Meta + GSheet Agent

This guide explains what the Meta + GSheet agent does, when to use it, what you need, and the exact steps to get the best value out of it. It also covers guardrails, how to use a Google Sheet as your source of truth, and known limitations.

What this agent is

One workspace that puts your Meta Ads data and creative insights in a single place. It helps you monitor performance, diagnose issues, and act on clear recommendations across Facebook and Instagram. You can also feed it a Google Sheet to join in offline or third-party data.

Use it to replace tab hopping, manual exports, and guesswork.

When to use it

• You need a single view of campaign performance and budget pacing across accounts.
• You want to find winning creatives and spot fatigue early.
• You want audience and placement context without drowning in pivot tables.
• You need clean, sharable reports for stakeholders.
• You want to combine Meta data with margins, stock, or lead quality from a Sheet.

What you need before you start

• Access to the correct Meta ad account.
• Optional Google Sheet for any extra data you want the agent to read. Share the Sheet by link with Viewer, Commenter, or Editor.
• Consistent naming for campaigns, ad sets, and ads. This reduces misclassification.

How to get started

1. Pick your Meta ad account.
2. Choose the agent: “Meta + GSheet” Agent.
3. Warm up the context.
   • For account questions, run "Get details of {adaccountID}."
4. If you will use a Sheet, paste the link and then, provide the tab and timeframe.
5. Ask your question in plain language. Examples are at the end of each capability section.

Using a Google Sheet as source of truth

You can attach a Sheet to help the agent make better decisions.

What can you put in the Sheet

• Source of truth data points if any
• Lead quality or revenue realized by campaign or ad set
• Prospecting or Retargeting labels when naming is inconsistent
• Any other structure data, that can help prism provide more nuanced analysis

If you want the agent to use external or offline data:

1. Paste the Sheet link in chat with open sharing.
2. Specify the timeframe and the tab to use.
3. Ask questions about that data, or ask the agent to blend in with Meta Ads insights.

Tip: Keep column names clean and consistent. The fewer ambiguities, the faster the agent answers. You can follow this template (opens in a new tab) for reference.

Capabilities

Campaign performance and analysis

• Real time monitoring of spend, impressions, clicks, CTR, CPC, CPM, ROAS across account, campaign, ad set, and ad.
• Active asset detection that flags what is spending and delivering, not just toggled on.
• Historical comparisons across any two periods with breakdowns.
• Conversion tracking for purchases, leads, and other supported standard events.

Example prompts

• “Show performance last 7 days vs prior 7 days by campaign. Call out top changes in ROAS and CPA.”
• “Which ad sets delivered zero impressions yesterday but had a budget allocated. List likely causes.”

Creative analysis

• Creative audit that ranks ads by ROAS, spend, CTR, and conversion rate to find winners and weak links.
• Visual asset analysis that extracts descriptions of images and videos to guide next iterations.
• Works across image, video, carousel, and dynamic creative.

Example prompts

• “List the top 20 creatives by ROAS in the last 30 days and describe common visual patterns.”
• “Flag creatives with falling CTR week over week and suggest variants to test.”

Audience and targeting insights

• Audience size estimation to predict reach and delivery potential before launch.
• Performance breakdowns by age, gender, geo, device, placement, and time of day.
• Saved audience review with overlap and waste indicators.
• Targeting configuration readouts so you can see exactly how each ad set is targeted.

Example prompts

• “Break down ROAS by placement for the last 14 days and recommend cuts or budget shifts.”
• “Show which saved audiences have declining conversion rate and why.”

Budget and spend management

• Account spending limit checks and current pacing.
• Budget pacing by campaign and ad set with alerts on underspend and overspend.
• Cost efficiency view that pairs ROAS or CPA with spend to find the best money multipliers.

Example prompts

• “Show underpaced campaigns vs plan for the month to date and propose daily budget changes.”
• “List ad sets with CPC up 30 percent week over week and give likely drivers.”

Operational excellence

• Automated rules tracking so you can see when rules changed bids or budgets.
• Activity history that logs edits with timestamps.
• Account health checks for status, billing, and delivery issues.

Example prompts

• “Summarize all automated rule actions taken in the last 72 hours and their performance impact.”
• “List account level issues blocking delivery and the actions to resolve them.”

Data visualization and reporting

• Line, bar, pie, and scatter charts generated directly from your data.
• Report builder that publishes to a shareable web URL.
• Custom dashboards that mix multiple metrics and timeframes for executives.

Example prompts

• “Create a dashboard showing ROAS, spend, and conversions by campaign for the last 30 days with week over week deltas.”
• “Publish a report link that compares this month to last month by placement.”

Practical marketing use cases

Daily operations

• Watch active delivery, catch zero impression ad sets, and fix pacing.
• Track spend vs monthly budget and move money to high ROAS areas.
• Surface top creatives for scale and kill obvious waste.

Strategic planning

• Study audience performance to refine targeting and exclusions.
• Compare formats and creative themes that win for your category.
• Map geo performance to your expansion plan.

Performance optimization

• Find underperforming ad sets to pause or rework.
• Reallocate to the highest ROAS campaigns and placements.
• Track funnel metrics by audience segment to tighten weak stages.

Client and stakeholder reporting

• Spin up clean charts and narrative highlights in minutes.
• Compare any two periods with callouts on what changed and why.
• Build dashboards that refresh without manual exports.

Guardrails and tips

• Always set a clear timeframe and currency in your question.
• Warm up account context first. The agent answers faster and with fewer follow ups.
• Name the Sheet tab and key columns when you paste the link.
• Keep campaign naming consistent. If you want Prospecting or Retargeting filters to be exact, provide a Sheet mapping.
• Prefer concrete asks. Example: “Flag ad sets over ₹X spend and ROAS under Y in the last Z days” is far better than “Give me CTR for last month”

Limitations and assumptions

• Custom events are not yet supported. The agent reports on standard Meta events such as Purchase and Lead.
• Prospecting vs Retargeting classification works best when you provide a Sheet with a mapping column.
  • If no mapping is provided, the agent will infer based on naming patterns. This can be wrong. Provide a mapping for reliability.
• If creative assets lack clear names or tags, visual pattern analysis will still work but insights will be broader.
• Reporting links are generated from the current analysis scope. Change the scope and regenerate if the audience or timeframe changes.
• Some metrics overlap (for example, Clicks (all) vs. Link Clicks). Prism defaults to Clicks (all) unless you specify.
• Creative or ad-level analysis may take longer because Prism needs to drill down across all assets.
• Default attribution is assumed unless specified otherwise.