One platform for advertising intelligence, seller-business context and controlled operations.

Every important figure carries its source label, the same six labels in the product and on this website: ADS, ADS HOURLY, SELLER, FINANCE, ASSUMPTION and DERIVED. A derived figure names the sources it was calculated from.

No hidden source.

When a data source is not authorized or not available, Terra PPC marks the result instead of filling the gap with a zero. A missing input never makes a number look better than it is: without finance data profitability is partial, without seller revenue TACoS is unavailable, and without inventory data the Inventory Guard is not evaluated.

Data from different product domains stays source-aware, but is normalized into one scoped decision context: the same names, units and period, for one organization, account, advertising profile, seller connection, marketplace and date range at a time.

Terra PPC is designed to organize authorized Amazon advertising data by account, profile and campaign structure, so performance can be read at the level decisions are made.

• Campaign, ad group, keyword, target, search-term and placement performance
• Ad Spend, Ad-attributed Sales, ACoS, ROAS, CPC, click-through and conversion trends
• Hour-of-day performance and budget pacing
• Target ACoS and Modeled Break-even ACoS comparisons where product economics are configured

Advertising results depend on the business behind them. Where authorized through the Selling Partner API, Terra PPC is designed to use selected seller-business context to explain performance, model profitability and protect advertising decisions.

• Seller sales and orders
• Sales & Traffic: sessions, page views, conversion and featured offer rate
• Finance & settlements: fees, refunds, adjustments, reimbursements
• FBA inventory states and days of cover
• Listing and catalog context, with applicable fee estimates for planning
• Marketplace participation
• Brand and retail analytics, where available

Non-PII context only: Terra PPC's analytics are designed without buyer names, email addresses, phone numbers or shipping addresses. Availability depends on approved roles, authorization and marketplace support.

Some inputs only the customer knows. COGS, freight, duty, packaging, other variable costs and the desired margin are configured in Terra PPC and labelled as assumptions — they never come from Amazon.

The economics engine turns customer-configured costs into the thresholds advertising decisions are measured against. Unit costs are versioned by effective date, so a supplier change moves break-even from the day it applies — and history keeps the cost that was true at the time.

The decision context assembles the eligible, scoped data a rule, a user or an AI-assisted request needs for the current operation — and nothing more. It answers one question: what information is available?

Advertising metrics

Spend, sales, ACoS, ROAS, CPC and conversion for the entity and period

Seller context

Sales and traffic, where authorized

Economics

COGS, modeled margin and break-even thresholds

Inventory

Days of cover for the advertised products

Listings

Listing status of the advertised products

Finance

Fees and refunds for historical profit

Scope

Organization, account, profile, seller connection, marketplace, period

Coverage

Which of these are available — and which are not

Terra PPC evaluates scoped performance, business context, economics, rules, strategies, and data readiness before producing a recommendation or eligible operational proposal. The evaluation is deterministic: the same inputs give the same result, and every result keeps its reason.

Rules are specific, deterministic conditions and actions. Strategies group coordinated rules with policy defaults — how cautious to be, which mode to start in — and are assigned per campaign. Publishing a strategy records a version, so an earlier one can be restored without rewriting history.

Automation turns analysis into action without hiding the logic. Each rule has conditions, an action, guardrails and a mode. Where seller context and customer economics are available, business-aware guards check inventory, listing status, cost changes and modeled break-even before a scale action continues.

• Conditions on metrics, minimum data thresholds and lookback windows
• Bid, budget, status, negative-targeting, search-term, placement and hourly workflows
• Guardrails: bid floors and ceilings, change caps, cooldowns and action limits
• Business-aware guards: Inventory, Listing, Cost Increase, Break-even and Financial Margin
• Manual, Review or Auto mode, chosen per campaign

Every automation cycle follows one canonical order of 18 steps. Scope and data readiness come first, safety next, then the proposals in a fixed stage order, the conflict resolver and the global guardrails — and only then the campaign's mode, the action and the audit record. Each candidate ends at exactly one step, with its reason.

Several eligible actions can compete for the same entity in one cycle, and a rule can meet a guard. Terra PPC resolves both deterministically: one proposal per entity survives, a guard can hold a scale action, and every suppressed action is recorded with its reason.

Automation is never unlimited. Limits apply per change, per campaign and per cycle, and the whole account can be stopped at once.

Global guardrails

Bid floors and ceilings, change caps and action limits for every automatic change.

Circuit breakers

Automatic execution stops when a cycle would exceed action, spend, budget, conflict or failure limits.

Data readiness

Stale or unavailable data stops a proposal instead of being read as zero.

Cooldowns

An entity changed recently waits before it can change again.

Action caps

A limit on automatic actions per campaign and per cycle.

Spend exposure limits

The share of spend and budget one cycle may move.

Undo / revert

Applied changes keep their previous value; the last automation cycle can be reverted.

The customer decides how much control to delegate, per campaign. The checks before the mode never change.

Manual

Direct control. Automation can be previewed, but people make the changes.

Review

Recommendations are created and wait for approval before anything is applied.

Auto

Eligible actions execute after policy and guardrail evaluation, and are recorded.

Every mode stays subject to scope, permissions, guardrails and the audit log.

Terra PPC AI/MCP lets compatible AI clients ask questions and prepare work in natural language. It is an interaction layer on top of the same controls — not a shortcut around them.

• Authentication and authorization on every request
• Scoped decision context only — no unrestricted database access
• Proposals pass the same decision engine, execution planner and guardrails
• Recommendations and actions still follow Manual, Review or Auto

ConceptualConceptual workflow. The answer uses synthetic values from the demo account.

Terra PPC is designed to record important recommendations, automation actions and operational changes with the fields needed to explain them later: what changed, the before and after values, the rule and rationale, the source, the mode, the guardrails, the time and whether the change can be reversed.

Agencies and multi-brand teams work across many accounts and marketplaces. Terra PPC is designed to keep every view, rule and action attached to one organization, authorized account, advertising profile, seller connection and marketplace at a time.

• Logical isolation between customer organizations
• Scope validation before a rule runs or an action is applied
• Marketplace-level reporting, each marketplace in its own currency

Advertising and seller-business connections are separate authorizations with separate scopes, each revocable on its own. Authorized access, least privilege, account isolation, data minimization, secure transport and traceability are part of the platform design. Data that is no longer needed for enabled functionality is designed to be deleted, and customers can request deletion of their account data.