The problem this solves
Agency reporting is the same ritual multiplied by the client roster. Each account manager burns the first days of the month assembling decks that are structurally identical and completely unshareable between clients, and the work scales linearly with every account the agency wins. Reporting week is the quiet tax on agency margins: senior people doing template work at scale, at exactly the moment clients are judging whether the retainer is worth it.
Client-facing adds a layer of stakes that internal reporting never has. A wrong number in an internal deck is an awkward correction; a wrong number in a client deck is a credibility event, and a stray reference to another client's data is far worse than that. So agencies double-check everything by hand, which makes reporting week even longer, or they trim the reports to the safest minimum, which makes the retainer look thinner than the work actually is.
The mission treats the roster as what it is: one report structure, many domains. Each client's deck is built from that client's data in isolation, every figure traces to the tool call that fetched it, and the whole run lands in a single index so the agency sees the book at a glance while each client sees only their own world.
How the mission runs
- Take the client roster from the thread. You provide the client domains and any per-client notes: what each engagement covers, which channels apply, and anything this month's deck should address. The roster is the only per-client input; everything downstream runs the same verified process for each account on the list.
- Pull each client's performance in isolation. Per domain, Search Console provides the organic story and AI Visibility reports how the client shows up in AI-generated answers, with Ad Libraries adding competitive advertising context where the engagement covers it. Each client's data is pulled and processed separately, so nothing can bleed between decks.
- Build each deck in a client-safe format. Each client gets a Gamma deck in the agency's standard structure: the month's headline, channel performance, competitive context, and next month's focus. Client-safe means scoped by construction - only that client's data, no internal margin notes, nothing about any other account anywhere in the file.
- Verify figures against their sources. Before anything is indexed, each deck's figures are checked against the tool calls that produced them, with date ranges and definitions pinned per deck. This is the gate that replaces reporting week's manual double-checking, and it applies identically to the first client and the fifteenth.
- Index everything in Google Sheets. A Google Sheets index lands with one row per client: the deck link, the month's headline numbers, and anything flagged for the account manager's attention before sending. The index is the agency's view of the whole book; the decks are each client's view of their own.
The prompt
This is the exact objective the agent receives. Swap the obvious placeholders for your own domain, segment or channel and run it as-is from the console, Slack, or the API.
What comes back
One finished Gamma deck per client domain, each built solely from that client's data in the agency's standard structure, with figures traceable to their source calls and nothing cross-client anywhere. Plus a Google Sheets index of the full run: deck links, headline numbers, and per-client flags for the account managers. Reporting week compresses to a review pass, and the review has evidence to lean on.
Make it yours
- Add per-client branding notes to the roster so each deck carries the client's name and framing conventions, ready to send without cosmetic rework.
- Schedule the run for the first of the month with account-manager review as the gate, so decks are approved and sent while competitors are still assembling theirs.
- Include a private internal appendix per client, delivered separately from the client deck, holding the candid notes the account team needs and the client never sees.
Frequently asked questions
How is cross-client data leakage prevented?
By construction rather than by proofreading. Each client's data is pulled and processed in isolation, decks are generated per domain from that domain's data alone, and internal commentary lives in the separate index rather than the client files. The failure mode of pasting into the wrong template has no equivalent here.
Do account managers review before clients see anything?
That is the recommended gate: the run delivers decks plus the index for review, and sending remains a human act. The index's flag column exists precisely for this pass, pointing account managers at the decks needing a narrative touch before they go out.
What happens when a client questions a number?
The number traces to the tool call that produced it, with the date range and definitions pinned in the deck's own lineage. The account manager answers from evidence in minutes instead of reconstructing a spreadsheet from three weeks ago, which is the difference between a wobble and a churn risk.