shishiQ: Questions in regards to white label
Hi, I looking at tier 4 - (white label) When using a personal API key, what are the additional costs for AI usage?
Is it possible for clients to add their own Claude API key?
Is it possible to put limits on chats per client?
Is the suggested $300/month resell price based on including AI costs or excluding?
Does the AI need to be trained before selling to a client, and what is the estimated time to do so?
Is the sales closing success rate dependent on how the AI is trained, or are there other factors on DMchamp's side that drive results and make this more predictable?
I read there is a sales closing rate of 15% — what are the factors this number is based on?
Asking to understand how to sell this in an honest and transparent way.
sohaib.dmchamp
May 28, 2026A: Hey shishi, great questions and love that you're asking how to sell this honestly. Quick answers:
On AI costs with your own Claude key, you pay Anthropic directly at their wholesale rates, and our prompt caching is applied to your key as well, which cuts the real cost per chat a lot. Most BYOK users see 40 to 80 percent savings vs the platform credits, depending on volume. That said, our new MAX tier is actually cheaper than BYOK in most cases, but since it's our own model there's no BYOK option for it.
Yes, clients (sub-accounts) can add their own Claude API key. Each sub-account has its own BYOK setup, so you can either run them all under your agency key or have each client bring their own.
Yes, you can set chat limits per client. Sub-accounts use feature flags that let you cap things like AI responses per chat, monthly credits, number of channels, context size, etc. You decide what each tier of your offer includes.
On the $300/month resell, that number is just an example. What you charge is totally up to you and depends on how you package it. If you go BYOK per client, you exclude AI costs (they pay Anthropic directly). If you sell credits as part of the package, you include them.
On training before selling, the guided onboarding scrapes the client's website and auto-generates the bot instructions, FAQs, and company info in a few minutes. From there, the recommended workflow is to spend 15 to 30 minutes refining the playbook and testing in our chat playground. You don't need to "train" a model in the ML sense, you're configuring a playbook and feeding it FAQs.
On closing rate, it depends on a mix: the quality of the playbook, the FAQ coverage, the channel (WhatsApp typically converts better than website chat), the lead source quality, follow-up sequences being enabled, response speed, and how clearly the client's offer is positioned. The platform handles the infrastructure, follow-ups, booking, and reasoning, but the playbook and lead quality drive most of the variance.
On the 15 percent figure, that's what our best-performing customers see in specific niches (inbound on WhatsApp with a tight playbook and live calendar booking). It's not a universal guarantee. A cold SMS list with a thin playbook won't hit that.