The questions that come up most when someone is trying to figure out which plan they actually need, what happens if they forget to cancel a trial, and where the boundaries are.
Picking a plan, the trial mechanics, what happens if you forget to cancel, and how upgrades and downgrades behave.
Starter and Research do. Seven days, full access from day one. You put a card down at checkout, but nothing gets charged for the first week. On day 7, Stripe charges the card unless you've canceled — and Stripe emails you 3 days before the charge so it isn't a surprise. Cancel anytime from /account/billing during the trial and you walk away clean. Pro and Enterprise don't have a trial.
Email us through the contact link on About within a reasonable window. We refund unintended trial conversions case by case, and the bar is low when you reach out fast and haven't really used the paid plan since the charge. We'll refund (full or partial) and cancel the subscription so it doesn't roll again next month.
Yes. Free → Starter → Research → Pro all keep your saved work. Nothing gets deleted on the way up, and nothing gets deleted on the way down either.
Nothing gets deleted automatically. Your saved dashboards all stay readable. If your count is over the new plan's limit (say, 35 boards on Pro and you drop to Free's 1-board limit), the dashboard page shows a banner naming the overage and you pick which boards to delete. The trash icon on every card stays visible while you're over the limit instead of only on hover, so you don't have to discover it. You can't create new dashboards until you're back at or below the limit. Inside any board, widgets past the lower tier's per-board cap also stay readable but become non-editable. None of your underlying data, watchlists, or exports get touched.
Same shape as dashboards: nothing gets auto-deleted. If you're over the new plan's watchlist cap (Starter 25, Research 150, Pro 750), the watchlist panel shows an overage banner and you remove items from individual indicator pages until you're back under the limit. Same pattern for calendar reminders (Free 3, Starter 25, Research 100, Pro 400) — the calendar page shows the banner and you clear reminders on individual events. Adds are blocked until you're back under the cap; existing items stay readable in the meantime.
Custom API limits, private backfills, SLA, SSO, audit logs, redistribution rights, licensed consensus feeds, and dataset-level legal review. The lane for institutions where one of those is a hard requirement instead of a nice-to-have.
Where Research is enough and where Pro starts to matter — model tabs, exports, the hosted runner, the labs that aren't out yet.
Free: model overview tab only. Starter: overview, graph, proof, and compare. Research and Pro: the same four tabs as Starter, plus Research unlocks the journal-grade export set.
Simple is point forecasts and guided workflows — that's what Starter gets. Research opens up empirical forecasting, forecast combination, diagnostics, model-expected release values, reproducibility metadata, and the published run limits that actually let you plan around a real envelope.
No. Research handles single-graph academic exports — journal-clean ZIP, SVG, CSV, LaTeX figure, source attribution, reproducibility manifests. Pro is for once you stop doing one graph at a time — hosted notebook execution, larger uploads, higher API and dashboard limits, and the published runner envelope.
No. Research lets you export notebooks and run them locally. The hosted runner — where we execute the notebook for you, within published runner limits — is on Pro.
Not yet. DSGE Lab and Agent-Based Lab are both shipping later, and both are Pro-only when they do.
How official-source series get mapped, what country coverage looks like, and where the export rules get drawn.
A series we've pulled from an official provider, tagged with the source ID and country, formatted to slot into charts, and stamped with the trust metadata you see on the indicator page. Provider data with the housekeeping done.
No. U.S. macro stays strong, but Research and Pro open up the global discovery catalog, country pages, and provider pages so you can work with official sources outside the U.S.
No. Every export bundle gets checked against the provider's rights metadata before it leaves. Some datasets are blocked outright; others need extra review even on paid plans. We don't let you redistribute data the source didn't license for redistribution.
What Macro Lab actually lets you do, and how Marco's monthly token budget plays out before it pauses.
Every limit on Research and Pro is published — run cap, upload size, storage, retention, runner sandbox. Not soft suggestions. They're the operating envelope, and they're stable enough that teams can plan around them.
You get a warning at 70%, another at 90%, and at 100% Marco stops responding until the next monthly reset. No surprise overage charges — the budget is capped, not metered up.
The pricing comparison shows the included monthly input and output budget per plan. Under the hood we meter across all model activity — cache reads, cache writes, eventually web-search units — so the budget you signed up for stays the budget even as Marco gains new tools.