All updates

The Decision Engine Stops Guessing

The Adstudio assistant answering a cost-per-conversion question: Google Ads and TikTok analyzed, Meta flagged as needing a reconnect, a validated finding in Turkish lira, and an unresolved hypothesis that produces an investigation instead of a task

Adstudio now reads every campaign through its own objective, reports money in the account's real currency, and compiles its platform expertise straight from the decision contracts instead of a hand-copied summary of them. Most of this release is about the difference between knowing something and sounding like you do.

Objective-First Analysis

  • Every Campaign Judged By Its Own Goal: A Reach campaign is no longer marked broken for not optimizing for purchases. Adstudio checks each campaign against the platform's own objective/goal compatibility matrix, so an Awareness or Traffic setup reads as valid instead of critical
  • Conversions That Match the Objective: Meta results are reported per action type — leads for lead campaigns, purchases for sales campaigns — instead of collapsing every objective into one conversion number
  • Goal Mismatch, Not Misconfiguration: If the goal you state disagrees with the campaign's declared objective, Adstudio says exactly that, rather than telling you the campaign is set up wrong
  • Honest Tables: Currency, date windows and unsupported metrics are labeled in the table itself — an empty cell now says why it is empty

Claims the Assistant Can Actually Support

  • Validated, Unresolved, Rejected: Every hypothesis carries its validation status. A task can only come from a validated hypothesis; an unresolved one produces a named next investigation instead of a confident recommendation
  • Confidence In Words: Numeric confidence scores are gone. Adstudio states high, medium, low or unknown, with the reason attached — a number implied a precision that was never there
  • Counted Is Not Examined: An audit that fetched sixty campaigns and reviewed three now says so, instead of reporting "no issues across 60 campaigns"
  • One Platform Failing Does Not Erase the Rest: A dead Meta token used to discard a finished Google audit and then suggest reconnecting the platform that was working. Failures are now scoped to the platform they happened on, and the answer comes back partial with the gap stated

Money in the Right Currency

  • Your Billing Currency, Everywhere: Spend, cost and value are read from the ad account's billing currency on Google, Meta and TikTok. A TRY-billed account is never rendered in dollars again
  • No Cached Guesses: Adstudio no longer caches an assumed currency, so a wrong guess cannot outlive the request that made it

Platform Expertise, Compiled

  • Seven Google Campaign Types: Shopping, Demand Gen, Display and App join Search, Performance Max and Video — every type carries its own expert checks and guardrails
  • Six Meta Campaign Types, 26 Checks: Meta's shared and campaign-specific checks are compiled with their guardrail wording intact, in the vocabulary the contracts define — creative contribution is measured, fatigue stays a hypothesis
  • No Invented Thresholds: Fixed 7-day and 30-day verdict boundaries have been removed. Comparison windows adapt to the account's own history rather than a number nobody chose
  • Faster Deep Analysis: Deep analysis and the account audit no longer hold every benchmark window in memory at once, so large accounts finish instead of stalling