Email and SMS Revenue Forecasting with Daniel Guerra

Why Projections Fail

0:00We pull up either last month's revenue or last year's revenue, you tack on whatever you're trying to grow, whatever your growth goals are, right? Your boss, your CEO, whoever tells you, "We need to be up 10% this year, 20% this year." You tack that on, compare it with what you did last year, submit the number you got, your top projections. That's just hoping and wishing that you can actually nail without any strategy. And the real cost of bad projections is actually it's not the number, it's the conversation you have to have after when you don't hit that number, and then you're like, "how far I was off. Sorry, I didn't actually adjust for this." And it's the awkward conversation you have to have with either your client, your boss, CEO, investors, you name it, whoever you're in the room with, and they're trying to figure out, "Hey, we missed goal, but if we have a system in place, we can actually arrive at to run it for 40 plus brands is so hard if I don't have access

0:42to like quad that's just running it. I couldn't do it, man. And remember that AI is a pattern recognition tool, and you are the judge man. You're the decision maker.

Football Story Setup

1:01You've seen the show, watched the movie, or you just heard it from actual football. Cool. It's all about you. so, on on side for about 4 to 5 months out of the year, I actually coach high school. and I'm going to introduce myself to this and I promise to tie it together kind of. but I'm going to start with a story. So, as a high school football coach, I started doing about 5 years ago.

1:27So, on our second year, cross-town rival might as well show up. Tough game. End up winning. You know, got the W. Awesome. Next year, same thing. You drive out. We think we're ready, right? Like as a team, you go to prepared. You know that their star quarterback graduated, their running back is out for the season, he's got an injury. Should be an easy win considering how you beat them last year and they lost key pieces. First quarter, they get the ball.

1:56They go down and score. All right, no problem. That happens. Okay, that's not a big deal. They get the ball right back, score again. By the time it's halftime, it's 46 to 6. We get smacked. We thought we had

Inputs Make Strategy

2:11The reason why I say that it is I am going to choose myself to coach any of you in the next four years of the football team. But what I want to do is talk about the fact that every Friday night my job is to project what our opponents are going to do and adjust our strategy based on a bunch of inputs. All right, we're going to call these things inputs and they feed into a system. All right, those inputs for football could be film study, personal injury reports, etc. right? They help me figure out how we can do and project out our performance without An incomplete scouting report means a wrong game and a wrong game and a wrong rap. So that is the speak.

2:47How does that play into retentions? Well, I'm also Daniel, how to data out of your standard copyright, the data guy or I can numbers guy. I talk about four weeks to the last four weeks. we For those of you that don't know, UC Anner runs our retention, purely retention drop list for 89% of your brands. Some of our brands are down here. One of our brands you see right here, Rockwell. Nice. So yeah, and I run a version of these projections for four weeks plus the model runs I write for agencies. All right, what I want to do is walk through what my system is and teach you guys how to build a system for your own specific business, brand, whatever it is that you do.

3:29Hopefully it applies to some of you or most.

Retention Projection Gap

3:34So I'm curious to see who raised their willing share. Do any of you currently run projections for your email and SMS for the follow-up for the next month? do you predict how much you're going to actually make your brand? Curious. One, two, I saw Melanie, yep. Maybe a few of you. Yep. I think most it's one of the most like untapped things that email and SMS retention marketers forget is projecting following months performances. Acquisition does this for every single month broken down by the week, by the day, and they had it probably through 2027 all the way through like Q4. And for some reason as email marketers and SMS marketers, we don't do that. And I don't know why. Maybe it's hard.

4:20So we're going to talk about that. Does this sound familiar? This is what I see when when we take over an account when I speak to other marketers. Usually what happens when we throw out projections. We talk about you open up your DSD. Right? And you pull up either last month's revenue or last year's revenue. You tack on whatever you're trying to grow, whatever your growth goal is, all right? Your boss, your CEO, whoever tells you we need to be up 10% this year, 20% this year. You tack that on, compare it with what you did last year, and you submit the number you got your top projections. That's what we're going to do. Email SMS, that's what your retention program has to be for for you this month.

