@kapa.ai facebook marketing connector AdAccount-level jobs returned

Job Failed

from Facebook (~30-70 seconds after submission) The connector fell back to a

[Split:campaign]

strategy: a separate

AdReportRun

job was created for each campaign × each day why this is happening? it causes slowness

This behavior is coming from the Facebook Marketing connector’s job‑splitting / fallback logic for Insights async jobs, and it’s expected (though it can indeed cause slowness). From the available information: 1. When an AdAccount‑level async job fails, the connector intentionally “splits” the job. ◦ The connector first submits an async

AdReportRun

at the account level. ◦ If Facebook returns

Job Failed

(often with messages like “too many rows” or timeouts), the connector treats this as “the job is too big” and falls back to a more granular strategy. [Too many rows; Ad-level timeout] 1. The fallback strategy is to split by campaign (and often by day). ◦ In the “too many rows” / timeout scenarios, the connector creates smaller jobs, e.g.

[Split:campaign]

and even per‑day intervals, to satisfy Facebook’s requirement to “use async jobs or restrict amount of ad IDs or time ranges.” [Too many rows; Ad-level timeout] ◦ This is exactly what you’re seeing: one

AdReportRun

per

campaign × day

. It’s a defensive mechanism to avoid Facebook’s “too many rows” / timeout errors. 1. Why it causes slowness. ◦ Many small async jobs (campaign × day) mean many round‑trips to the API and a lot of polling, which is slower than a few large jobs. ◦ This pattern was introduced specifically to handle accounts with large volumes (e.g., many ads or detailed breakdowns) where account‑level jobs frequently fail. [Infinite attempts / too many rows; Too many rows] 1. What you can adjust (within documented options). ◦ You can tune

insights_job_timeout

and

insights_lookback_window

in the source configuration to influence how quickly the connector gives up on a large job and how wide the attribution/lookback window is. Smaller windows and shorter timeouts can reduce the size of each job and may reduce how aggressively it needs to split. [FBM config schema] The knowledge sources do not provide a configuration flag to disable this

[Split:campaign]

behavior; it is the connector’s built‑in strategy to cope with Facebook API limits and errors. 👍 0 button 👎 0 button ℹ️ Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.