Decision intelligence for Google & Meta ads
It tells you what to do.
Not just what happened.
Every morning it reads your ad accounts, finds the spend that isn’t working, and hands you the decision - with the evidence, and a chat box to argue with it.
Free to try. No card. Read-only access to your ad accounts.
"Prospecting - Broad" (Meta) reached 3.1x more people at 62% of the original CTR over 12 days
148 search terms took spend without converting - ₹21,400 of spend on them
Below the minimum volume the signal engine requires to call a pattern
- CTR trend · CTR declining 12% vs previous 4 days
- CPA trend · CPA rose 75% - severe cost drift vs previous 4 days
- Conversion rate · Conversion rate fell 44% - strong drop vs previous 4 days
- Spend stability · Spend consistent (CV 0.16) over 8 days
- Frequency impact · Avg freq 1.06x over last 8 days - creative is fresh to audience
Meta reports insufficient data to grade this creative right now - that is Meta declining to measure, not a low ranking. Withheld, not zero.
Where to scale, protect, or investigate next - actionable first.
- Instagram83%
- Facebook17%
- What's happening
- Hold the creative and cap delivery on "Prospecting - Broad" (Meta) to the audience that still responds.§1
- Why
- CTR decay tracks expanded delivery, not the asset - the response inside the original audience is flat.§2
- Evidence
- 3 supporting points§3
- Historical precedent
- Seen 3 times · acted on twice · improved both§3
- Record strength
- Medium
- Investigate next
- Confirm frequency inside the original audience before you widen delivery again.§5
A deterministic summary composed from the evidence below - no prediction. Full detail in the sections that follow.
Hold the creative. The account has reached an audience ceiling, and new creative does not fix an audience problem.
Delivery expanded 3.1x over 12 days while blended CTR fell 38%. Inside the original audience CTR is flat. The asset is not fatiguing; it is being shown to people it was never built for.
- · the original audience definition has not been edited inside the window
- · conversion tracking has fired continuously for 12 days
- · Reach up 3.1x over 12 days; blended CTR down 38%
- · Original-audience CTR flat at 1.24% across the same window
- · CPM up 4% - inside normal weekly variance for this account
Hold the creative. Cap delivery on “Prospecting - Broad” (Meta) to the audience that still responds.
Check first: confirm the audience definition is unchanged before you narrow delivery.
Expected impact: recover blended CTR toward the original audience’s 1.24% · Low effort - under 5 minutes in the ad platform
Right now, part of your budget is buying nothing.
You just can’t see which part.
A creative that stopped converting four days ago. Search terms taking spend with zero conversions. An audience that stopped delivering and nobody noticed. Ads Manager will show you all of it - in twelve reports you’d have to know to open. Datsutra opens them for you every night and only wakes you for what changes a decision.
The Brief
Open it in the morning and there is nothing to interpret
A freshness line so you know the data is current, then a short list of things that need your judgment today. Nothing else competes for the top of the screen.
It opens on a judgment, not a chart
Needs your attention, then a short list. Each item names the call, the campaign it applies to, and the evidence line behind it.
It tells you when it can't see
Stale data is flagged in red at the top - "Google Ads, 23 days behind" - instead of being quietly averaged into a number you'd have trusted.
Three ways out, none of them a dashboard
Open the investigation, defer it to a date, or dismiss it. Deferring and dismissing are recorded too - the record is the point.
Creative analysis
It knows which ad is dying before the numbers do
Every creative scored out of 100 and ranked, with the retention curve underneath it. Hover a card - the one it's watching lights up.
- CTR trend · CTR declining 12% vs previous 4 days
- CPA trend · CPA rose 75% - severe cost drift vs previous 4 days
- Conversion rate · Conversion rate fell 44% - strong drop vs previous 4 days
- Spend stability · Spend consistent (CV 0.16) over 8 days
- Frequency impact · Avg freq 1.06x over last 8 days - creative is fresh to audience
Meta reports insufficient data to grade this creative right now - that is Meta declining to measure, not a low ranking. Withheld, not zero.
A health score, not a leaderboard
Every creative is scored out of 100 from real CTR, CPA, spend-stability and frequency trends, and labelled Active, Learning or Fatiguing.
Where viewers actually drop
The retention curve from plays to 100%, so you know whether the hook failed or the middle did - and whether it's worth re-cutting.
When Meta won't grade it, it says so
"Insufficient data to grade" is reported as Meta declining to measure, not as a low score. Withheld is never rendered as zero.
Audience analysis
Scale this one. Protect that one. Don't touch the third yet.
Thirty-one audiences, each carrying a verdict rather than a row of numbers - and a stated confidence, so you know which verdicts to lean on.
Where to scale, protect, or investigate next - actionable first.
- Instagram83%
- Facebook17%
Scale, protect, or find out
Every audience carries a verdict and a confidence, not a row in a table. "Needs evidence" is a real answer and the tool uses it.
Composition you can act on
Age and gender spend share, devices, regions, interests, platform split, and the hours your results actually land in.
Reach that means people
Meta's deduplicated account figure, not the sum of per-audience reach - which double-counts anyone you reached twice.
Datsutra Copilot
Disagree with it. Out loud.
You already argue with ChatGPT when it’s wrong. Do the same here - except this one is looking at your actual account. Type into the box and it defends the call, cites the evidence, or changes its mind.
