Media buying is the operational work of turning a campaign plan into paid advertising placements, agreeing how they will be delivered, and checking whether the agreed outcome is achieved. A team promoting a new service might negotiate a newsletter placement directly with a publisher or use a buying platform to access video inventory. In either case, it must approve the spend, supply the right creative, and verify delivery.
Start with the campaign objective, audience, and budget. Then choose a channel and inventory, decide how to buy it, approve the terms, and define how results will be measured. A small team may have one person doing all of this, but it should still record who owns strategy, financial approval, launch, privacy review, and reporting.
This guide treats media buying as an accountable operating process rather than a static list of platforms. It separates buying method from media channel, explains the brief and approval workflow, and shows how to design bounded AI and reporting controls without claiming unsupported integrations or automated outcomes.
What is media buying?
Media buying covers securing advertising inventory or placements, agreeing commercial and delivery terms, coordinating creative, and monitoring delivery against campaign goals. It is related to media planning and optimization, but each describes a different responsibility:
- Media planning sets the objective, audience, channel strategy, and budget allocation.
- Media buying negotiates or activates placements and records the accepted terms.
- Optimization adjusts a live campaign within approved budget, audience, and brand-safety constraints.
Keep those responsibilities visible even when one person performs them. A recorded budget owner, campaign operator, and reporting owner make it easier to identify who can commit funds, change a live campaign, or resolve a measurement discrepancy.
Separate buying methods from media channels
A channel is where an ad appears, such as display, video, connected TV (CTV), audio, social, search, radio, or out-of-home. A buying method describes how the advertiser obtains the inventory. Digital does not automatically mean programmatic, and a single channel can offer more than one buying route.
A direct deal is negotiated with a publisher or seller. It can give the buyer clearer placement control and agreed commercial terms, but requires coordination with the seller. Programmatic buying uses software to select and purchase inventory. A demand-side platform (DSP) helps advertisers manage buying; a supply-side platform (SSP) helps publishers offer inventory; an ad exchange connects buyers and sellers. An ad network can aggregate inventory and sell it to advertisers.
Within programmatic buying, the transaction arrangement matters. An open auction offers inventory through bidding; a private marketplace restricts access to selected buyers; programmatic guaranteed agrees on specified inventory and delivery terms through programmatic systems. These are buying arrangements, not channels. Google describes Display & Video 360 (DV360) as supporting reservation, programmatic, and programmatic-guaranteed campaigns across multiple media types in its DV360 product overview.
| Decision | Direct deal | Programmatic access |
|---|---|---|
| Control | Negotiate with a named seller and placement. | Choose auction or deal access and configure buying rules. |
| Operations | Manage proposals, terms, creative deadlines, and delivery with the seller. | Configure the platform, inventory, audience or context, and measurement. |
| Compare on | Placement certainty, total fees, reporting, and shortfall remedies. | Inventory quality, fees, access, controls, and comparable reporting. |
Neither route guarantees lower total cost or better results. Compare the full cost, access to the intended inventory, operating capacity, and measurement fit.
Choose the channel for the audience and objective; choose the buying method for the required control, inventory access, cost structure, and operating capacity.
Build the buying brief before contacting sellers or opening a DSP
A usable brief lets sellers respond on comparable terms and gives the buyer a checklist for platform setup. Record the objective, audience definition, budget and currency, flight dates, geography, channels, creative formats, measurement goals, brand-safety requirements, and any reach or frequency constraints. Mark each item as a must-have or a preference.
Name the people responsible before a proposal arrives: the brief owner, budget approver, buyer, creative approver, privacy reviewer when audience data or tracking is involved, and reporting owner. Keep the approved brief and accepted terms in the team’s designated campaign or work-management system. No universal campaign-management integration or shared data schema has been verified across buying platforms.
When reviewing a request for proposal (RFP), compare offers against the same requirements. An insertion order (IO) is a commercial and delivery record, not an informal recap. Check the dates, placements, format, billing terms, creative deadlines, reporting definitions, and any remedy for a delivery shortfall, such as a makegood.
