The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Back to feedPost
I almost let the agent guess at pricing.
Early on I let it answer “how much” questions with a range pulled from training data. Wrong move pricing changes, and a wrong number erodes trust fast. Now anything price-related routes straight to a human, every time.
→ Some questions are worth the extra step. Pricing is one of them.
AI Chatbot DevelopmentBusiness Workflow AutomationConversational AI DevelopmentElevenLabsRetell AiVapi
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Challenges
View all
Related posts
Two years ago, I was charging less than a grand for a website. This year, I’ve shipped identity systems for companies at their most critical funding rounds.
Same hands. Same tools.
The only thing that changed is I stopped punishing myself for being fast. When you bill hourly, you’re literally charging less as you get better.
I stopped selling "time" and started pricing for what a successful launch or a successful rebrand is worth to the company. Positioning isn't just what you say to clients. It’s what you charge them.
Charging hourly only makes sense in two cases:
1. When you're starting out (can't properly scope a project)
2. For revisions
Your A/B test said "winner" — but did revenue per visitor move?
CVR is the metric everyone watches. It's easy to track and moves fast. But CVR going up doesn't mean the variant made you more money.
I've seen checkout tweaks that lifted CVR 8% while AOV dropped enough to net a 6% loss in revenue per visitor. The variant "won" — and got shipped — because no guardrail was in the readout.
Before you call a winner😍 → Check revenue per visitor alongside CVR
→ Run a power check — underpowered tests call false winners ~30-40% of the time
→ If you tested multiple metrics, your p-values need correction (Benjamini-Hochberg)
The stat is only as useful as what you're measuring.
Demo AI automation workflow designed to help businesses capture, qualify, organize, and follow up with leads automatically. The workflow connects forms, AI, Google Sheets, Gmail, and team notifications to reduce manual work and improve lead response time.
