A
?
AssetsWorkflowsHandbookDesign system
← Handbook

Wins and failures

Alex's framing: wins pool into case studies, failures compile into an issue-management guide every AM learns from. The second half is the rare part. Agencies keep a testimonials folder and throw the failures away, so every AM rediscovers the same problem alone.

What came in

56 rows from a case-studies sheet: client, industry, services, testimonial, proof, case study note. 55 imported, 9 matching current portal clients and 46 belonging to past clients — kept by name with a null client_id, because a foreign key would have discarded exactly the history the exercise exists to preserve. "We have run ads for 46 interior firms" is itself the asset.

Three failures were already logged in there, honestly: two landing pages that "didn't perform well though" and a website that was "incomplete though". Each carries a lesson stub rather than an invented takeaway — the honest state is *someone who was there needs to write this*.

The material worth mining

Buried in prose, now queryable:

  • automation took reply rates from under 10% to almost 70% (Dary Design)
  • $500–$1,000 CAC against a $30,000–$50,000 ticket, 30%+ appointment show-ups (Assemble Concept)
  • 90% foreign-worker leads distilled to almost none (Design IOI) — a blacklisting case study
  • 300k views across IG and TikTok from a brand new account (M2 Decor)
  • "From Last Legs to $400K in 30 Days" (Renobook, already public)

The shape

client_outcomes: kind (win/failure), headline, detail, service, industry, metric_before, metric_after, evidence as an array of typed references, case_state, and — failures only — a lesson. Metrics are free text because the source says "less than 10%" and "almost 70%", and rounding those into numbers would be inventing precision the original never had.

case_state runs raw → has_proof → usable → published, so "what can we actually put in a deck on Monday" is a query rather than an argument. Currently 11 usable, 10 with proof, 34 raw.

One thing the concentration says

45 of 55 outcomes are Interior Design. That is either the moat or the risk, and it is worth naming out loud rather than discovering during a downturn.