Automatically RECOVER Failed Payments from Agency Clients Now (Exact AI System) — Transcript
Full transcript
- 0:00In the last couple of years, our failed
- 0:01payment recovery system has recovered
- 0:03over $270,000
- 0:05of revenue. Our recovery rate is almost
- 0:0760%. I think yours will probably be
- 0:09better than that, a lot better than
- 0:10that, because a lot of that is from
- 0:12people signing up, watching my YouTube
- 0:13videos, signing up to ReviewHarbor
- 0:15online, and then just canceling with
- 0:17fake cards and stuff like that. So,
- 0:18yours will probably be a lot higher than
- 0:19that. Mine used to be a lot higher
- 0:20before I started making YouTube videos.
- 0:22But, today I'm going to go over the
- 0:23exact failed payment recovery system
- 0:24that we've implemented in the last
- 0:25couple of weeks that has drastically
- 0:27improved our failed payment recovery
- 0:29rate, with our clients, etc. And so, if
- 0:31you don't know who I am, my name is Clay
- 0:32Lawrence. You can see last month, in
- 0:33June, we did over $57,000 in recurring
- 0:36revenue just helping local business
- 0:37owners get Google reviews and rank
- 0:39higher on Google. And so, the vast
- 0:40majority of these are low-ticket
- 0:42recurring revenue subscriptions, $100 to
- 0:43$300, $400 a month. When you have
- 0:46hundreds and hundreds of subscriptions,
- 0:47stuff happens. Cards fail, you know,
- 0:49Stripe can't process all the payments.
- 0:50Don't believe me, you can look up
- 0:51reviewharbor.com, you can look at our
- 0:53case studies, whatever. This is what we
- 0:54do, right? Failed payment recovery is
- 0:56absolutely huge. If you're just getting
- 0:57started, this probably isn't very
- 0:58important. As soon as you start to get a
- 1:00decent amount of customers paying you
- 1:01that sticky low-ticket recurring revenue
- 1:03that is just so juicy, it's so important
- 1:05to figure out how you can keep that
- 1:06revenue, especially if they're still
- 1:07happy. And so, I'm going to go over kind
- 1:09of our old process, what it looked like,
- 1:11and now how we're using AI to make this
- 1:13process 10 times better and way more
- 1:15automated. If you scroll down here, our
- 1:17old system here was made up of kind of
- 1:19complicated workflows that were super
- 1:21rigid and didn't have a lot of
- 1:22flexibility in there because it didn't
- 1:23have like human reasoning or AI
- 1:25reasoning to know what to do next. So,
- 1:27basically the structure was this. We had
- 1:28our Zapier account integrated with
- 1:30Stripe, right? And so, when a failed
- 1:31payment happened in Stripe, what it we
- 1:33would do is we would update the client's
- 1:35payment status in HighLevel. So, we'd
- 1:36add a tag to that client in HighLevel,
- 1:38and we'd say, "Hey, this client's
- 1:39payment recently failed." That in itself
- 1:41could be complicated because we had to
- 1:42make sure that the email in Stripe
- 1:44matched up with the email in HighLevel.
- 1:45There was no reasoning there, right? The
- 1:47next thing is we'd send text messages
- 1:48and emails every few days to try and get
- 1:50them to update their payment method,
- 1:51right? We'd just send them the update
- 1:53link. And what would happen is is a
- 1:54couple things. If they responded to the
- 1:56workflow, and they said, "Hey, you know,
- 1:57I'm getting a new card. My card got
- 1:59stolen. You know, I should have a new
- 2:00card in the mail in 7 days." If they
- 2:02responded to that text or email, we
- 2:04didn't want to keep following up with
- 2:05them. So, we kind of had to turn on stop
- 2:07on response uh because we just get
- 2:09annoying if they texted back and said,
- 2:10"Hey, I'm getting a new card." And then
- 2:12every day they keep getting texts and
- 2:13emails. Also, this is a pretty
- 2:14vulnerable spot to be in the client
- 2:16relationship because right now they're
- 2:17technically not paying you, right? So,
- 2:19we want to make sure we still leave a
- 2:20good impression. On the other hand, we
- 2:22had to have a whole 'nother workflow
- 2:23over here that when a new successful
- 2:25payment happened in Stripe, every single
- 2:27successful payment, we had to go match
- 2:29that and update all of our clients in
- 2:30HighLevel to say, "Hey, their payment
- 2:32succeeded." And then we would remove
- 2:34them from this other workflow
- 2:36automatically if that happened. And this
- 2:38workflow works, and it's fine. The
- 2:39problem is it's way too rigid. You're at
- 2:41this point in the customer journey where
- 2:43they're not paying you. Maybe their
- 2:44account's on the fritz or whatever, and
- 2:46you know, they haven't gotten the
- 2:46results they wanted, or maybe they're
- 2:48just, you know, lost their card. And now
- 2:49we're constantly bugging them and
- 2:50annoying them with these repeated
- 2:52reminders to try and update the card.
