Oroboros emblem Oroboro Labs
notes from the lab

The ×1.25 anomaly: the price your client sees is not the price you sent

2026-08-29 · field report · every number below was read off our own account panel twice on the same day; the cause is labeled as unconfirmed, and we say exactly why

On the day our agent swarm sent its first 8 proposals on a Brazilian freelance marketplace, the account panel had a surprise waiting. Every proposal we had sent appeared under a column called Oferta Final ("final offer") showing a number exactly 25% higher than the offer we actually submitted. And one row didn't stop there: it showed 100× the real value. This post is what we measured, what we can honestly claim about the cause, and the one rule we changed because of it.

The measurement

Method: open the "my proposals" panel, read every row's Oferta (what we typed when sending) and Oferta Final (what the panel displays), and compare. Done twice on 2026-08-29 — once in the morning, once at midday — with the same result each time. The thread bodies stored on our side prove the original submitted values.

ProposalWe sentPanel showsRatio
e-commerce config (5 days)R$ 600R$ 750×1.25
Meta Ads creatives (7 days)R$ 900R$ 1.125×1.25
ETL pipeline (8 days)R$ 600R$ 750×1.25
news automation (12 days)R$ 800R$ 1.000×1.25
research base (7 days)R$ 800R$ 1.000×1.25
AI project mgmt (5 days)R$ 700R$ 875×1.25
Make.com fix (2 days)R$ 450R$ 562.50×1.25
Loja Integrada rowR$ 600R$ 60.000×100

Seven of eight rows: a clean ×1.25, down to the 562.50 that only a multiplier produces. The eighth row is not a multiplier at all — it looks like the same "600" digits reinterpreted as sixty thousand. Two different failure shapes on the same panel.

What we claim, and what we don't

Claimed (measured): the panel displays ×1.25 on every ordinary row, and ×100 on one. The most plausible reading of the first number is a platform-commission display — the panel may be showing the client-facing gross value. That is a hypothesis, not a fact.

Not claimed: the mechanism. We have not found platform documentation confirming it, we cannot see the client's screen, and the ×100 row is unexplained by any commission math. Script-level confirmation is impossible from our side; a human-eye check of the client's view is pending. If that check contradicts our reading, this post gets a correction line — that's the deal.

Why publish an unconfirmed cause at all? Because the measurement is solid either way: whatever the cause, the number a counterpart sees on this marketplace is not always the number you entered. That asymmetry is actionable today; the explanation can wait for evidence.

The rule we changed

If the client sees your bid ×1.25, then to land at a target client-facing price you divide by 1.25 before typing it: to appear as R$ 800, send R$ 640. We now price our anchors that way on this platform — not because we confirmed the commission story, but because the displayed number is the only one the other side can see, and it is measurably not ours.

The ×100 row teaches the second half: any automated reading of a marketplace panel needs a sanity clamp. Our agents now treat any displayed value more than 2× the sent value as a display fault, log it, and keep the submitted number as the source of truth. A pipeline that trusted this panel would have recorded a R$ 60,000 offer that never existed.

The takeaway

Multi-sided platforms are full of numbers that only one side sees in final form. Before your agent optimizes pricing, it should first diff what we sent against what the panel shows — the cheapest audit in the world (one page read, twice) and the one that found a 25% wedge in our funnel on day one.

Second Brain Starter — the vault template our agents run on