Run a controlled buying and launch process
Move from an approved plan to a live campaign through named handoffs. For each stage, save the output in the campaign record so the next owner can verify what was approved.
Use deterministic preflight rules for binary checks: reject a record if dates are missing or reversed, currency is absent, creative specifications are incomplete, or the landing-page URL is not approved. Keep changes to spend, audience lists, IO acceptance, and live campaign status behind explicit human approval.
Choose a platform using current access and fit
Shortlist platforms against the campaign rather than a generic top-platform list. Check channel and inventory coverage, audience availability, buying arrangements, account eligibility, managed-service requirements, fees, reporting needs, privacy constraints, and operational support. Verify the product name and access conditions for the target market and account before making a recommendation.
- DV360: Google describes support for reservation, programmatic, and programmatic-guaranteed buying across multiple media types. Its product overview does not establish universal eligibility, pricing, minimum spend, or suitability for every advertiser.
- Amazon Ads Agent: Amazon states that Amazon DSP capabilities are now available within Amazon Ads Agent, with DVA+ advanced settings for programmatic control. Amazon describes display, video, audio, and streaming-TV buying across Amazon-owned and third-party inventory, but availability varies by campaign, market, account, and service model.
- Microsoft Invest: Do not treat it as a current standalone DSP option. Microsoft ended media buying through Microsoft Invest on February 28, 2026. Microsoft and partner inventory may still be available through third-party DSPs, according to Microsoft’s discontinuation announcement.
A product name in a recent roundup does not prove it is still standalone, available to your account, or sold on unchanged terms. Record the date checked, official source, market, eligibility, and campaign-relevant feature for each shortlisted vendor. Treat historical pricing, minimums, and service models as unconfirmed until checked directly.
Measure delivery without blending unlike numbers
Publisher, DSP, analytics, and CRM reports can use different definitions and attribution windows. Keep vendor-reported conversions separate from analytics or CRM conversions unless a documented reconciliation rule explains the difference. Before comparing reach, frequency, impressions, spend, clicks, or conversions, align the reporting grain, date boundaries, timezone, currency, attribution window, metric definition, and reporting status.
For a practical reporting design, define one normalized observation as one source row, one line item or campaign grain, one reporting window, and one metric. Keep the original export or report snapshot separately from normalized observations and campaign-level aggregates. Preserve the source platform, account, campaign and line-item identifiers, report name, report window, timezone, currency, attribution window, metric name and value, retrieval time, source-row hash, and transformation version.
For normalized observations, use a database uniqueness constraint and an atomic upsert on the chosen observation key, such as source platform, account, campaign or line item, reporting window, metric name, attribution window, currency, and metric-definition version. Include the campaign identifier when the source can contain multiple campaigns for one line item or report window. If separate retrievals must be retained, store each raw snapshot with its own retrieval identifier or content hash, then link normalized observations to the snapshot. A read-then-create check alone is not safe against concurrent jobs.
- Confirm the account, campaign, line-item identifier, and declared row grain.
- Align reporting dates, timezone, and aggregation period.
- Check currency and record any conversion method.
- Match attribution windows, metric definitions, and preliminary or finalized status.
- Preserve the source report, retrieval time, and source-row identity; quarantine missing dimensions.
- Enforce a unique observation key and use an atomic upsert for repeat or concurrent imports.
The fields below are an illustrative design, not a vendor export format or verified API schema. This example represents one normalized line-item observation for one metric and reporting window; raw snapshots belong in a separate record.
{
"source_platform": "example_dsp",
"source_account_id": "account-42",
"source_report_name": "line_item_daily",
"campaign_id": "campaign-17",
"line_item_id": "line-03",
"reporting_window_start": "2026-10-12",
"reporting_window_end": "2026-10-12",
"timezone": "America/New_York",
"currency": "USD",
"attribution_window": "source setting",
"metric_name": "impressions",
"metric_value": 12500,
"metric_definition_version": "v1",
"source_row_hash": "hash-of-source-row",
"retrieved_at": "2026-10-13T15:30:00Z",
"transformation_version": "v1"
}Before building an automated reporting pipeline, confirm that the specific platform supports the required export or API and that the source supplies the fields. The verified product pages reviewed for this guide do not establish a universal export schema or CRM connection.
