A worksheet to compare the same work, separate costs and request delivery evidence before choosing a provider.
Scope, costs and evidence for deciding the first engagement.
A ChatGPT Ads agency proposal should let you picture the work that happens next Monday: who prepares the offer, what the account needs, who changes the website and how you will know measurement works. If everything fits under “strategy, launch and optimization”, several conversations are still missing. Naming the channel does not complete the scope.
This guide helps you review proposals on a shared basis, request useful clarification and decide what the first engagement should cover. It includes a comparison worksheet and two fictional budgets, presented neither as CODE GPT Ads fees nor market prices. Technical references were reviewed on 25 October 2026; account-specific capabilities and conditions need separate confirmation.
Start by checking whether the agency understands what you sell, to whom, and what a useful sales conversation would look like. A presentation can describe the channel beautifully while saying very little about your business. Look for a specific hypothesis: which offer will be tested, where the visitor arrives and which action represents meaningful interest.
If that definition is still missing, prepare the 2027 campaign brief first. Comparing proposals written against different assignments is a remarkably efficient way to create misunderstandings. Give every provider the same offer, available destinations, account situation and description of what your internal team can handle.
Distinguish the provider’s tasks from conditions the platform must resolve. An agency can prepare messages, settings, destinations and agreed tests. Availability of a particular feature or an outstanding review requires its own check. Ask the proposal to identify those dependencies and explain what will be delivered while they remain unresolved.
The official client-account integration guide describes capabilities that require enablement. A public reference does not establish the advertiser’s access or a provider’s partner status. If a capability is essential to the project, request a verifiable confirmation of the scope rather than treating a logo or a label as evidence.
Also check what the agency considers a launch. The status and review documentation distinguishes configured activation, approval and delivery conditions. “We created the campaign” does not complete all those checks. The schedule should show its dependencies and the evidence needed to move to the next step.
Use this worksheet to request precision. It is an editorial framework, not an official agency ranking or a score that replaces your decision. Its purpose is to establish whether two proposals cover the same work. Add your own columns for each provider’s response and the question that remains open.
| Area | Deliverable to request | Open question |
|---|---|---|
| Account | Identity, owners and access checks. | Who retains administration, and what needs enabling? |
| Offer | Hypothesis, audience and approved conditions. | Which commercial conversation do we want? |
| Creative | Messages and assets defined by format and review. | What production and how many revisions are included? |
| Destination | Pages, adaptations and intended journey. | Who implements and checks changes? |
| Measurement | Event map, tests and evidence. | What does “tracking included” cover on this website? |
| Management | Owners, reviews and a change record. | What work happens during the engagement? |
| Costs | Line items, period, assumptions and extras. | What is outside the compared total? |
| Continuity | Inventory and handover to another operator. | What information lets us keep working? |
Identify the advertiser, the person administering the account and the person authorizing spend. As an operating principle, avoid making one external individual your only route into the account. Agree the permissions required for the work and how they will be reviewed when it ends. The brief can record responsibilities without becoming a store of passwords.
The official account reference helps distinguish identity, configured status and reviews. Use it to understand what has been checked, without treating those details as an ownership agreement between client and provider. The commercial arrangement should explain who keeps the resources and how continuity is maintained.
Include an inventory of destinations, creative, measurement sources, documentation and necessary settings. Ask the final handover to identify what is retained, where it is held and who receives it. An isolated export does not demonstrate that another team can operate the project. The inventory needs to explain relationships, dependencies and outstanding work.
Ask whether the scope covers outcome definition, implementation, consent behavior, testing and report interpretation. Installing a tag and verifying a successful enquiry are different tasks. If your website uses an external form, several domains or a particular checkout, those dependencies should appear before the price is agreed rather than becoming a surprise during implementation.
Request a successful case, a failed case and a repetition, with expected and observed results. If browser and server send the same action, ask how their shared identity will be checked. Our Pixel and Conversions API guide includes an acceptance plan to develop this part of the scope.
Agree where commercial quality is confirmed as well. An advertising event might represent a received enquiry while the CRM establishes fit later. An agency reporting both stages should explain the relationship and its limits. A platform total should not be relabelled “new customers” without establishing that connection to the business source.
Separate setup, management, media and additional work, then use the same time horizon and assumptions. A percentage-of-spend fee needs its calculation base, any minimum and what happens when spending changes. A monthly amount without a duration, or production without defined units, cannot close a useful comparison between providers.
