ByteByteGo
Founder · Alex Xu
Beat the incumbent agency head-to-head and won the account.
ByteByteGo's system-design newsletter for engineers, with over a million technical readers. Launched under hard brand constraints, went head-to-head with the incumbent agency, and won. They had a head start. It didn't help.
The brief
ByteByteGo is a system-design newsletter for software engineers, a developer-education brand with over a million technical readers. I launched paid acquisition under strict brand guardrails and went head-to-head with the incumbent agency.
What I did
- Launched with strict brand guardrails, uploaded buyer lists for lookalikes, navigated a US+CAN vs ROW split.
- Ran concept + copy across Confession, Stop X/Start Y, X-but-for-Y, UGC-podcast and cheat-sheet lead-magnet angles.
- Restructured the account into a BROAD + LEADS setup to cut cost and improve efficiency.
- Co-introduced cost-per-engaged-sub (CPES) and proved we beat the incumbent on the metric that maps to product value.
What I own here
The full loop, not a slice of it. On this account I run:
- 01 Creative strategy & execution
- 02 Media buying
- 03 Funnel performance & analysis
- 04 Client-facing meetings
- 05 Weekly reports
- 06 Biweekly analysis & reports
How the winning ad got made
Everyone says they have a process. Here's mine actually running, on the ad that helped take this account from the incumbent agency.
- 01
Steal the audience's words
Before a single dollar moves, I mine Reddit, Quora and the client's own readers for the exact language they use. For ByteByteGo: how engineers really talk about system design, interviews and keeping up. That becomes the message bank every line is pulled from.
- 02
Write like the brand, sell like direct response
One concept, five variations: same headline and CTA, a different promise each, all in the brand's voice. The selling instinct comes from my own direct-response years, when my copy either converted or I didn't get paid.
- 03
Buy against CAC, not vibes
Launched into a BROAD + LEADS structure and judged on cost per engaged subscriber, not clicks. Scaled the winner, killed the rest, and beat the agency we replaced on the metric that maps to product value.
- 04
Turn the win into a machine
Then the whole flow became a Claude Code skill: it researches, writes the variations, builds the Canva design and the designer brief. The next winner takes hours instead of days. I automate my own job so I can do more of it.
The ad that won the account Some ads I built
