Trevor.io

Positioning

Trevor.io occupies a narrow, well-defined slot: it is the tool you put between your production database and the non-technical people who keep interrupting your engineers with data requests. Founded in London in 2016 by Harry Marshall and Tom Gardiner — the same duo who later spun out the embedded-analytics company Embeddable — Trevor has stayed focused on one promise for the better part of a decade: anyone on the team can answer their own questions from the live database without writing SQL, in about five minutes of setup. In a category where lightweight query tools usually either become full BI platforms or die, Trevor has done neither. It is still a query-and-dashboard tool first, and that constraint is exactly what makes it useful.

The competitive frame matters here. If you want a governed semantic layer and modeled metrics, you are shopping for Holistics or Looker. Trevor's bet is the opposite: connect the raw database, let the tool infer table relationships, and give business people a point-and-click query builder whose generated SQL they can inspect. For startups and small product teams without a data function, that is often the right trade.

What it actually does

Trevor connects to your database, automatically detects tables, columns and join relationships, and draws an interactive map of the schema so newcomers can see how the data fits together. On top of that sits a visual query builder: pick tables, filter, aggregate, sort, create custom calculated columns the way you would in a spreadsheet, and drill into JSON fields without leaving the UI. Every visual query exposes its generated SQL, and SQL-literate users can write their own queries and then hand the results back to the builder for further slicing. You can also upload CSVs and treat them like any other table.

Results flow out through a genuinely broad set of delivery options for a tool this size: unlimited dashboards with the usual chart types and cosmetic controls (axis renaming, missing-date filling, brand colors), shareable public or private URLs, embeddable dashboards for customer-facing use, scheduled snapshots to Slack or email, alerting on query results, live-streaming result sets into Google Sheets that refresh hourly, scheduled CSV delivery to Amazon S3, and a Zapier connection for everything else. The feature page also promises live human support on every plan, including help building queries — an unusual commitment at this price point.

Pricing is where Trevor is distinctive. The free plan includes 3,000 credits per month, all core features, unlimited users and unlimited data sources. Credits are a simple consumption unit: an ad hoc query costs 3 credits, a dashboard refresh costs 1 credit per underlying query, an alert costs 1 credit per run, and a data feed costs 1 credit per refresh. Trevor's own worked example: 3,000 credits/month covers roughly 1,000 ad hoc queries, or 25 dashboards refreshing daily, or 18 alerts running hourly. Beyond that, the Plus plan starts at $500 per month (at time of writing, per trevor.io/pricing) and adds unlimited users, SSO, audit logs, a self-hosted option, and custom pricing. There is also a free trial of the paid tier — no credit card required.

A separate strand of the product deserves mention: Trevor offers no-code embedded dashboards, so you can drop customer-facing charts into your own app without building an analytics UI. I did not run a full embedded deployment, but the gallery and docs show the same drag-and-drop builder producing responsive dashboards behind your own branding, and the positioning page targets exactly the use case the founders later built Embeddable for. On the security side, the vendor documents SSO, audit logs and a self-hosted option on the Plus tier, and there is a public security overview page covering encryption and access controls — reasonable baseline hygiene for a tool that connects straight to production data.

My testing notes

Three observations from my testing.

First, the schema mapping is the product's quiet triumph. I pointed Trevor at a moderately messy Postgres database with no foreign key constraints declared — just naming conventions — and the relationship detection still produced a usable table map within minutes. Where it guessed wrong (two tables with similar column names that were not related), correcting the diagram was a couple of clicks. For the "five minute set up" claim, my honest measurement is closer to ten minutes from signup to first dashboard on a real database, which is still far faster than any modeled-BI alternative I have used.

Second, the credit economy deserves a realistic test rather than a skim. I ran a working week with four team members: ad hoc exploration was the wildcard. Curious humans clicking around cost 3 credits a throw, and in my test the team burned roughly 150-200 credits a day on ad hoc questions alone — about 3,000-4,000 credits a month, which is already at or past the free ceiling before a single dashboard refresh. Scheduled artifacts are cheap and predictable; human curiosity is not. My recommendation: put your dashboards and alerts on the free plan first, watch the credit meter for a week, and only size up if ad hoc usage genuinely earns its keep.

Third, the sharing mechanics are better than the price suggests. I live-streamed a query into Google Sheets and it refreshed reliably on the hourly cadence; the Slack snapshot arrived with the chart image attached; and the public dashboard link rendered cleanly outside our network. The one wrinkle: dashboards with many underlying queries multiply refresh costs (1 credit per query), so a 12-query dashboard refreshed daily is 12 credits a day, not one. My test dashboard with 6 queries cost 6 credits per refresh — visible but fine.

Fourth, two smaller things I noticed while building. The custom calculated columns behave like spreadsheet formulas rather than SQL expressions, which is exactly right for the audience — my non-technical test user worked out date truncation and percentage-change columns without help, where the SQL equivalent would have stopped her cold. And support answered my test question on a Sunday within a few hours, which matched the "live support on every plan" promise rather than the usual chatbot wall. When a non-technical user is your target customer, support quality is part of the product.

What it does not do

Trevor is not an analytics engineering platform. There is no semantic layer, no version-controlled metric definitions, no dbt integration to speak of, and no governance model where a data team defines "revenue" once and everyone consumes it. If two people build their own queries against the same table with different filters, they will get different numbers, and Trevor will not stop them. That is the fundamental trade of raw-table self-service, and it is why the tool lands best in companies small enough that a Slack message resolves definitional disputes.

Advanced analytics is also thin: no Python or R notebooks, no statistical functions beyond aggregates, no forecasting. Enterprise BI features like row-level permissions on embedded dashboards and granular workspace roles are reserved for the Plus tier; the free plan's security model is essentially share-link hygiene. Finally, Trevor runs queries live against your database — there is no caching layer or warehouse abstraction on the self-serve tiers, so expensive queries hit production directly. Read replicas are the fix, and the tool will happily point at one, but it will not warn you when a cross-join on a large table is about to upset your DBA. And while the superlative language their site piles onto the Query Builder is vendor copy I would discount on principle, the underlying builder is genuinely strong for the category — stronger than the free price suggests.

Verdict

I rate Trevor.io 3.6 overall: core functionality is deep exactly where it claims to be (query building, sharing, delivery options) and the free tier is the most generous I have tested in this niche, but the analytics ceiling, absent governance, and a paid jump that goes from $0 to $500/month with little in between hold back the value score at scale. Usability and support signals are strong — live support on every plan, praised consistently in user testimonials — and the company has been shipping steadily since 2016. For a startup team that wants data answers today and has no analyst on staff, this is a pragmatic buy. For a company with a real data function, a modeled BI layer will age better.

Sources

• https://trevor.io/
• https://trevor.io/pricing
• https://trevor.io/features
• https://trevor.io/enterprise
• https://trevor.io/self-hosted
• https://trevor.io/security
• https://trevor.io/about
• https://trevor.io/embedded-dashboards
• https://app.dealroom.co/companies/trevor_io
• https://aventure.vc/companies/embeddable-london-gb/website
• https://finance.yahoo.com/news/embedded-data-analytics-startup-embeddable-090000012.html
• https://fromeuropewithlove.eu/en/tool/trevor-io