- 2:53So, what actually ended up happening is
- 2:56when they would respond, we would have
- 2:57to go set a reminder to follow up in 10
- 2:59days or 7 days when they did get their
- 3:01new card in the mail or whatever to
- 3:03update it. And it was just a manual
- 3:04headache. And then eventually we might
- 3:06call them and say, "Hey, you know, can
- 3:07we get this updated?" etc. But it was
- 3:08just a very manual process. In the
- 3:10background, if you don't know, like
- 3:11Stripe subscriptions will just keep
- 3:13retrying the card every couple days to
- 3:15see if it updates. And one thing I would
- 3:16recommend is go check out your Stripe
- 3:18settings and make sure you have like
- 3:19optimization turned on. I think it's
- 3:21called payment optimization or something
- 3:23like that. Where basically they have
- 3:24Stripe has partnerships with the main
- 3:26credit card companies. And if somebody's
- 3:28card just expires and they just get a
- 3:29new card in the mail automatically, that
- 3:32credit card company will send Stripe the
- 3:34new data of that credit card so they can
- 3:35just keep the subscription rolling and
- 3:37update it automatically. We keep trying
- 3:39the payments, and sometimes they'd go
- 3:40through and they wouldn't have to update
- 3:42it. But like I said, the cons of this is
- 3:43the clients would respond, and we'd
- 3:44still have to manually follow up.
- 3:45Clients would get annoyed cuz they'd
- 3:47already messaged us, and we kept trying
- 3:48their payments in Stripe or whatever it
- 3:49was. And there was no multi-channel
- 3:51context. This is so important and it
- 3:53sounds a little bit nerdy, but what that
- 3:55means is like, let's say for example, a
- 3:56client emails me or my wife or somebody
- 3:58on our team directly and says, "Hey, I
- 4:00saw that text about the card being
- 4:02updated. We'll get to it eventually."
- 4:04HighLevel has no context on that, right?
- 4:05This workflow that we built out has no
- 4:07context. And so, it just keeps messaging
- 4:09them. It's not able to check the Zoom
- 4:10calls or check the phone calls we've had
- 4:12with them or check other channels to see
- 4:14where they've already kind of handled
- 4:15it, right? And so, all the messages were
- 4:18generic give us money vibes. You know,
- 4:20it's just like the same message over and
- 4:21over again. And a lot of times, some of
- 4:22these businesses were repeat offenders.
- 4:24So, every couple months their card was
- 4:26failing. And so, they got the same
- 4:28messages over and over again. It's just
- 4:29kind of annoying, right? And so, the new
- 4:31workflow, we've kind of incorporated
- 4:33this tool I've talked about before on
- 4:34the channel, which is Victor. Victor is
- 4:36like an AI agent that plugs into your
- 4:37Slack. And what you can do with Victor
- 4:39is you can connect it to all your tools.
- 4:41So, you can see Victor is connected to
- 4:43our Intercom support processes. It's
- 4:45connected to Stripe. It's connected to
- 4:46our Gmail accounts. It's connected to
- 4:48our HighLevel accounts, our HighLevel
- 4:49Summit accounts. Connected to all these
- 4:51different places where it can get more
- 4:53context to know whether we should
- 4:54message them or not. And so, the new
- 4:56workflow is easier to set up and it's
- 4:58more effective. Like literally, at the
- 4:59end of this video, you could probably
- 5:00take the transcript, if you go to the
- 5:01description and grab the transcript, you
- 5:03could probably take the transcript of
- 5:04this video, give it to Victor or give it
- 5:06to your AI tool, and then you will be
- 5:08able to just set it up. You can just
- 5:10reason with it, go back and forth, and
- 5:11this will just be set up in the
- 5:12background. And so, Victor, our Slack AI
- 5:14agent, we just have a failed payments
- 5:16channel, and every day it runs a daily
- 5:18check through both our Stripe accounts
- 5:19to check for failed payments, right?
- 5:21Check for failed payments today,
- 5:22yesterday, and any recurring ones that
- 5:24have been going on the last couple
- 5:25weeks. From there, what Victor does
- 5:26before it messages or do does anything,
- 5:29it goes and checks all of our channels
- 5:31for context, right? So, it checks my
- 5:32email, Carly's email, who she does a lot
- 5:34of our support. It checks our support
- 5:36email. It checks our text messages
- 5:38through HighLevel. It checks our phone
- 5:39call recordings. It checks our Zoom
- 5:41recordings through Fathom. It checks our
- 5:42Intercom support channel. It checks all
- 5:44of those places to see, "Hey, has this
- 5:47client already let us know that they're
- 5:48working on it, right?" And if they've
- 5:50already messaged us yet, then we just
- 5:51can respond politely and say, "You
- 5:53awesome, no worries, looking forward to
- 5:54it." And then create a reminder based
- 5:57off of what they said to then continue
- 5:59to follow up, right? So, then it kind of
- 6:01puts them back into the follow-up
- 6:02process here. But, if they haven't
- 6:04messaged us yet, it'll send a very,
- 6:06very, very personalized text and email
- 6:08through directly through HighLevel.
- 6:10It'll grab results from their account.
