StrategyNew

    Supabase Started Metering Your Logs. The Bill Arrives in 2027.

    On 25 September Supabase turned log ingest and log query into metered resources. Nothing is enforced yet - the meters run until early 2027 with the limits switched off. That gap is the whole problem: a number that cannot hurt you today is a number nobody watches.

    Supabase shipped two things on 25 September 2026. One of them was a Postgres minor release its own changelog marks as a breaking change. The other barely reads like news: usage-based pricing for logs. It has no deadline attached, it changes nothing about how your app runs today, and it is the one most Lovable-built projects will walk into unprepared.

    What actually changed

    The Supabase changelog entry introduces usage-based pricing for logs in two parts. Log ingest meters the volume of log data your services generate. Log query meters how much log data you scan, on a quota that scales with your ingest.

    The entry describes this as a soft launch. Usage is metered and visible in your project's usage dashboard, but limits are not yet enforced. A grace period runs through early 2027. During it, the usage dashboard shows log ingest and query meters with an "Upcoming" label, and warns if a project approaches its limits - informational only, with no enforcement behind it.

    Supabase also states that more than 90% of projects are within the included limits and will not be affected.

    Why "90% of projects" is not an answer about your project

    It is a true statement about a fleet on the day it was written. It is not a forecast about an application that has not launched yet.

    Log volume is a function of traffic and error rate. Both are supposed to rise if the product works. A prototype with eleven users and a founder clicking through it generates a log volume that tells you nothing about the same codebase with two thousand users and a payment path that occasionally fails. The projects sitting comfortably inside the limits today include a large number that have not yet done the thing they were built to do.

    The second problem is quieter. A prototype's logging configuration was not chosen. It was inherited.

    The settings that drive the number are on your side of the line

    This is where Lovable projects differ from a fully managed platform. Lovable designs your schema and runs your migrations, but the Supabase project it writes to is one you own, on your account and your billing. The database configuration is yours.

    Supabase's own guidance points at Postgres settings for tuning: log_statement and log_min_duration_statement. It has also already moved fleet-wide defaults down, including disabling connection logging (log_connections) across most plans.

    Read that last part carefully. Supabase reduced default logging across the fleet before introducing the meter. That is a platform telling you, in advance, that the default was generating more volume than anybody needed.

    Logs, before and after 25 September 2026

    QuestionBefore 25 September 2026Now
    Volume of logs your services produceNot a billable dimensionMetered as log ingest
    Scanning logs to debug an incidentNot separately accounted forMetered as log query, on a quota that scales with ingest
    What happens if you exceed the limitNothingNothing yet - metered but not enforced until the grace period ends in early 2027
    Where the number appearsNot surfaced as a costProject usage dashboard, labelled "Upcoming" during the grace period
    Who chose the logging levelInherited defaultsInherited defaults, unless someone changes log_statement and log_min_duration_statement

    The calendar is the real story

    The logs meter is not an isolated event. It is one of several things that landed this month on somebody else's schedule rather than yours.

    The same Supabase changelog, on the same date, carries a PostgreSQL 15.19 / 17.11 minor release flagged as a breaking change, bundling five upstream security cycles and closing 44 CVEs, with four changes that may require action before upgrading. Next.js listed a September security release for 30 September, covering 16.3.7 and 15.5.27, addressing nine vulnerabilities - one critical, two high, five medium and one low.

    None of these arrive through your build tool. No prompt surfaces them. They arrive on a changelog page that somebody has to be reading.

    What "production-ready" usually leaves out

    The category of fixed-price build offers - ours included - advertises some version of production-ready, publishable, or launch-ready. Read closely, that claim is nearly always about a moment: on handover day the app deploys, auth works, payments clear, the pages render.

    That is a real and useful claim. It is also a claim about a day, not about a year. The things it rarely covers are the ones that arrive afterwards on a platform's calendar: who reads the changelog in month three, who owns a database setting that becomes a billed line in 2027, who presses the upgrade button, and who ran the detection queries first.

    When you are comparing offers, the useful question is not whether a shop says production-ready. Almost all of them do. It is what happens in month four.

    Four things to do this week

    1. Open your Supabase project's usage dashboard and find the log ingest and log query meters, shown with an "Upcoming" label during the grace period. Write today's numbers down. You now have a baseline.
    2. Look at log_statement and log_min_duration_statement on your database. If nobody set them deliberately, then nobody chose your ingest volume.
    3. Confirm whether connection logging is on for your project. Supabase disabled log_connections across most plans - confirm it for yours rather than assuming.
    4. Put one calendar entry in early 2027, when the grace period ends, and re-read your numbers then. That single entry is the difference between a meter you watched and a bill you discovered.

    Getting these four answered

    IdeaFlo has completed 80 jobs on Upwork, turning Lovable and Vite prototypes into applications that hold up in production. If you want the four checks above run against your project, alongside the rest of the pre-launch inventory, the Strategy Audit is $499 one-time and is credited toward any build. If the app needs to become a real application first, the Publishable MVP is $4,999 one-time and ships ready to publish, and the Founding Client package is $2,999 one-time with a 14-day build guarantee and a full refund if the deadline is missed. Both carry the revenue guarantee: first paying customers within 90 days of launch, or a full refund.

    Book a Call

    Sources

    Related posts