4:51Cool? That's awesome. That's not it. That's right? That's just hoping and wishing that you can actually nail without any strategy. And the real cost of that projections is actually it's not the number. It's the conversation you have to have after when you don't hit that number. And then you're like, "Oh, it's how far I was off. Sorry, I didn't actually adjust for this." And it's the awkward conversation you have to have with either your client, your boss, CEO, investors, you name it. Wherever you're in the room with and they're trying to figure out, "Hey, we missed goal." And they're usually blaming the acquisition first, and then they come to you and they're like, "Well, you know, we sent the traffic." And it's all a mess. It's the bad conversations you have to have, right?

5:33So here's an example. I'll give you guys

Client Crisis Example

5:35an an exact example. This happened in March of this year. We had a client, I'm not going to name them obviously. They were down 43% just in March. We took over the program in December. Great month, January. Good month, February. great month, March, down 42%. And suddenly they're like, "We want you guys out." Whoa. I didn't even see that coming, right? Revenue's down 42% and I had zero idea why. No clue. I'm not joking. We hop on like three calls and we're like, "I don't know what's going on. Like we projected we're going to hit these goals. Let's figure it out together, right?" And they're like, "You guys need to rebuild your strategy cuz whatever you're doing is not working."

6:13But it wasn't strategy. It was actually we were projecting off bad numbers. Not only were we projecting off bad numbers, they actually increased pricing significantly the month before and they didn't even let us know. Two things that actually deeply affect the way that you project your performance for following month. And the issue was actually that we were were projecting using last year's revenue for March. Last year's revenue actually had a massive discrepancy within Clavio, which is our ESP, and it was over inflating a number that wasn't actually accurate. And so we were projecting that we could grow on a month that was inaccurate and we had no idea.

6:46So what we did is we adjusted the projections based on this discrepancy that we found. We retained the client. Now they're a client, you know, we're still with them. The problem was the problem with B is if you don't have a system to track down your projecting on, then you're just guessing and just shooting in the dark and probably jumping in a hero or pulling up whatever your ESP's reporting platforms.

The Projection System

7:07So how do we fix bad projections? It's actually super easy. Actually, this is my favorite slide. It's super easy. You just have to account for revenue for recipient and then deliverability. And then you got to go to segment health, SMS overlap, pricing changes, and there's list growth, inventory, paid acquisition, seasonality, campaign gigs, promotional linkages, and any negative flows that you may have added along the way. Yeah, I mean that's it, right? The playbook. it feels like chaos is an international system of bliss, right?

7:40And that's the general way that last slide, that's what it feels like since when to not bring in the numbers. Right? But if you have a system in place, we can actually arrive at it. Right? So, I'm going to walk through these. I'm going to go step by step. And it's not about how I do things. My My entire conversation is not about like, "Oh, this is how Daniel at this one agency runs it for what his specific, you know, ICP or brands or whatever." It's It's a system that you can take back and you figure out how that works for you and your stage of your life cycle or whatever you do for your business.

8:10That's the whole point. Right? It's not not me trying to teach you this is how

Step One Baselines

8:14you do it. It's this is how you set it up for yourself. So, this is the easiest up to skip. You got to establish your baselines. You don't know where you're going if you don't know where you've been, my classic saying. So, one is where are you measuring your numbers? That's actually one of the first conversation we have with with a client is, "Hey guys, if you want to be successful for you, we have to know where are you looking at your numbers?"

8:36If you don't know where you're looking at your numbers, where you're going to measure your success, you have no idea where you're trying to go and then therefore your projections are going to be inaccurate cuz you don't even know where you've been there. So, establish your baselines. Figure out your attribution settings. What's your percentage of attributed revenue? Right? I wrote a few examples here. It's going to be different depending on the brand, right? Subscription brands were 81.

8:55it's going to be vastly different than Ruggable, who don't offer necessarily a subscription program for us, but very important that we get repeat purchases. That's not as easy for us. and then you got to establish your goals, right? Where are for growth revenue targets? What's our channel mix? All these things super important, and I mean it. Write it down on a piece of paper if that's what it takes your first time through. I know most of you think you already have all of this written out somewhere in a sheet. Put it down all in one sheet. If you don't have a single source of reference, it's all going to be jumbled cuz I'm about to walk you through the next slide of where everything that moves the numbers. It's a messy slide. I actually made this slide because there's so many things that affect the numbers, but you need all these things. And so I wrote them down. Jimmy, I don't know if you guys are distributing these afterwards, but in theory, right? They're here. And I

Collecting Inputs Tiers

9:44I've broken down to three three tiers. Okay? One is readily accessible data. And I'm going to talk about like the Klaviyo or MCP. It doesn't have to be that, right? Whatever Whatever ESP you're in probably has the ability to pull these numbers and they're readily accessible for you right off the bat. Easier things like your 12-month campaign revenue, these three things like your RPE, broken down or blast 120 days, campaign types. If you're not tagging your campaigns, you should start. If you didn't, you can retroactively export them and import them to whatever whatever it is that you're going to be using.

