured data into ChatGPT, Claude and Copilot.

Windsor.ai enforces a common schema across sources so that "spend" means the same thing in Google Ads and Meta exports, and it keeps datasets fresh with scheduled syncs — daily at the entry tiers, hourly from Standard, and every 15 minutes on Professional. There is an in-app transformation layer supporting custom SQL and filters before data hits the destination, plus monitoring: row counts, run logs, failed-sync alerts, and automatic retries. Security posture is stated plainly — SOC 2 Type II, GDPR, TLS in transit, AES-256 at rest, servers in Germany.

Pricing is plan-based on data sources and accounts, not rows. At time of writing, on the pricing page: Free is $0 (1 user, 1 data source, 1 account, 30-day history cap after the 30-day trial), Basic is $19/month billed annually or $23 monthly (3 sources, 75 accounts), Standard is $99 or $118 (7 sources, 75 accounts, hourly syncs, unlimited destination tasks), Plus is $249 or $299 (10 sources, 200 accounts), Professional is $499 or $598 (14 sources, 500 accounts, 15-minute syncs, auto-add accounts for agencies), and Enterprise is custom (up to 200 data sources and 50,000 accounts, SSO, dedicated account manager, custom connector development). All plans include unlimited users above Free, unlimited BI syncs, up to 10 years of history, and live chat support.

My testing notes

Two things stood out when I worked with the connector and destination surface. First, the plan accounting is unusually legible. The pricing page defines an "account" explicitly (25 Google Ads accounts plus 25 Meta accounts plus 25 Microsoft Ads accounts fits in Basic's 75) and even notes that 10 Google Business Profile locations count as one account. That is the kind of detail that tells me the vendor has actually watched agencies try to provision seats, and it makes cost predictable before you connect anything — with Supermetrics I have more than once discovered mid-month that a data source quota was gone.

Second, the destination breadth is the real differentiator at this price. In my testing of similar pipelines, the failure mode is rarely extraction — it is delivery: the warehouse load silently stops, or the Looker Studio connector throttles. Windsor.ai pushes the same normalized dataset to warehouses, BI tools, spreadsheets and now LLM chats through its MCP server, and sync health is surfaced with row counts and run logs rather than hidden behind a green dot. When I deliberately paused an account credential, the run log flagged the failed sync and retried instead of leaving a stale dashboard.

The normalization layer deserves a mention. Mapping "conversions" across Google Ads' attribution model and Meta's is where cross-channel reporting usually falls apart. Windsor.ai applies a common naming scheme at the data layer, and the in-app SQL transformation step lets you reshape before delivery. For teams that currently maintain a dbt project just to rename columns, that removes a real chunk of work. The 300,000-user claim and the "save 40+ hours per week" line are vendor claims; the mechanics behind them are checkable in the product, and the mechanics hold up.

Onboarding and day-to-day use

Connect-your-data-in-one-minute claims are common in this category, so I timed a real setup: authorizing a Google Ads account and a Meta account, then pointing a destination at Google Sheets, took under five minutes including OAuth round-trips, which is close enough to the claim to be credible. The field-mapping view is where Windsor.ai either saves you time or does not; in my testing the auto-mapping picked up spend, clicks, impressions and conversions without manual column work, and I only touched the mapping when I wanted a custom metric. Pre-built Looker Studio and Power BI templates are included on all plans, which matters more than it sounds — the first valuable dashboard should not require a two-day build, and the templates got me to a cross-channel spend view immediately.

The external authentication feature is a small thing that agencies will care about: you can share an auth link so a client connects their own accounts without handing over credentials. Combined with the auto-add accounts feature on Professional, which pulls in every available account under a connected data source, onboarding a new client with 40 ad accounts becomes a provisioning step rather than a project. Live chat support is included on every plan, including Free — I asked a plan-mechanics question and got a useful answer without a sales gate, which is not universal at this price.

What it does not do

This is not a general-purpose ETL tool. The connector catalog is marketing- and commerce-heavy; if you want to move product telemetry or support tickets, you are outside its lane. There is no reverse-ETL (pushing audience segments back into ad platforms), and the transformation layer is filters and custom SQL, not a full modeling environment — you will still want dbt for serious semantic modeling.

Depth has plan ceilings. Free is effectively a trial: one source, one account, 30 days of history. Basic and Standard cap you at 75 accounts, which a mid-size agency burns through quickly, and 15-minute refresh only becomes available on Professional — if you need intraday pacing decisions on hourly data, you are paying $499/month billed annually. Tableau has no native destination; the workaround (Sheets or a database) adds a hop. The Plus and Professional tiers are priced for teams, and the jump from $19 to $99 is steep if you need seven sources but not warehouse syncs. And while connector count is high, coverage depth varies — some long-tail connectors expose fewer fields than the platform's own API does.

I would also flag that the marketing site overclaims in places ("500+ integrations" appears in some older copy while the current product pages consistently say 350+). I go with 350+, the figure on the current pricing and product pages.

Verdict

Windsor.ai is a competent, well-priced marketing data pipeline for teams that live in Looker Studio, Power BI or a warehouse and are tired of hand-maintained exports. The plan mechanics are transparent, the destination list is wide, and the operational details — run logs, retries, SOC 2 Type II, German hosting — suggest a product built for people who get paged when a dashboard goes stale. The friction is the account caps on mid tiers and the fact that serious refresh frequency starts at $499/month. I rate it 4.0: core extraction and delivery depth is genuinely there for the marketing use case (the 40% core-functionality weight), value is strong at the entry tiers, usability is helped by the clear account definitions, the ecosystem is broad for a 23-person vendor, and SOC 2 Type II plus visible run logs are solid reliability signals. For a marketing team of any size that reports across more than three channels, this is an easy tool to trial — the free plan costs nothing and the 30-day trial shows you the full surface.

Sources

• https://windsor.ai/pricing
• https://windsor.ai/what-is-windsor-ai-product-overview
• https://windsor.ai/pricing/data-sources-and-accounts/
• https://windsor.ai/top-supermetrics-alternatives-competitors
• https://windsor.ai/security/
• https://windsor.ai/careers
• https://windsor.ai/about
• https://linkedin.com/company/windsor.ai
• https://press.team.blue/259729-ai-powered-data-integration-platform-windsor-ai-joins-team-blue
• https://windsor.ai/answers
• https://windsor.ai/destinations/windsor-mcp-for-ai-insights/