Self Serve Banner Ads

Official boundaries for uploaded display assets, flexible formats and online claims

Google Ads publishes uploaded display-ad specifications for its network, including supported file types, dimensions, animation rules and eligibility conditions. Those requirements are Google-specific and do not define every self-serve platform or guarantee inventory and results. IAB's New Ad Portfolio provides flexible-size creative guidance intended for varied screens; each receiving seller still determines what it accepts. FTC online advertising guidance states that claims must be truthful, not misleading and supported when needed, with clear and conspicuous disclosures where required in the United States. These sources frame three separate gates: technical asset acceptance, adaptable creative design and claim responsibility. Verify the live platform, market, account, contract and destination. Do not use upload approval as evidence of accessibility, legal compliance, viewability, traffic quality or conversion.

• Google Ads uploaded display specifications - Google-specific formats, dimensions and eligibility, not universal self-serve rules
• IAB New Ad Portfolio creative guidelines - industry flexible-size guidance whose acceptance must be checked with each seller
• FTC online advertising and marketing - United States claim, substantiation and disclosure guidance

Assign account ownership

Name the legal entity, account administrator, billing owner, campaign operator and emergency contact. Use individual credentials, approved authentication and least-privilege roles.

Do not share an administrator login for convenience. Test access revocation and recovery before funding. A self-serve interface moves responsibility to the advertiser and needs an accountable operating record.

Define the campaign job

State the audience problem, message, destination action, accepted outcome, observation period and largest budget the test can authorize. Distinguish awareness, qualified response and sale.

Avoid optimizing a vague engagement label. A served impression, viewability signal, click, usable visit and approved outcome belong to different stages and should not share one success definition.

Check account eligibility

Confirm business verification, payment, market, format and policy requirements in the live account. Record rejected documents and platform instructions with dates.

Public sign-up access does not guarantee campaign or format eligibility. Do not build a production plan around an interface option that the account cannot use.

Map available inventory

Document websites, applications, placement types, devices, markets and source controls exposed for the selected campaign. Preserve estimates separately from delivered evidence.

Inventory availability is not guaranteed reach, quality or context. Sample rendered placements and retain unknown sources. Use narrow caps and exclusions while transparency remains incomplete.

Collect current asset specifications

Retrieve file type, byte limit, dimensions, aspect ratio, animation, border and destination requirements from the receiving platform. Validate the exported file.

An industry size name can render differently or be unsupported. Do not infer that Google specifications apply to another seller. Store the rule version with each creative family.

Create a claim record

Attach exact wording to its supporting evidence, product and market scope, material conditions, owner, approval and refresh trigger. Include price, comparison and performance statements.

Platform approval does not substantiate the claim. Remove or narrow language that cannot remain truthful in the rendered banner and destination. Do not hide a conflicting condition in tiny text.

Design the asset family

Plan how source, proposition, qualification and action adapt across fixed, flexible and responsive containers. Create related layouts instead of stretching one composition.

Preserve semantic order and map each export to its source copy. A correction should identify every dependent rendition. Avoid a large library whose live versions cannot be traced.

Preview extreme rendering states

Test narrow, wide, small, mobile and image-limited examples supported by the platform. Check cropping, text, logo, action, close controls and motion.

Do not approve only the best preview. Responsive systems can combine assets differently. Remove components that create contradictory or incomplete messages and state remaining uncertainty.

Verify accessibility

Check contrast, text resizing, motion, color dependence, action spacing and destination focus, labels and errors. Use representative devices and assistive conditions.

Upload success is not accessibility evidence. Automated checks can find some defects but do not establish full usability. Pause paths that hide conditions or block completion.

Connect the destination

Map every creative ID to a verified URL and fallback. Test redirects, parameters, consent, page readiness, forms, payment and confirmation under representative conditions.

Keep offer and qualification consistent. Count broken or unexpected destinations as campaign loss even when the platform reports a valid click.

Configure location and language

Record provider location option, exclusions, language setting and delivered distribution. Review complete journey language with qualified owners.

