550 million views in 365 days. 1.5M new subscribers. 3.2M watch hours. $9,537 in estimated revenue from the channel alone, not counting what the inbound leads generated on top of that. This is one client account running on the same system the whole time. No viral tricks. No paid promotion. Just content going out consistently while they focused on their actual work. The part that stands out to me most is the subscriber growth, 188% more than the previous year, because that's compounding audience that keeps growing even when nothing new is posted. Curious if anyone else here has seen a similar compounding effect when content runs consistently over a long period versus sporadic posting?

Hey everyone 👋 Over the past few weeks I built and deployed an AI reporting system for a digital marketing agency client. They use it now, and they’ve already asked for a second experiment platform to be added. What it does: One plain-language prompt pulls A/B test results from GrowthBook, writes a verdict and builds a client-ready deck in Google Slides, then saves it to Drive. There’s also a chat interface for follow-up questions about an experiment. Stack: FastAPI, Make.com, Groq, Google Cloud Run, and Postgres for per-thread conversation memory. 3 things I learned: 1. I didn’t use an autonomous agent for the core flow. The tool calls run in a fixed order and the LLM only writes up the results. For client-facing reports, I wanted predictable output over clever behavior. The “agentic” part is limited to the chat follow-ups. 2. Pull data from the source, don’t paste it. Numbers come from the API, so nothing gets mistyped or misread before a client sees it. That’s the real answer to “why not just use ChatGPT?” 3. Hybrid beats pure no-code or pure code. Make.com handles the glue and the delivery, and custom code handles the parts that need reliability and fallbacks. What’s next: adding Convert as a second source, then a demo video. Question for the group: I’m now looking to sell this to other digital marketing agencies, and I’d love your advice from your own experience. My plan so far: a demo video on LinkedIn, then personalized DMs to agency owners and heads of growth, then a free 15-minute audit call. I’m targeting agencies of roughly 5-30 people who run experiments for clients. For those of you who’ve sold to agencies: - What worked best for getting the first few conversations? - Do agencies buy a finished product, or a custom build per client? - How did you price it, a one-time fee, a monthly retainer, or both?

You want a successful AI agency. Clients who pay you well. Work you’re proud of. The freedom to choose how you live. How long have you wanted that? And how much has actually changed in that time? You’ve watched the videos. Saved the posts. Tried the tools. But you’re still unsure what to sell, who to sell it to, or how to get someone to pay you. Another month passes. You’re still saying, “I’m working on it.” Let's get real for a second. Last week, how many business owners did you speak to? How many follow-ups did you send? Did you put a single offer in front of anyone who could buy it? Or did you spend the week preparing? Getting everything "perfect" before going to market. Changing your website. Testing a new tool. Rewriting your offer for the 5th time. Each one gives you something to do without risking a no. But a business owner can’t buy an offer you never show them. And you can’t learn why your pitch falls flat if you never make it. At some point, you have to put your work in front of someone who can say yes or no. You also need to look at how you’ve set up the environment around you. Who reviews your sales calls? Who tells you when your offer is unclear? Who notices when you spend another week planning instead of speaking to buyers? If the answer is nobody, you’re left to spot every mistake yourself. That can keep you stuck for far longer than you need to be. Find people who will look at your work and tell you the truth. People who can show you where you lost the buyer. Help you fix a weak proposal. Ask if you sent the follow-up you said you would. Then give yourself time to act on that feedback. Put your phone away. Pick the task you’ve been avoiding. Finish it. You don’t need to figure out your whole future today. You do need to stop letting the same week repeat. How much longer do you want to keep saying, “I’m working on it”? If you’re ready to take the next step, comment "ACTION" below. I’ll DM you and set you up with someone on my team to talk through where you’re stuck and whether we can help

I am making an AI Automation where I need to make a summary for a report and submit it via PPT to the end user so I have made it but it's not accurate report generation and I have tried the prompting technique but it's not working but I have noticed that the data is not completely labeled like there are some metrics that are unknown from the platform so it cannot actually understand it entirely so my question now is that how do I tend to that like what options could I have one is to drop of those metrics from the experiment (it's from the field of digital marketing) or shall I ask the client to actually help me identify those metrics or what more options do I have For refrence I have structured the data for the server side of the job it's just the unlabeld part of it that is the issue here I have made it using make.com and fastapi server (via code) it's a hybrid setup workflow

Hey guys I have been meaning to ask you guys that does technical events like AWS community day or dev days or KCD helps with agency promotion or not and I am also a student so I can learn there too ! So either way it would be a win-win for me but I wanna know that if it does help then how can I get the maximum out of it !!