Case study · Agency operations · Boutique influencer agency, four account managers

An agency's creator roster that updates itself every morning in Google Sheets

A four-person agency keeps a roster of about 400 creators in a Google Sheet: beauty, fitness and food, mostly US. It used to go stale within weeks, so every pitch started with someone re-checking handles by hand, and client decks carried follower counts from last quarter. Now the roster sheet is connected to Zaver once, a Profile metrics run is scheduled every morning at six, and formulas flag drops, private or renamed accounts and engagement below the floor. The weekly top-up is one ask in Claude: find the gap, append the rows, get public contacts. The sheet stays the system of record; nobody learns a platform.

Google SheetsZaver Radar: Profile metrics → Google Sheet, scheduledClaude or ChatGPT with the Sheets connectorZaver Radar over MCP

400creators in the roster sheet, refreshed every morning at six by a scheduled run
100credits per full refresh at the real price of 250 credits per 1,000 profiles
20creators ranked from 51 screened in one weekly top-up search
0 hrsof manual re-checking per pitch since the schedule went live; it used to take about half a day a week per account manager

The job

The roster is the agency's product. Every pitch, every client deck and every rate negotiation starts from it, and every one of them starts with the same chore: open the sheet, open Instagram, re-check followers and recent Reels for the twenty creators in question, fix the ones who renamed or went private, and type the numbers back in. Across four account managers that was roughly half a day a week each, and the numbers were stale again by the next pitch. A platform would have solved it by replacing the sheet with a seat licence and a learning curve. The agency wanted the opposite: keep the sheet, lose the chore.

How it ran, step by step

Every scene below is a redacted rendering of the real exchange: the ask as typed, the tools that ran, and what came back. Handles, names, emails and prices are masked.

  1. 1

    Day 1 · 20 minutes

    Connect the roster sheet and fill it once

    Connect my roster sheet and fill followers, average reel views, engagement rate, posts, category and full name for every handle in column A, writing from column F

    In Radar, Google Sheets takes the sheet link once and the service account gets editor access. The run form (shown below, from the live product) asks for the connected sheet, the tab, the column with handles, the first data row, the first output column and the metrics to write: followers, following, posts, average reel views over the last 12, engagement rate, verified, category and name. Leave the last row blank and it reads until the first empty cell. A notes column takes errors such as a private or renamed account. 400 profiles cost 100 credits. From Claude the same thing is one ask that calls the data tool.

    run_data_tool profile_metrics_sheetget_run

    Radar: the Profile metrics → Google Sheet run form with connected sheet, tab name, handle column A, first data row 2, output column, metrics checkboxes, overwrite, notes column and the Queue run and schedule buttons
  2. 2

    Day 1 · 5 minutes

    Schedule it for six every morning

    The same run goes on a schedule in Radar: daily, 06:00, overwrite existing values, last row blank so new rows are picked up automatically. By the time the first account manager opens the sheet, every row is from this morning. The cost is the same 100 credits a day, about 3,000 a month, which fits the Growth plan's 5,000. A smaller roster or a weekly cadence fits Starter.

    Radar · Schedules

    Chat scene: the first run summarised, then the daily 06:00 schedule confirmed
  3. 3

    Day 1 · 15 minutes

    Add the columns that turn numbers into decisions

    Three formula columns next to the refreshed data: change versus the baseline followers recorded when the creator was added, engagement against a floor per tier, and a flag that reads "dropping", "below floor", "check account" (from the notes column) or blank. Conditional formatting colours the flag. Nothing else changes; the account managers' own columns (niche, owner, city, last campaign, rate card) stay where they were.

    Google Sheets formulas

    Google Sheets scene: the roster with refreshed metrics, change versus baseline, engagement floor and flag columns
  4. 4

    Every morning · 2 minutes

    The morning check, in one ask

    Read the Roster tab: who dropped more than 5% followers since baseline, whose engagement is below the floor for their tier, and which handles errored? Draft a one-line note to each owner.

    With the Sheets connector, Claude reads the tab, groups the flags by account manager and drafts the notes. A creator who renamed gets a lookup by name in Zaver and the handle fixed in column A; a creator who went private is parked; a creator whose Reels fell off a cliff gets a note before a client asks. The check takes two minutes and happens before anyone opens Instagram.

