Nupact detects, thinks, and acts across search, social, and programmatic. Each change is drafted with the evidence attached and applies at the autonomy level you set.
Add a second product line and four more countries and the account outgrows the morning check. The big campaigns get looked at, the rest get sampled, and drift in the unwatched part surfaces weeks later in the numbers.
You operate across platforms but can’t see performance in one place. Every comparison means export, reconciliation, manual assembly. Decisions get made on incomplete pictures, and nobody can prove what’s incremental.
Log into one platform to adjust bids. Switch to another to reallocate budget. Reconcile pacing in a spreadsheet. Each action is small, they compound, and execution that should take minutes takes hours out of your best marketer’s week.
Nupact unifies your data into one layer purpose-built for agents to analyse and act on.
Six workflows, automated end to end
Anomaly · Google Ads · Search-DE. Clicks today on pace for 40% below a normal Monday, at an unchanged CPC. The shared €2,400 budget is on pace to run out. No other established campaign or market is off its baseline.
| Search-DE, today | baseline | on pace | delta |
|---|---|---|---|
| clicks | 2,400 | 1,450 | −40% |
| cost | €2,280 | €1,380 | −40% |
| CPC | €0.95 | €0.95 | 0% |
| conversions | 96 | 58 | −40% |
Ruled out: bids and competitors (CPC flat), demand (budget spent in full), tracking (feed fresh). Correction drafted: Search-DE-Generic on its own €300 test budget, the shared €2,400 back to Search-DE, DE budgets €2,700 a day while the test runs. Conversion value at risk across DE: about €2,210 a day.
Google published Monday’s impression share: Search-DE 41% against a 28-day level of 68%, lost to budget 2% → 31%, lost to rank 30% → 28%. Monday’s diagnosis holds.
Detection runs within minutes of a threshold breach, against a per-market baseline, on every account and every dimension. Monday’s drop was in Slack with its cause at 14:40, a week before the weekly report would have shown it.
Threshold breaches reach the channels your team already lives in. No dashboard to keep checking.
| Channel · campaign | Budget | Daily | Status |
|---|---|---|---|
| Google · Search-DE “restock” | €13,600 | €324 | Proposed |
| Meta · Restock-Prospecting-DE/AT | €16,800 | €400 | Proposed |
| Meta · Retargeting-DE | €6,200 | €148 | Proposed |
| TikTok · Spark-DE | €8,400 | €200 | Proposed |
| Google · PMax-AT | €3,000 | €71 | Proposed |
Structure, budgets and targeting built from the brief, gaps flagged, everything held until you approve.
Nupact drafts the move and holds it. Approve, adjust, or snooze; budget moves, bid changes and pacing corrections apply the moment you click. Bye bye, Meta Ads Manager.
Every Meta test ad is read each day at every step of the funnel, against your account’s recent winners, matched on landing page and format where possible. Only clear evidence stops an ad or moves it to scale. Everything else keeps testing.
At a German DTC brand on Meta, half of all stops came within a day of launch. The verdicts would have saved 12% of the test spend, net. How the backtest works
Nupact watches each scaled ad for fatigue and drafts the brief for a recut or a replacement. Winners go to the next market with pixel, catalogue and copy remapped. New assets are tagged on arrival, so the next brief starts from what won. Creative fatigue
| When | Action | Trigger & reasoning | Conf. | Reviewer | Outcome | |
|---|---|---|---|---|---|---|
| 08:41 | Meta · Prospecting-DE daily €400 → €520 | CPL 22% under target for 6 days | 88% | A. Weber | Approved | Revert |
| 03:12 | Google · Search-FR daily €300 → €260 | Pacing 31% ahead of the weekly plan | 94% | System · guardrail G-2 | Autonomous | Revert |
| Yesterday | TikTok · Spark-DE rotation 3 ads → 5 ads | Frequency 6.1, CTR down 34% in 14 days | 76% | M. Rossi | Pending | Review |
| Yesterday | Meta · Prospecting-DE audience 1% LAL → 3% LAL | Rejected: creatives not ready for colder traffic | 62% | A. Weber | Rejected | — |
The full history of what changed, when, and why, rejected proposals included. Queryable, exportable, permanent, a Revert control on every row.
Stale feeds pause automation on their own and page the team. Silent breakage is where paid media loses money.
Conversion feed stale for 6h (Meta CAPI). Autonomous actions paused themselves and the team was paged in Slack. The machine stops itself before it can optimize on a broken signal.
Nupact acts, then watches what the action did. The outcome feeds the next recommendation, so month six is sharper than week one. The change log is what fuels it.
The context stays. Nupact's memory belongs to the company: every decision and what came of it survives handovers and agency switches, so the account never starts over with its next owner.
A change goes live.
The result comes back.
The account’s history grows.
The next proposal starts sharper.
| Channel | Spend | Conversions · platform-reported | Conversions · your tracking | Gap | CPA · reconciled |
|---|---|---|---|---|---|
| Google Ads | €412,300 | 9,140 | 7,890 | +16% | €52 |
| Meta | €389,750 | 11,230 | 8,410 | +34% | €46 |
| TikTok | €96,200 | 3,020 | 2,140 | +41% | €45 |
| The Trade Desk | €171,400 | 2,410 | 2,060 | +17% | €83 |
Platform-reported and reconciled numbers sit side by side; the gap stays visible in every report.
Every channel, ad server and tracking source reconciled. Each platform’s self-grading shown against your own data.
Nupact designs and reads the geo holdouts and counterfactuals on its own, so the channel decision survives the budget meeting. How the tests run
Enriched conversions returned via Meta CAPI and Google Enhanced Conversions, with dedup, match-rate monitoring and a dry-run first.
Choose the level per campaign and per action type. Promote the routine actions when the proposals keep being right; keep the sensitive ones on approval forever.
Nupact watches and reports. Every signal surfaced with its evidence, nothing touched. A complete deployment on its own, and where most teams start.
Nupact prepares the move and holds it for your click. This is where most teams run day to day.
Nupact acts inside the guardrails you set. Every action lands in the change log with its reasoning, reversible from the log.
Built and hosted in the EU. Your data is hosted in the EU, aligned with GDPR and the EU AI Act, with every action audited.
A system can run your incrementality tests, but someone has to decide which question is worth the forgone sales, and no system builds the creative engine that keeps TikTok fed. Those calls need someone who has made them before.
Nupact brings in hand-picked senior operators and writes their brief from your live data before the first call, so an expert’s first hour feels like their second month. If this bench doesn’t exist in your team, this is how you get one without hiring it.
Attribution and incrementality, plus the tracking that quietly stopped reporting.
Multi-market Google, Meta and programmatic, run by operators with a record of growing revenue from ad spend.
Testing frameworks and production velocity, for a channel that needs volume to work.
An interim head of performance who owns the number while you hire.
Connect your ad platforms through their official sign-in. The first account audit lands the same day: budget leaking into stale assets and broken tracking, each with the fix drafted.
Nupact learns your accounts’ normal: pacing patterns, seasonality, market structure, your KPI definitions.
Alerts and digests start flowing. Promote actions to propose-and-approve whenever you’re ready, per campaign, per action type.
Bring one live account and we walk through what changes when the control layer sits on top of it. Pilots start from that call, scoped on live spend, with a direct line to the people building the product.