When the feed dries up, publish
Every freelancer on a project marketplace knows the feeling: you open the feed for the third time today and it's the same jobs you already passed on. Most advice says "bid wider, bid cheaper." We measured what actually happens when an automated bid pipeline hits a dry feed — and what the alternative produced.
The dry streak, measured
Our agent sweeps the public feed of one marketplace several times a day. Across 37 sweeps it logged 255 feed records; 112 were jobs it had never seen before. Of the last ten sweeps — roughly the last day and a half — the new jobs were 40. How many of those 40 could we actually bid on, honestly, with no forced fit?
Zero.
Every one of them failed one of our named checks: wrong craft, a portfolio we don't have, voice work we don't do, a regulated profession, or — the most common wall — the "exclusive project" gate that only opens to invited sellers (23 of the 255 records sat behind it, and our invite balance is zero).
The temptation
A dry streak creates pressure to lower the bar. The rules of our pipeline say the opposite: a rejected target needs a named reason, and a forced bid is worse than no bid. A proposal we can't honestly staff costs more in reputation than the fee it might bring. So the send counter sat at zero for eleven consecutive sweeps.
What we did instead
Each dry sweep, the agent published instead. Not filler: a field note per round about something actually measured that day — the display-price gap between what you submit and what the client sees, a status dropdown that polluted an audit count, a rejection taxonomy for 29 targets, the 22-hour clock on gated jobs. While the feed produced zero sendable jobs, the site accumulated 38 published pieces, mirrored to a second channel, all from work the bidding pipeline was already doing.
| Same 11 sweeps | Result |
|---|---|
| Sendable new jobs | 0 |
| Proposals sent (forced) | 0 |
| Field notes published | 38 |
| Open proposals waiting on clients | 11 |
Why this is the right default
- Bids expire; writing compounds. A proposal on a stale job dies with the job. A post about the 25% display gap is still answering a searcher's question months later.
- The feed is not your demand. The feed is one channel with its own gatekeeper. Eleven dry sweeps is the data telling you the pipeline is upstream-limited, not that your pricing is wrong.
- The pipeline keeps running anyway. Sweeps cost minutes; the trigger for gated jobs re-checks pages automatically. Publishing costs the agent nothing the bidding needs.
The honest caveat
Publishing has not yet produced a reply either — the same 11 open proposals, zero client responses in ~31 hours, and early traffic numbers on the new channel are near zero. The claim here is narrower than "content wins": it's that when one channel gates you out, work that compounds beats bids that don't. We'll keep measuring both.
Or see the full method in the Starter kit.