    Sheets connectorfind_creator_by_name

    Chat scene: the morning check listing flagged creators grouped by account manager with draft notes
  5. 5

    Weekly · 10 minutes

    Top up the gaps with discovery and contacts

    We have 12 fitness creators in Texas and need 20. Find creators 20k to 150k with workout reels and a mostly US audience, skip anyone already on the roster, get public contacts for the five best, and append them to the Roster tab

    Discovery screened 51 profiles and ranked 20, with two gym and community accounts flagged as brands. Contact enrichment on five of the top picks returned a public business email for two; the other three list only a link-in-bio or ask for DMs, which the sheet records as the contact route. The new rows go in with today's date as the baseline, and tomorrow's six o'clock run fills their metrics without anyone touching the schedule.

    discover_creatorsenrich_creator_contactsSheets connector

    Chat scene: Texas fitness discovery results and contact lookup, handles and emails masked, appended to the roster
  6. 6

    At pitch time

    Check audiences only for the creators you are about to pitch

    For the five fitness creators we are pitching to the client, run audience checks: US share, real-follower share, engagement versus tier, fair price. Put the results on a shareable shortlist.

    The daily refresh keeps public metrics fresh for everyone; audience reports are spent only on the handful going into a deck, at 8 credits each on Growth. The results go on a shareable shortlist the client can open without a login, with the roster's own rate card next to Zaver's fair price.

    get_creator_insightsshare_shortlist

Who did what

The marketer

  • Owns the sheet and its columns
  • Sets the schedule once
  • Reads the morning notes and decides
  • Approves which creators go into a pitch

Claude

  • Runs the first fill and the weekly top-up
  • Reads the sheet and drafts the morning notes
  • Fixes renamed handles by name lookup
  • Builds the shortlist for the client

Zaver, over MCP

  • Profile metrics into the sheet on a schedule
  • Discovery with brand flags
  • Public contact enrichment
  • Audience and fair-price reports
  • Shareable shortlists

What it costs: the schedule against doing it by hand

The roster did not need a platform seat to get fresh. It needed one scheduled run and a few formulas.

ItemBy hand, beforeScheduled refresh
Keeping 400 rows currentAbout half a day a week per account manager, re-checking the creators in the next pitchA daily run at 06:00: 100 credits a day at 250 per 1,000 profiles, about 3,000 a month
PlanNone, and the numbers were still stale between pitchesGrowth at $299 a month billed annually covers the daily cadence; Starter covers a weekly one
Renamed, private or dead handlesFound when a pitch failsWritten to the notes column by the run; fixed by name lookup in the morning check
Adding creatorsPlatform search, export, paste, re-formatOne ask: discovery, contacts, append; the next run fills the metrics
Where the roster livesThe sheetThe same sheet, same columns, same sharing

Time figures for the manual column are the agency's own estimate for its four account managers.

Outcome

Since the first morning, every pitch starts from numbers that are hours old instead of months old, and the half-day chore is gone from four calendars. The weekly top-up keeps each niche and city at its target count, with contacts attached as rows arrive. Audience reports are spent only on the creators going into a deck. The sheet, the columns and the sharing stay exactly as the team built them; the only new thing on anyone's screen is a flag column that is usually blank.

Run the same flow

Connect Zaver to Claude or ChatGPT at /mcp, connect Gmail in the assistant's connector settings, and start with the first ask. The playbooks cover each step on its own.

Keep a creator roster up to date automaticallyTurn your scouting sheet into a live trackerAudit an inherited creator listGoogle Sheets add-onSheet templatesConnect Zaver to Claude or ChatGPT

Questions

Is this a real customer?

Yes. A boutique agency with four account managers runs its creator roster this way. The roster shown is a redacted rendering with masked handles, and the weekly top-up on this page (a Texas fitness search and contact lookup) was run in Zaver on 30 October 2026.

Does the run change my own columns?

No. It writes only to the output columns you choose, starting at the column you name, and only the metrics you tick. Your niche, owner, rate and notes columns are untouched. Overwrite can be switched off if you want to keep old values.

What happens to new rows?

Leave the last-row setting blank and the run reads until the first empty cell in the handle column, so rows appended during the day are refreshed at the next run.

What about handles that changed or went private?

The run writes the error to the notes column instead of failing the batch. A renamed creator can be found by name in Zaver and the handle corrected.

How many credits does it take?

250 per 1,000 profiles per run. A 400-row roster refreshed daily is about 3,000 credits a month, inside the Growth plan's 5,000. Weekly refreshes of the same roster are about 430 a month, inside Starter's 2,000.

Can I do this without Claude?

Yes. The connection, the run and the schedule are set up in Radar's web app; the formulas are plain Google Sheets. Claude or ChatGPT adds the morning check and the one-ask top-up.

Reconstructed from the working session and published with the customer's permission. Brand, creator identities, contact details and commercial figures are redacted. Scene images are renderings of the redacted messages, not screenshots of the original thread.