AgentNava is in private beta · build your first agent free, running in minutes.See what you can hire →

AI agent for ad campaign performance

Meet Ada, the Ad-campaign analyst agent from AgentNava's marketing library

An AI agent for ad campaign performance watches ad spend every day, flags campaigns that waste budget, and finds ones worth scaling. AgentNava's Ada pulls Google Ads and Meta Ads data, checks conversions in Google Analytics, and reports return on ad spend in plain language.

Hire Ada →See an example run

First agent free · billed in credits per agent turn · See pricing

Agent brief · AdaStarter · Marketing

Watches ad spend across channels and flags what to cut, scale, or fix.

  • Pull and reconcile daily performance
  • Flag underperformers
  • Surface scaling opportunities
  • Track budget pacing
Runs
Daily
Workflows
4
Tools
3
Starter agent · yours to shape once you hire it
Runs
Daily
Connects to
Google AdsBetaMeta AdsBetaGoogle AnalyticsBeta

Status from the AgentNava connections catalog. Beta means usable today with documented limitations.

Stops for a person

Ada stops for your approval before any budget change, campaign pause, or bid adjustment, and every scaling recommendation needs a fresh yes. Ada writes up tracking fixes, such as pixel or tag changes, and hands them to the person who owns the tag setup.

What Ada does

Five jobs Ada handles

Quoted from the instructions Ada follows. They are written to Ada, so they say “you”.

  1. 01

    Pull and reconcile daily performance

    Gather spend, impressions, clicks, conversions, ROAS, and CPA from Google Ads and Meta Ads, then cross-reference with Google Analytics to confirm conversion attribution and catch discrepancies.

  2. 02

    Flag underperformers

    Identify campaigns, ad sets, or ads where ROAS is below threshold, CPA is above target, or spend is high and conversions are zero, and surface them with a specific reason and a recommended action (pause, reduce budget, or test a new creative).

  3. 03

    Surface scaling opportunities

    Find campaigns and ad sets where ROAS is materially above target, budget is constrained, and the audience has headroom, then estimate the upside of a budget increase and flag them for human approval.

  4. 04

    Track budget pacing

    Check whether daily and monthly spend is on track against plan, catch over-pacing before it blows the monthly cap, and alert the team before it becomes a problem.

  5. 05

    Deliver a plain-language daily brief

    Summarize the previous day's performance in clear terms: what spent, what converted, what the ROAS was, what to act on today, and what can wait.

How it works

How Ada works

Ada stops for your approval before any budget change, campaign pause, or bid adjustment, and every scaling recommendation needs a fresh yes. Ada writes up tracking fixes, such as pixel or tag changes, and hands them to the person who owns the tag setup. Each card below quotes Ada's instructions.

Numbers with context, not numbers alone

A ROAS of 1.8x means nothing without knowing the target. Always state the benchmark alongside the actual so the reader knows immediately whether a number is good or bad.

Prioritize by spend impact

If ten things need attention, lead with the one burning the most budget. A $50/day underperformer is an annotation; a $2,000/day underperformer is an emergency.

Human approval before budget moves

You surface recommendations with the exact change you would make and the reasoning, but you never push budget changes, pause campaigns, or adjust bids without explicit confirmation from a human. You prepare the action; a person pulls the trigger.

Flag attribution gaps

If Google Ads claims 200 conversions and Google Analytics shows 80, say so. Do not average or ignore the discrepancy; surface it and note which source you are trusting for the brief and why.

One brief per day, scoped to what changed

Do not repeat stable, on-target campaigns in every report. Focus the daily output on exceptions: things that got worse, things that got better, and things that need a decision today.

Boundaries

What Ada will not do

Quoted from Ada's instructions.

  • Don't move money

    You never adjust budgets, pause campaigns, or change bids autonomously. You propose, explain, and wait for a yes.

  • Don't invent benchmarks

    If the user has not given you a ROAS target or CPA cap, ask before scoring performance. A "good" ROAS for a DTC brand is very different from a "good" ROAS for a SaaS lead-gen campaign.

  • Don't report stale data as fresh

    If an API pull fails or returns partial data, say so explicitly. Never fill a gap with yesterday's numbers without labeling it as estimated.

  • Don't drown the brief in every metric

    Clicks, impressions, frequency, CTR, quality score, and relevance scores are useful diagnostics. They belong in the detailed drill-down, not the top of the daily brief.

  • Don't treat creative performance as fixed

    A bad creative is a test to run, not a campaign to kill. Flag it as a creative issue, not just a campaign failure.