Apply privacy and measurement controls before targeting
Apple’s App Tracking Transparency (ATT) framework applies when an app collects and shares data with other companies for tracking across apps and websites. An app must request authorization and check the resulting status before enabling tracking-dependent behavior. Apple’s documentation explains the ATT authorization framework and its broader privacy and data-use requirements.
For an app-based campaign, document the proposed tracking purpose, SDK inventory, data-sharing recipients, authorization status, and measurement fallback. Check whether the activity meets Apple’s definition of tracking, verify the app’s implementation, and allow only behavior supported by the authorization status. Do not use hashed email or device fingerprinting as a substitute for required permission. ATT defines an app-level policy and technical requirement; it does not establish a universal percentage change in revenue, targeting, or conversions.
Use AI for bounded operations, not spend approval
AI can help extract proposal terms into a comparison record, classify campaign notes, or flag unusual contract wording. It should return candidate values with source passages, not decide whether an offer is acceptable. Deterministic parsing and allowed-value checks are better for dates, currency codes, required fields, and approved formats. A buyer verifies commercial terms, and the budget owner approves the commitment.
| Trigger | AI job | Validation and action | Fallback |
|---|---|---|---|
| A publisher proposal arrives | Extract vendor, placement, dates, cost, currency, delivery metric, and source passages. | Validate the structured fields and compare every value with the source document. Send a verified comparison record to the buyer. | Route missing or conflicting terms to the buyer and seller; do not infer a value. |
| A report import contains anomalies | Summarize flagged discrepancies after deterministic checks. | Require valid identifiers, dates, currency, metric definition, and row grain before an upsert into the reporting store. | Quarantine the row and retain the raw snapshot for analyst review. |
| An app campaign proposes tracking | Optionally summarize the data-flow inventory. | Privacy or engineering owner checks Apple’s tracking definition and authorization status. AI cannot authorize the behavior. | Block tracking-dependent activity and use an approved measurement fallback where available. |
Require valid structured output, check date and currency formats, and compare each extracted value with the source document. Reject malformed or incomplete output. Do not let AI accept an IO, commit funds, change material live settings, upload customer data, or determine privacy compliance. Evaluate any operational use by review time, missing-field rate, exception rate, reconciliation differences, and on-time launches, not by assumed campaign-performance gains.
Hypothetical example: a buyer receives a publisher proposal and asks an AI tool to extract dates, cost, placement, and metric definition. The result enters a proposal-comparison record in the team’s chosen work-management system only after schema validation and source review. If dates conflict or currency is absent, the buyer resolves the exception with the seller; the system does not silently fill the gap.
{
"vendor_name": "Example Publisher",
"placement": "Newsletter sponsorship",
"start_date": "2026-11-02",
"end_date": "2026-11-08",
"proposed_cost": 4200,
"currency": "USD",
"delivery_metric": "impressions",
"metric_definition": "publisher-defined delivered impressions",
"source_excerpt_by_field": {
"proposed_cost": "Illustrative source passage",
"delivery_metric": "Illustrative source passage"
},
"needs_review": true,
"exception_reason": "Illustrative output; buyer must verify against proposal"
}For teams designing supervised workflows, see AI agent consulting.
Make the buying process accountable
A strong media-buying operation connects the campaign objective to an approved placement, preserves the terms behind the commitment, and measures delivery on comparable definitions. Keep the approved brief, proposal comparison, IO, creative approval, launch record, raw report snapshot, normalized observations, and report provenance together, with named owners for each decision.
Recheck product access and terms when selecting vendors because platform names, inventory, features, and service models change. For support designing operational processes around campaign work, explore ConsultEvo’s services.
Research note: Vendor capabilities and policy details in this guide were checked against the linked official sources on October 13, 2026. Recommendations for approval gates, reporting fields, uniqueness constraints, and AI review are operating-design guidance, not vendor-provided templates or confirmed integrations.