- 6:12And so, we have a back-end database, but
- 6:14you can just connect Victor to all your
- 6:16client sub-accounts so it can have the
- 6:17data as to how many reviews they've
- 6:18gotten, what, you know, if you're
- 6:20providing leads, how many leads have you
- 6:21gotten them, all that sort of stuff. And
- 6:22these messages are super personalized.
- 6:24You can see like it says, "Hey, their
- 6:25name, quick heads-up, review us payment
- 6:27for their business name, uh didn't it go
- 6:28through? We've gotten you 78 reviews.
- 6:30We'd hate to see that stop. Probably
- 6:32just a card thing. Feel free to update
- 6:34it here." And then it sends the update
- 6:35card link, right? This is a real one
- 6:37that was sent out today. And so, from
- 6:40there, if there's still no luck, we
- 6:41still keep sending them personalized
- 6:43reminders up to 7 days when the very
- 6:45last thing it does is add a human in the
- 6:47loop and say, "Hey, maybe you should
- 6:48give this person a call." And so, that's
- 6:49the very last thing. It'll just notify
- 6:51us in that failed payments channel. It
- 6:53makes it so much easier. And as you can
- 6:55see from this graphic, these are real
- 6:56numbers that we pulled from our Stripe
- 6:57accounts. This has saved over $272,000
- 7:01of revenue for us in the past couple
- 7:03years. And if that seems high, well,
- 7:05well, think about it this way. If
- 7:06somebody's paying you $279 a month and
- 7:07their card doesn't go through, if you're
- 7:08able to get their card to go through
- 7:10again and they stay with you for another
- 7:11year, that's not $279 that you saved for
- 7:14that 1 month, thousands and thousands of
- 7:16dollars that you saved across the
- 7:17lifetime value of the customer. And so,
- 7:19it's super important, but the business
- 7:21model we run, which is low ticket
- 7:22recurring revenue, it can become an
- 7:24absolute headache. And so, literally, if
- 7:26you take a screenshot of what this looks
- 7:29like, put the transcript into your AI
- 7:31tool, whatever you want to use. If you
- 7:32want to use Victor, Victor's link will
- 7:34be down below in the description, and
- 7:35you can use that. Plugs directly into
- 7:36Stripe, HighLevel, your Slack, all of
- 7:38the things that you need to plug into.
- 7:40And now, it's just a much more personal
- 7:41vibe, right? We're not annoying our
- 7:43customers anymore because we're not
- 7:44sending them messages that they've
- 7:45already messaged us. It's more dynamic
- 7:47and less rigid than the workflows. The
- 7:49main reason you want to use AI in
- 7:51certain workflows is when you need very
- 7:52specific if this then that like human
- 7:55reasoning almost to figure out what the
- 7:56right next action is. And so this kind
- 7:59of takes that off our plate and now we
- 8:00don't have to ever worry about
- 8:01scheduling any sort of reminders. The
- 8:03only thing we might have to do is that
- 8:04if we haven't got it in 7 or 10 days,
- 8:06it'll send us a reminder saying, "Hey,
- 8:08this person might need a phone call."
- 8:10And that's all we have to do. Other than
- 8:11that, it's taking care of it. And so
- 8:13it's just this loop that it keeps going
- 8:14through making sure we can try this
- 8:16payment. And Stripe's still trying it in
- 8:18the background obviously, but a lot of
- 8:19times you might have to update that
- 8:20card. And so this is our failed payment
- 8:22recovery. I know it seems kind of like a
- 8:24boring subject, but it's super duper
- 8:26important as you grow as you scale to
- 8:28you know recover these payments from
- 8:30clients that are already paying you. I
- 8:31do this sometimes when like if we're at
- 8:33a certain revenue mark for the month,
- 8:34I'll go through all our failed payments
- 8:36and like try to retry them cuz it's like
- 8:37the easiest money that you can get if
- 8:39you want to get to that next payment
- 8:40mark because it's pretty much the
- 8:41easiest money you can get. You're
- 8:42already providing value. They already
- 8:44said they trusted you enough to pay you
- 8:45once or get put the card on file at
- 8:46least. So this has been really helpful
- 8:48for us at Review Harvest. Just want to
- 8:49share it. If you enjoyed this video,
- 8:50feel free to sign up with any links down
- 8:52below in the description. That'll
- 8:53support the channel. I'll keep making
- 8:54videos like this. And also if you sign
- 8:56up with my link to HighLevel down below,
- 8:57you'll get an extended trial. You'll get
- 8:58a link where you can copy and paste all
- 9:00my automations, all my stuff directly
- 9:01into your HighLevel account and learn
- 9:03about how I built this business to over
- 9:05$55,000 a month in recurring revenue. So
- 9:07if you like this one, check out the next
- 9:08one here.
About this transcript
This page contains the full transcript of Automatically RECOVER Failed Payments from Agency Clients Now (Exact AI System) by Clay Lawrence, generated from the public captions YouTube serves with the video. The transcript has 2,373 words across 340 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.
What you can do with it
Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.
Free YouTube transcript tool
YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.