Why a run failed and how to fix it

Every failed run shows a short message explaining what happened. This article covers what each one means and what to do next — most of the time, starting a new run is all it takes.

A temporary problem on our side

You'll see a message like "We hit a temporary problem on our side" when something went wrong in our systems, not with your setup. Our team is notified automatically. If a Try again option is shown, it's safe to use right away; otherwise, wait a few minutes and start a new run.

We couldn't start this run

A run can fail to start for a reason tied to your account or setup without more detail to share. Check anything you changed recently — your API key, catalog or retailers — then start a new run. Contact support if it keeps happening.

The run never started

This run never started, so we stopped waiting on it rather than leaving it stuck. You can start a new run right away.

The run stalled while refining results

This run stalled while we were refining its results, so we stopped it. You can start a new run.

Results were incomplete for the quality check

This run's results came back incomplete, so the quality check couldn't run on them. You can start a new run.

Results couldn't be queued for a quality check

We couldn't queue this run's results for their quality check. You can start a new run.

Quality check unavailable for this run

The quality check wasn't available for this run, so we stopped it. You can start a new run.

We couldn't read the run's results

We couldn't read this run's results after it finished. You can start a new run.

Stopped by our support team

This run was stopped by our support team because it was stuck. You can start a new run.

Your saved API key couldn't be used

We couldn't use the API key saved for this workspace. Check it in Settings → API key, then start a new run.

The price-check service wasn't available

The service that runs price checks wasn't available for this run. You can start a new run.

The run hit its cost limit

We stopped this run automatically because it reached the cost limit for a single run. Start a new run with fewer products.

The run cost more than expected

We stopped this run automatically because its cost crossed this workspace's cost-stop multiple of the estimate. You can adjust the multiple in Settings → Billing, or start a new run.

The run hit a service-side budget

We stopped this run because it hit a cost budget on the agent service itself, or because part of it used a model we don't have pricing for. You can start a new run.

We tried several times and it kept failing

We tried this run several times and it kept running into problems, so we stopped it. You can start a new run.

Too much to work through at once

This run had too much to work through at once. Try again with fewer products.

Declined for a billing reason

This run was declined for a billing reason before it could finish. Check the billing details in Settings → Billing and any saved API key, then contact support if everything looks correct.

The service was too busy

The service was too busy to finish this run. Try again in a few minutes.

Too many requests at once

Too many requests were made at once, so this run was stopped. Try again in a few minutes.

Something went wrong while processing

Something went wrong while this run was being processed, so we stopped it. You can start a new run.

A needed service wasn't reachable

A service this run needed wasn't reachable, so we couldn't finish it. You can start a new run.

An unexpected problem occurred

We hit a problem while this run was working, so we stopped it. You can start a new run.

We couldn't confirm your own key was used

We couldn't confirm your own API key was used for this run, so we stopped it. Check the key in Settings → API key, then start a new run.

The models this run needed weren't available

The models this run needed weren't available right now, so we stopped it. You can start a new run.

None of your saved keys worked for a step

None of your saved API keys work for a step this run needed. Check your keys in Settings → API key, then start a new run.

Go further

• Use your own AI provider key — connect, check or remove a key.
• Run a check & read a recommendation — trigger a new run.
• Billing & credits basics — how research units and cost limits relate.