Google Ads Partner Operating Contract

This is a partner operating guide, not a campaign-setup tutorial. Platform instructions change; the obligation to connect search intent, spend, destinations, calls, CRM evidence, and company ownership does not.

Outcome

Capture relevant search demand and turn it into a transparent learning loop from query and click through qualified conversation, opportunity, and sold work where evidence is mature enough.

The partner owns paid-search decisions and the quality of the paid-search evidence. It does not own the company's project-fit standard, Sales follow-up, website production, CRM architecture, or business claims unless those are separately assigned.

Readiness before launch or scale

The company and partner should be able to name:

• the service, market, project-fit boundary, and intent worth buying;
• the destination and promise for each meaningful campaign group;
• calls, forms, confirmation states, and the accepting owner;
• what counts as a raw inquiry, qualified opportunity, appointment, proposal, and sold project;
• the current measurement limits and source-of-truth disagreements;
• the approved budget boundary, decision window, and evidence that would change the plan; and
• company ownership of the account, billing relationship, data, landing pages, numbers, and exportable history.

If a lead path cannot pass end to end, the campaign is not ready merely because ads can run.

Baseline and onboarding evidence

The initial review should extend beyond the ad account:

Area	Minimum evidence
Search demand	Search terms, match behavior, negatives/exclusions, geography, schedule, device and network context, and the intent each group is meant to capture
Ads and promise	Current message, offer, proof, claims, extensions/assets, and alignment with the destination
Destination	Page job, project fit, mobile experience, call/form path, privacy, confirmation, and response ownership
Calls and forms	Number/form mapping, spam treatment, recording/transcript consent where applicable, notifications, and a tested receipt
Measurement	Agreed conversions, tag/event sources, duplicate prevention, attribution windows, offline or CRM feedback, data freshness, and known gaps
Sales and CRM	Source persistence, qualification/disposition, stage meanings, owner, follow-up evidence, and lost-reason feedback
Account control	Company administrator, partner access level, billing owner, change history, exports, recovery, and exit requirements

Access records may name the account, role, administrator, and approved password-manager reference. They never contain passwords, recovery codes, or private secret links.

Decisions the partner owns

• Campaign and query structure tied to real search intent
• Budgets and pacing inside the approved boundary
• Search terms, exclusions, targeting, and bid/automation choices
• Ad and asset testing with message match
• Paid-search diagnostic interpretation
• Recommendations for destinations, tracking, call handling, CRM feedback, and the next test
• scale, hold, repair, or retire recommendations with evidence and uncertainty

The partner should document material changes and explain the decision, not simply report that an optimization was made.

The measurement seam

A platform conversion is not automatically a qualified lead. Keep these layers visible:

1. ad-platform response;
2. identifiable inquiry;
3. qualified opportunity using the company's fit rule;
4. Sales progress using event-defined stages;
5. sold work and commercial outcome; and
6. accounting actuals where relevant.

Preserve original source details and acknowledge attribution limits. Uploading deeper outcomes can improve platform learning only when identities, consent, matching, deduplication, stage meanings, and data quality are reliable. Do not send private call or CRM content to an unapproved system.

Working rhythm and handoffs

A useful review answers:

• What changed in search demand, spend, query quality, destinations, or follow-up?
• Which inquiries were qualified, poor fit, duplicate, spam, unreachable, or still unknown?
• What can the evidence support now, and what is still too immature?
• Which change is recommended, who owns it, and what will verify it?
• What did Sales, Website, Lifecycle, Operations, or Measurement fail to receive?

Cadence follows spend, risk, and conversion lag. A fixed weekly or monthly meeting is not the operating model; a dependable decision and handoff loop is.

What good looks like and how to exit

The seat is operating well when:

• campaigns can be traced to intent, destination, spend, and downstream outcomes;
• lead quality is reviewed with Sales rather than inferred from cost per form;
• material changes and decisions are documented;
• broken tracking, routing, or destination conditions trigger repair instead of hidden optimization;
• the company can independently access accounts, billing, numbers, reports, and history; and
• another qualified operator could continue from the returned artifacts.

Conversion-definition drift, lost source detail, mismatched landing promises, broken calls or forms, unexplained changes, or incomplete company control are reasons to repair the relationship or reconsider the seat.

On exit, return the current structure, change and decision history, query/exclusion logic, ads/assets, audiences, conversion specifications, reports, landing-page notes, unresolved risks, and access removal confirmation.

Use this with Paid Readiness and Learning, Marketing Asset Control, and the Vendor Pilot and Exit Scorecard.