What’s happening isn’t actually n8n expiring your connection it’s Google. When you set up OAuth for a self-hosted app, Google puts you in what’s called a “testing” publishing status by default. In that mode, any refresh tokens they issue are only valid for seven days. That’s why your connection keeps dropping after a week. Since you mentioned your project is already set to production, there are two other things worth double-checking: 1. Verify your app in Google Cloud Console: Even if you switch to production, you may still be restricted if the OAuth consent screen hasn’t been fully verified and published. Once verified, Google issues long-lived refresh tokens instead of the temporary 7-day ones. 2. Redirect URI consistency: Make sure the redirect URI in your Google Cloud OAuth client settings matches exactly what n8n is sending. A tiny mismatch (extra slash, wrong domain, http vs https) will cause token issues after refresh. 3. Credential overwrite option (for self-hosted setups): If you’re worried about OAuth resets, n8n supports [credential overwrites via environment variables or REST API]. That way you can control client ID/secret centrally and avoid accidental mismatches. So the real fix here is getting your Google app verified and locked into production mode with proper redirect URIs. Once that’s done, refresh tokens should last indefinitely (until revoked manually or by inactivity).

For us, the method that's been most effective and scalable has been cold DMs blasts. I know that might sound basic, but here's why it works so well. We can reach thousands of prospects in minutes, not hours like with cold calling. It costs almost nothing to run, and the response rates are actually 39% higher than email. Plus, people check their texts immediately. The process is pretty simple. We scrape leads from Google and public directories, import them into our CRM system, and then send them through automated Email workflows. The key is having really targeted lists and messaging that doesn't sound like spam. We've been able to do thousands of outreach messages in a matter of minutes with this approach. I've also been testing LinkedIn outreach, but honestly it's much more limited in scale. LinkedIn caps you at around 20-30 connection requests per day even with premium accounts, so it's better for higher-value prospects. But when we do use it, we focus on building genuine conversations rather than pitching immediately. The key there is having a solid profile that shows credibility when people look you up. What I've learned is that most people fail not because they picked the wrong channel, but because they're not consistent enough with any single channel. They'll try LinkedIn for a week, then jump to cold email, then try Facebook ads. Pick one method, master it completely, then scale it before moving to the next. The other thing that's been huge for us is having our positioning dialed in. We're not selling "AI chatbots" or "automation." We're selling complete business transformation systems that solve specific problems. When your positioning is right, the channel becomes less important because people actually want what you're offering. Referrals are obviously gold when you can get them, but you can't build a predictable business on referrals alone. You need a system that consistently brings in new prospects, which is why we focus so heavily on the SMS outreach. It's predictable, scalable, and we can measure everything.

What you want to do is use the IF node. That’s the built-in way to add conditional logic to your workflow. You set up the IF node with a condition for example, “if a field equals X” or “if a number is greater than Y” and then connect its two outputs: one branch for when the condition is true, and another for when it’s false. From there, you just build the different paths your workflow should take depending on the outcome. If the logic you need is more complex than the standard comparisons in the IF node, you can always drop in a Function node (or Code node) and write a little JavaScript to handle the decision-making. But most of the time, the IF node gets the job done quickly without needing any code. So in practice, it looks like this: Trigger (or input node) → IF node → branch A if the condition is true → branch B if it’s false. That’s all there is to it, and you can chain multiple IF nodes together if you need more layers of decision-making.