This exercise uses invented amounts and three months of management. Both proposals allocate €1,200 per month to media. Taxes and unpriced extras are excluded; it does not assume an advertising configuration accepts or spends exactly those amounts. It is a budget-reading exercise, without an investment recommendation or a claim about actual agency fees.
| Cost line | Proposal A | Proposal B |
|---|---|---|
| One-off setup | €900 | €300 |
| Monthly management | €350 | €600 |
| Three months of management | €1,050 | €1,800 |
| Known service fees | €1,950 | €2,100 |
| Planned media, three months | €3,600 | €3,600 |
| Known total | €5,550 | €5,700 |
We checked these sums in a local calculation: B’s known total exceeds A’s by €150, although its setup costs less. That does not establish which proposal is better. If B includes production that A charges separately, information is still missing. An unpriced extra should remain an open item; treating it as zero would manufacture a complete comparison.
The breakdown of ChatGPT Ads pricing, setup and management develops these cost lines. For media-cost interpretation, also read CPC, CPM and bidding strategies. Fees, authorized budget and commercial outcomes answer different questions. Put them beside each other without pretending they are interchangeable.
Before commissioning, ask to see the report structure and an explanation of the decision it supports. You can review an anonymized example or a clearly labelled fictional template. Look for period, settings, changes, outcomes and next action. An upward graph without context has considerable talent for decorating a meeting, but rather less for explaining the next investment decision.
The reporting documentation distinguishes metrics and processing timelines. The proposal should identify the report and explain how attribution windows, timezone and comparable periods are retained. Ask the team to separate observed numbers from hypotheses explaining their movement, particularly when recommending an increase in investment.
To review interventions, agree an operating record containing the resource, change, reason and owner. The audit-log reference describes recorded changes to account resources. That record can add technical context; it does not replace the commercial agreement or establish by itself that a particular change was authorized.
Write down what your team needs to receive to accept the first phase. For example: reviewed offer and messages, checked destinations, documented settings and agreed measurement tests. If a platform dependency remains open, ask for its owner and next action. Acceptance of the provider’s work should correspond to the actual commissioned scope rather than an implied promise of platform approval.
Assign an owner to each deliverable and decide how corrections are recorded. “Ongoing optimization” needs an operating routine: which signals are reviewed, what changes can be proposed and when a decision follows. A monthly meeting can form part of the service, but putting it on the calendar does not yet describe the work between meetings.
Return the questions that affect comparison to each provider: unpriced work, unassigned responsibilities, production limits, dependencies and outcome definitions. Ask for one consolidated version. Combining several email replies with an old proposal can produce a scope that nobody entirely recognizes. The document used for your decision should be understandable without reconstructing the correspondence.
Then consider how that scope fits your team. A business with its own development and analytics capacity may need a different engagement from one delegating everything. Our ChatGPT advertising service explains how we prepare a pilot. Tell us which offer you want to test and what you need to delegate to define a comparable first assignment.
Suppose a proposal includes “landing page and tracking” without saying whether new development is involved. Before discussing the price, return a question that helps complete the scope. The following message is an editorial example to adapt to your project, rather than a real negotiation or a set of conditions every agency must adopt.
To compare the engagement, please specify which page is delivered, which changes are included and who implements them. Identify the action being measured, the tests we can review and any dependencies outside scope. Separate this work from media spend and subsequent management so our internal team can review the same assignment.
The answer should identify a deliverable and a way to check it. “Adapt the existing form and verify receipt” describes different work from designing a new landing page with a sales-system integration. If both proposals use the same label for those assignments, the comparison issue still sits in the scope, even though you already have two prices on the table.
Then try a continuity exercise on paper: imagine somebody else reviews the project next week. Can they locate the account, offer version, destinations, measurement map and open decisions? If they would need to reconstruct everything from scattered messages, request a clearer inventory. This exercise reviews the proposed documentation; it establishes neither actual access nor a completed technical handover.
Keep the answer within the consolidated version, alongside any changes to cost or schedule. The person approving the work and the person reviewing delivery can then use the same document. It is a modest project discipline that helps everyone agree what has actually been commissioned before the work starts.
That depends on the work and the calculation base. Compare deliverables, period, minimums, revisions and extras before comparing the fee model. The pricing label does not explain how much work it covers or the amount you will eventually authorize. Ask for a calculation using your agreed assumptions.
Not by itself. Ask the provider to distinguish verifiable results, examples and scenarios. A forecast needs assumptions and limits; a real case needs enough context for interpretation. Neither guarantees that your offer, market and customer journey will behave the same way when the campaign is run.
You can propose a separate scope identifying the deliverables and the team that activates or operates afterwards. Agree the handover so documentation and tests remain useful to that team. Project continuity deserves defining even when the first assignment is small or covers only part of the journey.
We use your offer, destinations and available team to define scope, deliverables and responsibilities.
What you sell, which markets you work in and what you want to learn from the first pilot. If the scope is not clear yet, we will define it with you.