THE PROBLEM
Every Monday morning needed Slack pings to four people.
Triple Whale showed one CAC. The Glew dashboard showed a different one. The accountant's P&L showed a third because of timing differences with refunds. The COO ended up emailing four people every Monday to ask 'so which week was good?'
Cash flow visibility was a sheet the analyst updated weekly. Real cash position was a guess.
WHAT WE BUILT
One database. One set of definitions. One screen.
Week 1: piped Shopify, Stripe, Meta and Google Ads into BigQuery on a daily schedule. Sat with the accountant for an hour to agree what 'CAC' actually means.
Week 2: built the dashboard. Per-brand views, consolidated view, SKU profit after ads + returns + refunds — the part Triple Whale doesn't show you.
Week 3: cash forecast. 90 days projected from receivables, recurring spend and inventory POs already placed. The COO calls it 'the runway timer'.
Week 4: Slack alert layer. Four thresholds, one channel.
I cancelled Triple Whale and the Glew dashboard the day we shipped. I still pay for Klaviyo. Now I argue with my own numbers, not three vendors'.
THE OUTCOME
Three subscriptions cancelled, one dashboard the founder trusts.
- SaaS dashboards cancelledTriple Whale + Glew$1,250/mo cut
- Slack pings to figure out 'how was last week'4 → 0−100%
- Cash forecast horizonnext Friday → 90 days+13×
- Metric definitions in dispute3 → 1−
WHAT WAS MESSY
What broke in the first three weeks.
Refund timing was the hardest part. Shopify counts refunds the day they're issued; Stripe counts them the day they settle. The accountant's P&L counted neither. We added a 'refund pending' state to the cockpit so all three views reconcile.
Cash forecast first version overshot by 11% because we hadn't modeled return rate seasonality. Two-week calibration fix.
WHAT'S NEXT
Same warehouse now powers an inventory reorder engine and a custom returns dashboard.
Because everything's in BigQuery, every downstream tool became cheaper. The same data feeds a returns dashboard (which products refund most, by SKU and channel) and an inventory reorder engine that the COO is testing now.