PRICING

Priced by how hard the target is, not how many pages it has.

One billing unit — the credit — with a monthly pool and a published overage rate. What decides your number is the difficulty of the sites on your list, carried by a simple per-record multiplier. Start with a free sample from your own targets, no card.

Get a free sample See the plans

NOT SELF-SERVE YET Pricing is stated in credits because that is how metered billing will work. Today every plan begins with a scoping conversation; self-serve billing arrives with the API.

PLANS

Four ways in

Prices below are what you pay for the plan. What sits inside a plan is scoped with you first — because a target list is the only honest input to a quote.

  • Aura Sample

    Freeyour sources · 48 hours

    The honest way to start: real records from your own target, delivered in two days, yours to keep. No card.

    Volume
    Up to 100 records
    Targets
    Open & Rendered targets
    Cadence
    One run
    • One target, one extraction run
    • Up to 100 records in your requested schema
    • JSON or CSV — the real output, yours to keep
    • A short report: what was reachable, what was not
    • No card and no commitment
    Get a free sample

  • Aura Prism

    $99per month

    Continuous extraction from whatever targets you need, delivered in the shape you asked for.

    Volume
    25,000 credits / month
    Targets
    Open & Rendered targets
    Cadence
    Weekly or monthly runs
    • Any targets you like — maintained by us as they change
    • Your schema — your field names, units, and nesting
    • JSON, NDJSON, or CSV delivery
    • File drop or webhook delivery
    • Validation before every delivery

    Beyond the cap$4 per 1,000 credits

    Scope a Prism plan
  • Most common

    Aura Spectrum

    $399per month

    Many sources, normalised into one schema so cross-site comparison needs no cleanup.

    Volume
    150,000 credits / month
    Targets
    Open & Rendered targets
    Cadence
    Daily runs
    • Many sources under one normalised schema
    • Currency, unit, and date normalisation across sources
    • Deduplication and change detection between runs
    • File drop, webhook, or hosted endpoint delivery
    • Everything in Prism

    Beyond the cap$3 per 1,000 credits

    Scope a Spectrum plan

  • Aura Atlas

    From $1,500per month

    The defended tier. Hardened targets, real volume, and engineering attached to your account.

    Volume
    Scoped credit pool
    Targets
    Open, Rendered & Defended targets
    Cadence
    Any, including near-real-time
    • Akamai, DataDome, and PerimeterX-class targets
    • Warehouse table, queue, or hosted API delivery
    • Custom cadence and change-detection windows
    • Replay any past window without re-fetching
    • A named engineer and an agreed SLA
    • Everything in Spectrum

    Beyond the capQuoted against your scope

    Talk to engineering

BILLING One unit across the whole product: the credit. A credit is one delivered, schema-passing record at the Open tier — harder targets spend more per record. Blocks, retries, timeouts, and failed runs are our cost, never yours.

COMPARE

Every plan, line by line

The cards show the headline; this is the whole ladder. Difficulty tier still moves the final number — see below.

Sample Prism Spectrum Most common Atlas
Price Free $99 / mo $399 / mo From $1,500 / mo
Sources 1 target Any Any Any
Included credits / month 100 records (one run) 25,000 150,000 Scoped
Run cadence One run Weekly / monthly Daily Any · near-real-time
Open & Rendered targets
Defended targets (Akamai / DataDome / PerimeterX)
Your schema — field names, units, nesting Sample
JSON · NDJSON · CSV delivery JSON / CSV
File drop or webhook
Hosted API endpoint
Warehouse table or queue delivery
Cross-source normalisation (currency, unit, date)
Deduplication & change detection
Replay any past window without re-fetching
Validation before every delivery Sample
Named engineer & agreed SLA
Overage — $4 / 1k credits $3 / 1k credits Quoted

WHAT MOVES THE NUMBER

Difficulty, not volume, decides the price

The same 100,000 records can be a routine job or months of reverse-engineering work depending entirely on what is guarding the page. So a credit buys one Open-tier record; a harder target simply spends more credits per record. That multiplier is the whole price story — there is no separate build fee.