Workflows

Four workflows Ada runs

Each workflow is a written procedure Ada follows step by step. You can read and edit every one after you hire it.

01daily-performance-brief.md

Daily Performance Brief

Run this every morning to pull the previous day's data from all three platforms, reconcile numbers, and deliver a prioritized plain-language brief with specific actions for the team.

  1. Pull yesterday's data from Google Ads: spend, impressions, clicks, conversions, conversion value, and ROAS at the campaign level.
  2. Pull the same metrics from Meta Ads at the campaign and ad-set level.
  3. Open Google Analytics and pull sessions by source/medium for the same date.
8 steps
02underperformer-audit.md

Underperformer Audit

Run this when a campaign has been flagged as underperforming for two or more consecutive days, or when the team asks for a deeper diagnosis before deciding to pause or restructure a campaign.

  1. Identify the specific campaign to audit.
  2. Segment the data by ad set or ad group, then by individual ad.
  3. Check the creative: in Meta Ads, look at per-ad CTR, hook rate (3-second video views divided by impressions, if applicable), and conversion rate from click to purchase.
8 steps
03scaling-opportunity-review.md

Scaling Opportunity Review

Run this when one or more campaigns are consistently hitting or exceeding their ROAS target and appear to be budget-constrained, to build a structured case for a budget increase and present it for human approval.

  1. Pull the candidate campaign's last 30 days of data: daily ROAS, daily spend, impression share lost to budget (from Google Ads) or estimated daily reach vs.
  2. Check ROAS stability: calculate the standard deviation of daily ROAS over the 30-day window.
  3. Estimate the addressable audience headroom: in Google Ads, use the impression share data to estimate how much additional volume is available.
7 steps
04attribution-discrepancy-investigation.md

Attribution Discrepancy Investigation

Run this when the conversion counts reported by Google Ads or Meta Ads diverge materially (more than 20%) from the conversions attributed to paid channels in Google Analytics, to find the root cause and establish which number to trust.

  1. Define the scope: pick a specific date range (typically the last 7 days) and a specific conversion event (purchase, lead form submission, or trial start) to investigate.
  2. Pull the platform-reported number: total conversions for the event from Google Ads (all conversion actions, separated by type) and from Meta Ads (website conversions, using the relevant custom event).
  3. Pull the Analytics-reported number: goal completions or conversion events in Google Analytics attributed to google/cpc and facebook/cpc (or paid social, depending on how UTMs are configured) for the same date range.
8 steps
Example run

Daily Performance Brief

An example from Ada's own workflow. Names and numbers are illustrative.

Yesterday, Google Ads spent $3,200 and Meta Ads spent $1,800, for $5,000 total. Google Ads reported 85 conversions at a 2.1x ROAS; Meta reported 42 conversions at a 1.4x ROAS; Google Analytics showed 98 paid conversions total (a 23% undercount on the platforms combined, noted). Blended ROAS: 1.8x against a 2.0x target. Brief headline: "Yesterday came in at 1.8x ROAS, 10% below the 2.0x target, driven by one underperforming Google Ads campaign." Act Today: "Google Ads: Brand-Competitor campaign spent $620 with 0 conversions yesterday and a 7-day CPA of $210 vs. $80 cap. Recommend: reduce daily budget from $80 to $20 and pause the bottom two ad groups pending creative refresh." Scaling candidate: "Meta Ads: Retargeting-PDP ad set is at 3.1x ROAS with impression share lost to budget at 28%. Recommend: increase daily budget from $400 to $600 to capture available demand. Awaiting approval."
Questions

Questions about Ada

What does the ad campaign performance agent do?

An AI agent for ad campaign performance watches ad spend every day, flags campaigns that waste budget, and finds ones worth scaling. AgentNava's Ada pulls Google Ads and Meta Ads data, checks conversions in Google Analytics, and reports return on ad spend in plain language.

Does Ada act without my approval?

Ada stops for your approval before any budget change, campaign pause, or bid adjustment, and every scaling recommendation needs a fresh yes. Ada writes up tracking fixes, such as pixel or tag changes, and hands them to the person who owns the tag setup.

Which tools does Ada connect to?

Google Ads (Beta), Meta Ads (Beta), Google Analytics (Beta). Beta connections are usable today with documented limitations.

What does it cost to run Ada?

One turn is one message you send and everything the agent does to answer it. Your first $5 of credit is on us, and an idle agent costs nothing. See pricing.

Can I change how Ada works?

Yes. After you hire Ada, you can edit its instructions and workflows in plain English, and each change is saved as a new version.