- Ask why it reached that conclusion.
- Ask what evidence it used, or what to test first.
- Tell it something about your business it should factor in.
- Commit, challenge, or defer - all three are recorded.
Press Challenge it in the panel to see one of those exchanges. An advisor that has to defend its answer is worth more than one that only asserts it.
- What's happening
- Hold the creative and cap delivery on "Prospecting - Broad" (Meta) to the audience that still responds.§1
- Why
- CTR decay tracks expanded delivery, not the asset - the response inside the original audience is flat.§2
- Evidence
- 3 supporting points§3
- Historical precedent
- Seen 3 times · acted on twice · improved both§3
- Record strength
- Medium
- Investigate next
- Confirm frequency inside the original audience before you widen delivery again.§5
A deterministic summary composed from the evidence below - no prediction. Full detail in the sections that follow.
Hold the creative. The account has reached an audience ceiling, and new creative does not fix an audience problem.
Delivery expanded 3.1x over 12 days while blended CTR fell 38%. Inside the original audience CTR is flat. The asset is not fatiguing; it is being shown to people it was never built for.
- · the original audience definition has not been edited inside the window
- · conversion tracking has fired continuously for 12 days
- · Reach up 3.1x over 12 days; blended CTR down 38%
- · Original-audience CTR flat at 1.24% across the same window
- · CPM up 4% - inside normal weekly variance for this account
Hold the creative. Cap delivery on “Prospecting - Broad” (Meta) to the audience that still responds.
Check first: confirm the audience definition is unchanged before you narrow delivery.
Expected impact: recover blended CTR toward the original audience’s 1.24% · Low effort - under 5 minutes in the ad platform
Trend analysis
Every number against the period before it
Campaigns, creatives and audiences, always read against the previous window of the same length - as a table when you’re scanning, as a trend line when you’re arguing about direction. A figure with no comparison isn’t a finding, so the tool won’t show you one.
Google and Meta stay on separate cards. Never blended. A combined ROAS is a comfortable number that hides which platform is carrying the account.
| Campaign | Spend | Conv. | CPA | vs prev. |
|---|---|---|---|---|
| Prospecting - Broad | ₹4,86,000 | 214 | ₹2,271 | +34% |
| Retargeting - 30d | ₹1,92,000 | 186 | ₹1,032 | -8% |
| Search - Brand | ₹64,000 | 121 | ₹529 | -3% |
Decision memory
It grades its own calls, including the ones it got wrong
Confidence is stamped at the moment you commit. The verdict arrives when the review window matures - Improved, Worse, Neutral or Inconclusive - measured against a snapshot taken before you acted.
More than one account
Running several brands? The portfolio is the whole point
Four brands you own or fourteen you're paid to run - the failure is identical: the account nobody opened this week is the one quietly losing money. Datsutra reads all of them every day and ranks by what's at stake.
- ROAS
- 2.41x
- Monthly target
- ₹35L
- Last sync
- 1h ago
- Open decisions
- 11
- Status
- Critical
One portfolio, every brand in it
How many need attention, how many decisions are open, how fresh each connection is. The account nobody opened this week is the one the portfolio raises first.
Assign an owner, keep the record
Each brand has someone accountable and a decision history that survives handover. When a marketer leaves, the reasoning doesn't leave with them.
One inbox across all of them
Every open decision in one table - client, type, severity, priority, platform - so triage happens once instead of eleven times.
The Client Portal
Coming soonYour client sees the work, not the workings
A read-only space carrying your logo and your colours. Your client signs in and sees how their business is performing and what has been done on their account this month. Only what you publish appears there.
- Weekly and monthly reports assemble themselves from the record you have already built.
- Review, approve, publish. Download as PDF or CSV.
- Cases, signals and internal reasoning never cross into it - the portal is a presentation layer, not a window into your workspace.
- 2 SepHeld the top creative and capped delivery to the responding audience
Response inside the core audience was still strong - the fall was in expansion, so we protected the winner instead of replacing it.
- 26 AugRebuilt the prospecting structure around three intent tiers
Budget now follows demonstrated intent rather than sitting flat across the account.
- 19 AugPaused two underperforming placements, moved budget to search
Search was clearing cost per acquisition at roughly half the blended rate.
How it works
Three steps, then it runs without you
Connect the accounts
Authorise Google Ads or Meta, then point each discovered ad account at the brand it belongs to. Read access only.
It investigates overnight
Signals detected, cases created, and only what survives scrutiny reaches you. Everything else stays in the log.
You decide - it remembers
Commit, challenge or defer. Every decision is recorded with the confidence it carried, then graded when the review window matures.
Where it stops
What Datsutra refuses to do
An advisor you can trust is defined by what it won't do under pressure.
It won't spend your money
Datsutra proposes; you commit. It never touches a budget, pauses a campaign, or executes anything on its own.
It won't manufacture certainty
When the data is too thin to conclude, it says so and makes no recommendation - instead of dressing a guess up as an answer.
It won't blend platforms to look tidy
Google and Meta are shown separately, always. A blended ROAS hides which platform is actually carrying the account.
Who it’s for
One account or forty, the job is the same
From ₹8,000/month
One client account included; add more as you grow. Founding offer: 50% off, locked for 6 months.
Connect an ad account tonight.
Read the brief tomorrow.
Free to try. No card. Google Ads and Meta, read access only.