Why not just use social/Google's reported ROAS?

Platform ROAS counts "conversions" they define, which often includes page views, add-to-carts, or phantom signups from iOS 14+ modeled data. Tycoon Agent uses real Stripe charges only, so her ROAS is typically 30-50% lower than what social tells you — and accurate.

What attribution window does she use?

30 days by default — long enough to catch SaaS trial-to-paid conversion, short enough to be reactive. You can ask for 7-day or 90-day if your sales cycle is different.

What if a customer touches multiple ads before buying?

Tycoon Agent uses last-touch attribution by default (matches what GA4 reports), but can run a multi-touch view on request — first-touch, linear, or position-based. The recommendations don't change much for early-stage businesses.

Can she actually pause the campaigns or just tell me to?

Both. She drafts the change and asks for confirmation. On approval, she pauses via the platform APIs and posts confirmation to Lark. You can also keep her in advisory mode if you want to execute manually.

Ask Tycoon Agent: What's my paid ads ROAS this month?

You'd think this needs a marketing analyst and a multi-touch attribution stack — Tycoon Agent has the answer with real Stripe revenue, not platform-reported "conversions."

The short answer

Tycoon Agent answers "what's my paid ads ROAS this month" by pulling spend from Google Ads, social ads Manager, and LinkedIn Campaign Manager, then matching it against Stripe revenue tagged with the same UTM campaign IDs. She doesn't trust the platforms' reported conversions — those count fluffy events like "page view." Instead she joins ad clicks → GA4 sessions → Stripe customer creation → 30-day revenue, then divides revenue by spend per campaign. The output is a Lark table: campaign name, channel, spend MTD, revenue MTD, ROAS, and a recommendation (scale, hold, kill). She also flags any campaign with high spend and zero attributed revenue — the silent budget killers. You stop trusting the social network's optimistic dashboard and start making real allocation decisions.

How Tycoon Agent actually does it

Pull spend from each ad platform
Google Ads, social ads Manager, LinkedIn Campaign Manager via Composio connectors — month-to-date spend by campaign.
ComposioGA4
Match clicks to Stripe customers
Join ad UTM tags to GA4 sessions, then to Stripe customers by client_id — 30-day attribution window.
StripeGA4
Calculate true ROAS
Revenue from real Stripe charges ÷ ad spend, per campaign. Ignore platform-reported conversions.
Bucket by recommendation
ROAS > 3x → scale. 1-3x → hold and optimize. <1x or zero attribution → kill or pause.
Send to Lark
One-screen summary: total MTD spend, blended ROAS, top 3 winners, top 3 losers, suggested budget shifts.
Lark

What it looks like in chat

YouWhat's my paid ads ROAS this month?

Tycoon AgentMTD as of today: $8,420 spent, $14,230 attributed Stripe revenue → blended **1.69x ROAS**. **Scale (3):** Google Search "ai api gateway" → 4.2x ROAS, $1.8k spend. LinkedIn "founders" lookalike → 3.8x. social retargeting → 3.1x. **Hold (2):** Google brand defense (1.4x but defensive). YouTube pre-roll (1.6x, slow build). **Kill (4):** social "AI tools" interest broadcast → $1,240 spend, $0 revenue, 28 days running. Google display network → 0.3x. LinkedIn job titles "CTO" → 0.6x. TikTok creator promo → 0.0x, 14 days. **Suggested shift:** Pause the 4 losers ($1,840/mo saved), reallocate to Google Search winner. Projected blended ROAS at month end: 2.4x. Want me to draft the campaign pause + budget shift?

What you get back

Lark message with MTD spend, real Stripe-attributed revenue, blended ROAS, per-campaign recommendation, and a suggested budget reallocation you can execute today.

Cadence: Weekly Monday + on-demand.

What's my paid ads ROAS this month?