  • Open

    1 credit / record

    Server-rendered pages with no meaningful defenses. The data is in the markup; the work is normalising it.

  • Rendered

    3 credits / record

    JavaScript execution, infinite scroll, virtualised lists, per-device markup. The page has to run before it can be read.

  • Defended

    10 credits / record

    Akamai, DataDome, and PerimeterX-class protection: TLS and HTTP/2 fingerprinting, JS challenges, behavioural scoring. The multiplier carries the reverse-engineering work — available on Atlas.

What each tier actually means at the wire:

target difficulty select a tier

The data is already in the page — the work is tidying it up.

defended targets are quoted, never rate-carded

THE PROCESS

Three steps, no procurement theatre

  1. Tell us your targets

    Which sites, which fields, how often, and where the data needs to land. A few sentences and a couple of example URLs is enough — you do not need a spec document.

  2. We assess and sample

    An engineer places each target in a difficulty tier, tells you plainly what extraction involves, and delivers real records from your real sources — before any commitment.

  3. Pick a plan, or get a quote

    Open and Rendered targets fit the published plans. Defended targets are quoted against scope on Atlas. Either way you have seen real output before committing.

QUESTIONS

What people ask before scoping

What is a credit, exactly?

One credit is one delivered, schema-passing record from an Open target. Harder targets spend more credits per record — three on a Rendered target, ten on a Defended one — because the work behind them is greater. It is one unit for everything, so a mixed pipeline still bills against a single pool, and there is no separate build fee: the reverse-engineering cost lives in the multiplier.

What happens when I run out of credits for the month?

Records above your pool are billed at the plan’s published overage rate per 1,000 credits. We flag it on the run that crosses the line rather than at the end of the month, so the pool is never a surprise, and unused credits do not roll over.

Why are defended targets quoted instead of listed?

Because an honest per-1,000 rate needs someone to have looked at the target first. Two sites behind the same commercial bot-defense platform can differ by an order of magnitude in effort. We assess yours, then quote it.

Is this self-serve? Can I sign up and start?

Not yet, and we would rather say so here than after you have entered a card. Metered credit billing is how it will work — that is why pricing is already stated in credits — but today every plan starts with a short conversation and a free sample. Self-serve signup arrives with the API.

What if a target turns out to be impossible?

We tell you immediately, stop billing for that source, and either re-engineer access or remove it from scope. Quietly shipping degraded data is the one failure mode we will not accept.

Do you offer annual pricing?

Yes — 15% off on an annual commitment, or 20% if the year is paid upfront. Annual plans get the same monthly credit pool, billed once.

What counts as one record?

One structured object in your schema — a product, a review, a listing, a post. A page that yields fifty products is fifty records, not one; a page that yields nothing is zero. Credits are spent on what lands in your delivery, not on pages fetched.

Do failed or blocked runs cost me anything?

No. Blocks, retries, timeouts, and re-fetches are our cost of doing the job, and they never spend a credit. You are billed only on delivered, schema-passing records.

How quickly can you start, and when do I see real data?

A free sample against a single target usually turns around in a couple of days. For a full plan, scoping and the first validated sample come before you commit — you approve real records, not a promise. Exact turnaround depends on the target list and is agreed with you up front.

Can I change my schema or add sources later?

Yes. Field names, units, and nesting are yours to revise, and there is no per-source limit — add as many targets as you like; they simply draw from the same monthly credit pool. A schema change is a mapping update on our side, not a rebuild you pay for again.

Where does the data land, and who owns it?

You do. Delivery is file drop, webhook, hosted endpoint, or — on Atlas — a warehouse table or queue. We retain only what is needed to run and replay your pipeline, and we do not resell or reuse your extractions.

Can I cancel, and what happens to a running pipeline?

Monthly plans stop at the end of the paid month with no penalty; annual commitments run to term. When a pipeline stops we hand over the most recent delivery and, on request, a final full export.

Send us the hard one first.

Pick the target that defeated your team or your last vendor. We will tier it, tell you what extraction involves, and show you real records — a free sample from your own source — before you pay for anything.

Talk to our team