I've spent tens of thousands of $$$ with @gelato and mostly it's been great
But now they've delayed 100+ orders
There's no indication anything will arrive by Christmas
They quietly shorted the shipping window by a week w/ no announcement
And I'm stuck in "ai support bot hell"
storybrand X hero's journey
- a character the customer can relate to
- put them in a hole "it hurts"
- make it worse "in fact it was killing me"
- get them out of the hole "this product/service saved me"
- happily ever after
- cta: if you're in that hole, get this
also created a lil shell helper called `did()` which adds annotations to my posthog analytics
easy/dumb way to see what happened near interesting moments in the charts
i've broken my checkout too many times
so i've just rolled out playwright
every time i deploy, it runs a test purchase using the airwallex demo card in my shiny new staging env
db latency from @vercel to @prisma accelerate to @supabase, before and after localising
vercel in us-east-1
prisma is everywhere (cf anycast)
before: supabase in ap-southeast-2
p50 = 1.2s, p99 = 4.4s
after: supabase in us-east-1
p50 = .23s, p99 = 1.9s
Dark patterns for dealing with meta - a brainstorm
1. Send conversion values reduced by 50% to pixel and capi. See if metaโs skimming algorithm adjusts
2. Every time I get a sale, increase ad budget by 10% immediately. Any hour without a sale, decrease it
What else?
@SamMendelsohnW6 I have had thoughts like these and dismissed them as sour grapes.
But maybe Iโll change my pixel & capi to send meta conversion values of 50% the real value and see if my CPA goes down ๐
Results of my headlines & desires test
Looking primarily at CTR, side-eye on CPC
โ got a shortlist for further testing
Interesting to see the low-CPC & low-CPM outlier top the charts ๐
- cbo with 'autoshutoff' in name
- automated rule to kill any ad with >$3 spent and campaign name contains 'autoshutoff'
- one adset, usual audience suggestion (no restrictions)
- one ad per image
- daily budget $150
all ads should be done in a couple days, ready to analyze
today i'm probing market desires... figuring out what do ppl click on most
running $3 each in ad spend to 55 ugly AF ads
only diff is the headline
using traffic goal instead of sales ๐ณ๐คฏ
bc i want fast signal about interests & desires
don't care about readiness to buy rn
2/2
- export images
- create bulk import csv with same headline and primary text for all rows
- use facebook ads 'bulk import ads' tool in new cbo campaign
next: campaign setup