AgentXray vs PureBids (2026): Ads-API Optimizer vs Settlement-Verified Profit
Last updated: 2026-08-21
TL;DR
PureBids is the closest tool to us we have found: Amazon EU focus, an MCP connector so your own Claude, ChatGPT or Gemini can read your campaigns, simulation mode on by default, and a published methodology. €99/mo flat, no commission. The difference is the source of the money: PureBids connects to the Amazon Advertising API, so its profit figure is ad data plus hand-entered COGS plus assumed fees. AgentXray reads the settlement — the fees Amazon actually deducted. On our own account the referral rate moved from 5.63% to 12.61% inside one month; a tool working from an assumed category percentage cannot see that, and every 'profit' it showed for those weeks was wrong by the difference.
PureBids (ANGAHE creating moments LLC, Wyoming) sells Amazon PPC optimisation built for European sellers, in Spanish, across ES, DE, FR, IT, UK, NL, BE, PL and SE. It is genuinely good at things we do not do: dayparting by day of week, placement modifiers, a 60-day negative scanner, campaign creation from an ASIN, an organic rank tracker, and a Monday morning email that tells you what to pause and what to scale. It publishes its decision rules — 14-day window, minimum 20 clicks, 7-day cooldown, 48-hour attribution buffer — which most tools do not. We think that is the right instinct and we have copied it. AgentXray (Avanta Global EOOD, Sofia) is built by an active Amazon seller and answers a narrower question: what did a unit actually earn, measured against the money Amazon deposited. This page is written from their own published pages, read on 2026-08-21, and says where they are ahead of us as plainly as where we are ahead of them.
The tools compared
AgentXray
From $29/mo — Flat; tiers separate order volume, not features. No commission on ad spend.
Measurement layer for Amazon sellers: settlement-verified profit, break-even, and per-arm net contribution.
Strengths
- Profit from settlement line items — the fees Amazon actually deducted
- Detects a referral-rate change inside the window instead of averaging over it
- Every proposed change priced against that seller's own break-even CPC
- Applied changes judged 14 days later on net contribution, and allowed to say 'inconclusive'
- One-click Amazon authorisation — we never ask for your API credentials
Best for: Sellers who want the number their bank statement agrees with, and every ad decision priced against it.
PureBids
€99/mo — €899/yr (≈€74.92/mo). 7-day trial, no card until day 8. No commission on ad spend.
Amazon EU PPC optimisation with AI diagnosis, weekly report and an MCP connector. Built for Spanish-speaking sellers.
Strengths
- Wider PPC toolset: dayparting, placement optimisation, rank tracker, campaign wizard
- Published decision methodology (14d window, 20 clicks, 7d cooldown, 48h buffer)
- Monday email with prioritised actions — no dashboard visit needed
- Spanish interface, documentation and support end to end
- MCP connector with six ready-made skills, read-only by default
Best for: Spanish-speaking EU sellers who want an opinionated PPC manager with a broad feature set.
Feature comparison
Where the money number comes from
| Capability | AgentXray | PureBids |
|---|---|---|
| Fee source | Settlement line items | Advertising API + entered COGS |
| Referral-rate change inside a window | Flagged (referral_stable=false) | Not documented |
| Fee-tax (GST/VAT charged on fees) | Measured and corrected | Not documented |
| Judging metric for an applied change | Net contribution after 14 days | ACoS against a target |
| Verdict may be 'inconclusive' | Yes, by design | Not documented |
| Settlement history depth | 89 days (Amazon's own cap) | Not applicable |
Getting connected
| Aspect | AgentXray | PureBids |
|---|---|---|
| Amazon Ads authorisation | OAuth — one click | You supply Client ID, Secret and Refresh Token |
| Needs an Amazon developer account | No | Yes |
| Seller Central data | Connected (settlement, orders, inventory) | Not part of onboarding |
| Free trial | Free tier, no card | 7 days, no card until day 8 |
Acting on it
| Capability | AgentXray | PureBids |
|---|---|---|
| Changes reach Amazon without a human | Never — approval is a browser session | Optional full automation |
| Assistant can apply its own suggestion | No — the token cannot authenticate approval | Simulation mode by default |
| Dayparting | No | Yes, per day of week |
| Placement optimisation | No | Yes |
| Organic rank tracker | No | Yes |
| Campaign creation wizard | No | Yes, from an ASIN |
Pricing model
| Aspect | AgentXray | PureBids |
|---|---|---|
| Headline | From $29/mo | €99/mo |
| Commission on ad spend | None | None |
| Feature gating by tier | None — volume only | None — single plan |
| AI questions metered | No — you connect your own assistant | Included assistant |
Verdict by use case
- Breadth of PPC tooling: PureBids
- Spanish-language product and support: PureBids
- Hands-off automated bidding: PureBids
- Knowing the fees Amazon really charged: AgentXray
- Catching a fee rate that moved mid-window: AgentXray
- Connecting without handing over API credentials: AgentXray
- Proving a change worked, or admitting it is unclear: AgentXray
Where we do not win: PureBids has more PPC features than we do, it is a complete Spanish-language product where we are not, and if you want the tool to drive the bids itself it will. Those are real advantages and a seller who needs them should buy it. Where we do win is narrow and, we think, load-bearing: their profit figure is assembled from ad data, entered costs and assumed fees, and ours is read from what Amazon deducted. The two agree until a fee moves — and fees move. Running both is not absurd: let them optimise, let us tell you whether the optimisation paid.
Frequently asked questions
- Both of you connect to Claude over MCP. What is different?
- What the assistant can see. Theirs reads advertising data; ours reads advertising data alongside the settlement, so when you ask whether a campaign made money the answer is priced against the fees actually deducted from that account. Ours also carries a tool whose only job is to list what it cannot tell you, so a gap is never returned as a zero.
- Is PureBids' 90-day methodology better than yours?
- It is published, which most are not, and we copied that instinct. On the rules themselves we differ in one place that matters: they judge an applied change against an ACoS target, and we judge it on net contribution after 14 days. ACoS improves when you cut a profitable arm, so a tool optimising toward it will occasionally report a win for a change that cost the seller money.
- Why do you not offer 24 months of history?
- Because Amazon does not. Measured on 2026-08-21 against a live account: asking the Reports API for settlement reports older than 90 days returns 400 InvalidInput — 'RequestedFromDate is more than 90 days old' — while 89 days returned 14 reports. Any deeper history therefore comes from something other than settlement, and we would rather have 89 days of what Amazon charged than two years of what it estimated.
- Why will you not take my Amazon Ads API credentials?
- Because holding another company's Client Secret and Refresh Token makes us responsible for a key we did not issue and cannot rotate. Amazon's own answer to this is OAuth, and it is the path we use.
Research sources
- PureBids home, pricing and comparison pages (purebids.app) — accessed 2026-08-21
- AgentXray settlement-measured economics — internal, 2026-08
- Amazon SP-API Reports settlement retention limit — measured live, 2026-08-21