A platform signal does not prove residence, nationality or comprehension. Preserve unknowns and separate language from country. Narrow conclusions to the delivered cell.

Set scheduling and frequency

Choose delivery times and caps that match offer availability, support and customer impact. Document the identity scope used for frequency.

Do not describe device or cookie frequency as exact person exposure. Review distribution and complaints, not only an average. Start a new period when the identity method changes.

Control budgets and bids

Set account, campaign and daily or lifetime limits supported by the platform, plus an external authorized ceiling. Record currency, pacing, bid model and stop owner.

A platform forecast is not a spend or outcome guarantee. Monitor reconciled spend and pause before uncertainty can exceed authority. Require approval for material increases.

Test billing and funding

Reconcile deposits, invoices, credits, taxes, fees, currency and unused balance. Confirm who can change payment methods and request refunds.

A funded balance is not media consumed. Keep transaction and delivery ledgers separate. Escalate unexplained deductions before adding funds.

Define source quality actions

Create allow, observe, cap, exclude and pause states using the source level the account exposes. Preserve samples, timestamps, server evidence and reasons.

One signal cannot prove fraud, brand safety and commercial fit. Use several observations and apply narrow action while investigating. A clean sample does not certify all future supply.

Configure measurement

Document provider event, click parameter, analytics request, server event, deduplication, timezone, attribution window and accepted first-party action.

Test known, missing, duplicate, blocked and delayed cases. An attributed conversion is not causal proof or automatic business acceptance. Keep unmatched records visible.

Review data responsibilities

Map identifiers, audience inputs, conversion fields, recipients, purpose, retention and deletion. Use the least data needed and test non-consenting paths.

Self-serve access does not grant permission to upload or combine every dataset. Escalate sensitive data, profiling and cross-context matching to qualified review.

Run a capped launch cell

Use one audience hypothesis, source scope, asset family, destination and accepted outcome with a fixed budget and stop conditions. Preserve all settings before launch.

Do not use the first cell to prove the whole platform. Report the delivered mix and technical loss. A broader test needs renewed limits when inventory or format changes.

Read the response path

Separate eligible requests, served impressions, rendering or viewability evidence, clicks, usable visits, completed actions, approvals and reversals under named definitions.

Find the first material loss. A high click rate can conceal accidental interaction or a weak destination. Do not reward the earliest metric without downstream evidence.

Calculate complete cost

Include media, asset production, resizing, landing work, verification, data review, support and staff monitoring. Reconcile delayed adjustments and refunds.

Compare full cost with matured accepted value. Cheap impressions are not economical when source quality, destination loss or operating burden is high.

Set immediate pause rules

Pause for unsupported claims, broken destinations, uncontrolled spend, failed suppression, severe rendering defects or unresolved customer or data harm. Name restart evidence.

Apply the narrowest safe pause while preserving the record. Do not keep a campaign live merely to gather more data about a material defect.

Audit change history

Record budgets, bids, targeting, assets, destinations, sources and roles with effective times. Store external notes when the interface history is insufficient.

Start a new evidence period after material changes. A result cannot be attributed to a setting when the team cannot show when it became active.

Prepare account exit

Export campaign, transaction, source and conversion records needed for review. Test removal of users, audiences, payment methods and integrations.

Do not wait for suspension or contract end. A recoverable account preserves evidence and limits continuing access. Record any data or balance that cannot be retrieved.

Close with a self-serve decision

Document account, campaign, assets, sources, spend, destination, measurement, accepted outcomes, complete cost, incidents and unresolved limits.

Choose continue, revise, retest or stop with a next exposure, owner and rollback condition. Self-serve should increase control, not remove governance.

Test creative upload states

Upload one valid asset, one intentionally invalid file and one replacement to a non-production campaign. Record validation messages, review states, processing delay and which identifier changes.

Do not equate successful transfer with approval or eligibility. The exercise should show how an operator finds a failed rendition and prevents an obsolete file from continuing to serve.

Control asset naming and versions

Use a naming convention that connects campaign job, size family, language, claim record, destination and revision. Store editable sources and exported checksums under assigned ownership.

