AI creative project report template for clients
Separate accepted work, recorded effort, estimated avoided work and campaign results in a client-facing report.
Report accepted work, recorded effort and spending, estimates of avoided work, and campaign results as separate measures. Name each measure's source and label it as observed, calculated, estimated, a hypothesis or untested. This gives the client a basis for deciding what to commission next.
You can copy the plain-text client report form and fill it with project records.
Copy the report template
Use this as a starting point. Replace each bracket with a project record or mark it unknown.
Project and reporting period: [name, dates]
Agreed scope: [concepts, creator work, versions, placements, account routes]
Accepted files: [file names or links, versions, acceptance dates]
Open items or scope changes: [item, owner, next decision]
Creator work: [activity, time, source of record]
Staff effort: [activity, hours, time-log source]
Fees and pass-through costs: [amount, currency, invoice or receipt]
Tool or media costs: [amount, currency, account record]
Comparison baseline: [measure, source, date, limits]
Effort difference: [calculation; observed, calculated, estimated, or hypothesis]
Campaign delivery: [asset, spend, exposure, reporting window]
Attributed results: [platform, attribution setting, reported result]
Incrementality: [test design, estimate, uncertainty; or not measured]
Unknowns and next decision: [what is still unclear, who decides, by when]
Start with the commissioned work
Restate the agreed output in one short paragraph: the concepts, creator input, languages, placements and account routes included. Then show the amount accepted, pending and removed from scope. Link the client to the current delivery files rather than leaving them to identify the latest version.
Distinguish a concept from an adaptation. A new argument, a new opening and a new aspect ratio are different kinds of work. Meta's creative training distinguishes concept and format diversification, which is a useful basis for reporting those categories separately. It does not establish a value multiplier for either one. Meta creative-diversification guidance.
The report can show that one accepted concept produced several usable placement files. It should not describe every file as an independently developed idea.
Filled hypothetical example
These fictional records show how to keep delivery, effort and results separate.
| Measure | Project record | What it supports | What it does not support |
|---|---|---|---|
| Delivery | Two concepts; four accepted placement files | The agreed files were delivered | Four separately researched concepts |
| Creator review | 2 hours in a creator-submitted time log | Recorded review time | Total creator effort unless other work is logged |
| Staff effort | 18 hours in the current log; 24-hour planning estimate | Six fewer hours than estimate (25%) | Proven cause or cash saving |
| Campaign | One file spent C$1,200 and received eight platform-attributed purchases | Eight purchases under the account's attribution settings | Eight additional purchases caused by the ad |
| Next step | No holdout or lift test was run | Incremental sales remain unknown | A claim that the ad added sales |
The 24 hours are an estimate, not a measured prior job. Attribution is not incrementality. A lift study needs a suitable control design and may not be available for every account. See how to measure incremental sales.
Give the client a delivery table
| Work | Planned | Accepted | Still open | Decision needed |
|---|---|---|---|---|
| Concepts | Agreed ideas | Ideas ready for use | Unsupported or unfinished ideas | Which to develop next |
| Executions | Defined scripts and edits | Accepted masters | Review or repair | Approve, revise or remove |
| Adaptations | Formats and languages | Accepted versions | Missing checks | Assign reviewer or narrow scope |
| Account setup | Named account routes | Verified destinations | Permission blocks | Client or creator action |
Use the terms agreed in the proposal. Do not invent a new measure at reporting time because it makes the volume look larger. If the client changed the scope, show that separately so delays and added work have a fair explanation.
Report spending and effort on consistent terms
Show the agency fee and agreed pass-through expenses in the client report. If you share internal delivery costs, make the accounting basis clear. A fixed service fee and a reimbursement of actual third-party cost are different commercial arrangements.
For tools, distinguish the subscription allocation, credits consumed and any added charges. HeyGen documents account-dependent credit rates and limits. That makes the actual project account and its usage record a stronger basis than a generic per-video estimate. HeyGen credit guidance.
Include failed attempts in internal delivery analysis even if the client buys only accepted outputs. Otherwise the agency may underestimate the next job. Avoid making the client read every rejected render; group the causes that affect future decisions, such as product-detail repairs or changes after script acceptance.
Show where creator work went
Record creator effort in separate activities: briefing or interview, source recording, script review, performance review and corrections. Distinguish time the creator actively worked from time a project waited for the next response.
Less filming does not establish less total creator work. A campaign may replace recording sessions with detailed review. If the creator records their own hours, identify that as self-reported time. If the agency only scheduled a session, do not treat the whole calendar block as measured work.
If review records show that the brief lacks creator context, consider another interview as a testable next step. Record whether it reduces that kind of revision before describing it as a better workflow.
Calculate avoided work only against a credible alternative
Use a comparison that matches what the client would otherwise have commissioned. Record the source of the baseline: a prior comparable job, a current quote or an explicit planning estimate. Label its weaknesses.
In the filled example above, the baseline is a planning estimate. The gap is a difference from that estimate, not proof of time saved against a measured alternative. The production method's effect remains a hypothesis; team experience, brief complexity and review behaviour may also differ.
If those six hours allow staff to do other work, report released capacity. Call it a cash saving only when spending actually fell on the stated basis. Multiplying time by a charge-out rate does not mean the client's invoice or the agency's wage bill fell by that amount.
Where there is no credible baseline, report actual effort and cost without a savings percentage.
Keep campaign performance in its proper place
The creative delivery report can link to the media report and identify which assets ran. If an asset never received meaningful exposure, say that performance remains unresolved. An accepted file is ready for use; it is not a proven winner.
When the team wants to compare executions under a planned test, use the platform's current testing tools and the media buyer's analysis. TikTok's split-testing guidance describes a controlled comparison between two versions; ordinary delivery figures answer a narrower question. TikTok split-testing guidance. Google's Conversion Lift guidance describes comparing exposed and control groups to estimate incremental conversions; eligibility and uncertainty matter. Google Ads Conversion Lift.
End with a decision the client can make
Recommend the next action based on the observed problem. If review consumed most of the schedule, improve the source brief and feedback process. If the product interaction failed quality checks, record the real footage. If several versions remain untested, discuss the media plan before commissioning more.
End with the accepted delivery record, the next decision and the evidence behind both. Keep estimates and hypotheses labelled so the client can tell what the project showed from what it may show later.
Connect this decision to the campaign
Connect accepted assets, rights costs, media results and renewal dates to the pilot evaluation plan. Use the incremental-sales guide when the client asks whether the ads caused additional purchases; delivery counts cannot answer that.