Names should help investigators without exposing confidential audience data. Retire superseded assets deliberately, because a platform library can preserve an old message after the landing page changes.

Review malware and destination safety

Scan files and landing routes under approved security practice, check unexpected scripts or redirects, and define escalation for platform or browser warnings. Preserve evidence without opening unsafe content casually.

Creative approval is not a security assessment. Pause the affected asset and destination when behavior changes, credentials are requested unexpectedly or a third-party resource cannot be trusted.

Exercise exclusion behavior

Configure a small, observable exclusion such as a placement, device or audience suppression and confirm the delivered record after sufficient processing time. Note product-specific delay and hierarchy.

A saved setting is not proof that delivery respected it. Treat unexplained exposure outside a material exclusion as an incident and constrain spend while the account owner investigates.

Separate preview from live evidence

Archive platform previews for composition review, then capture authorized live examples with timestamp, source, device and creative identifier. Compare text, crop, motion and destination.

The preview environment may not reproduce every seller container or browser state. Do not use one clean example to certify all inventory; continue sampling the delivered mix.

Manage language variants

Give each language asset an accountable reviewer, approved claim source, destination and fallback. Verify that platform labels and geographic controls do not substitute for comprehension review.

Machine translation or an English master can miss pricing, tone and material conditions. Pause a variant when reviewers cannot support the complete rendered journey for its intended audience.

Reconcile pacing anomalies

Compare authorized ceiling, platform setting, hourly or daily delivery, invoice record and account timezone. Investigate bursts, delayed reporting and spend after a pause request.

A daily average can conceal short periods that exceed operating capacity. Define the evidence and response time required before assuming that a changed budget has taken effect.

Document support escalation

Save the official support route, case identifier, timestamps, requested evidence, response and interim containment for billing, policy, delivery and security problems. Restrict sensitive material in tickets.

Support availability does not transfer decision ownership. The advertiser still needs a safe local action when the provider has not answered or cannot restore a required control.

Approve scale by delivery cell

Expand only the source, asset, device, market and destination cell whose mature accepted outcome and complete cost remain inside guardrails. Set a new cap and review date.

Do not scale the entire account from an aggregate win. Untested inventory and formats remain new exposures, and their performance or customer risk may differ materially.

Check browser cache behavior

Replace an asset and inspect authorized delivery after cache and review delays on representative browsers. Record whether the old creative can remain visible and for how long.

A successful library update may not remove every served copy immediately. Keep urgent claim corrections behind a campaign pause until the operator can verify the live state.

Review ad-label context

Sample placements to see whether advertising identification, creative borders and nearby publisher controls remain understandable in the rendered environment. Escalate ambiguous blending.

The advertiser cannot assume every page surrounds the asset consistently. Avoid designs whose meaning depends on a publisher label or background that may not appear.

Validate click target geometry

Test the intended clickable area, close control and destination on touch and pointer devices. Capture accidental or blocked interaction patterns during the bounded launch.

A reported click can arise from confusing geometry rather than interest. Pause assets that obscure controls or create a material mismatch between visible action and actual target.

Reconcile rejected creative

Record rejection reason, affected rendition, submitted evidence, appeal, replacement and final status. Link the resolution to the claim or specification record.

Do not repeatedly resubmit a disputed asset without understanding the rule. A later approval does not erase the need to review the message and destination independently.

Set end-of-campaign cleanup

Pause delivery, verify spend settlement, export evidence, revoke temporary users, remove unused audiences and confirm destination or tracking cleanup. Record residual balance and open cases.

Completion in the interface is not operational closure. Continue monitoring delayed charges, conversions, complaints and access until the documented maturity and retention windows end.

Self-serve banner launch matrix

The account proceeds only when ownership, asset, spend, delivery and outcome evidence are present.

Retained self-serve banner resources

Existing self-serve campaign, advertising-guidance and account links and graphics remain below in their established order. They do not verify present eligibility, inventory, price, quality, compliance or performance.

Self-serve banner ad questions