10:17Okay? So, this is the second step that that I I'm just collecting and organizing your inputs. So, all we're doing right now is we're going to collect all our potential inputs and then we're going to organize them. Tier one, those are readily accessible. Most of these I no longer pull in tier one. They're just there. I just like reconnect or whatever I do with my, you know, platform MCP and I go, "Pull these things." And then it does it. It's awesome. Tier two, this is contextual data you do have to go get. I promise you it's not that easy.

10:43But once you do it and once you figure out where the data lives, it becomes so much easier to do it month after month. Eventually, you can run it to where you're actually able to pull it automatically. Okay? So, these are things like paid acquisition traffic or how much they're going to spend. Yeah, I have a couple brands that use Domo, other ones use Airtable, other ones use sheets, other ones use different things, right? Where their acquisition team tracks these things. I have to go through, figure out where where do these numbers live and as soon as I get access to them, I'm pushing them over into whatever my AI system that I'm using is, because that's an important thing that I'm going to need to use I need to have to run my projections.

11:17Then, you have things that you're all probably touching on day-to-day basis, right? The planned promotional calendars, how many emails you're going to send, what changes you're going to do to your flows. All those things that you have access to, feed them into it. The last thing that I want to talk about is tier three, which is the thing that is actually my last point in this slide right here. It's refined for feedback. Okay, that's step six. So, that's actually what I call tier three, it's retroactive recalibration. Right? So, at some point things are going to go wrong, your numbers are going to be bad. Okay? And we're going to talk about what happens there. But, what you have to do is go back, take those numbers, figure out what went wrong, and tell it that it's wrong and why it went wrong. That's actually the third tier and most important, in my opinion, of like getting actual projections, is going back and telling your AI system, right?

12:05Because all of this is only possible with AI. I'm I'm just going to say. Like, to run it for 40 plus brands is so hard if I don't have access to like Claw that's just running it. I couldn't do it, man. So, this is all AI talk, I promise.

AI Prompting Demo

12:19Okay, so here's what what it actually looks like, right? So, this is three prompt to AI, okay? Super simple, right? I use a super simple prompt right here. I plug in the MCP, I typed in an historical prompt, and I said, "Hey, analyze the last four months data for campaign and flow revenue. Create a calendar projecting maximal revenue for campaigns, flows, and SMS based on historical performance." Super simple, but it's connected to an account, right? Boom, it's actually a calendar. That calendar tells me, "Oh, here's how much you're projected to make here, and here's how much in campaigns and flows and blah blah blah."

12:46That's awesome. That's a really badass looking calendar that's actually contextually blind, and when I deliver that to my boss or my client, whoever it is, they'll be like, "Okay, I'm going to hold you to that $7.76 million you told me you could make." I have no idea if it actually is going to hit $7.6 million.

Judgment Over AI

13:04Okay, so this is step four, analyze the output. This I actually I'm going to shout out a friend, her name's Casey, she said this at one of her talks, and it like hit me like a ton of bricks, so I decided I'm going to, you know, pass that over you guys cuz it's the way that we have to think of things when we're talking about analyzing AI. You have to draw a line. Draw a line in the sand, and remember that AI is a pattern recognition tool, and you are the judge man, right? You make You're the decision maker.

13:31And sometimes the issue is we're allowing AI to be the decision maker and we're flipping the script and we're going, "Well, AI told me so, it must be right." The judgment line is remembering that AI produces the numbers and you decide if they're right. Okay? You make the call. So, throw the inputs in, see what it spits out, see if it's even close to what you've even done historically or if the numbers are even close to accurate, and then you make the judgment call of what's necessary and what other inputs you need to address to make sure that you capture your data.

14:05I coach this all the time. It's one of my favorite things to say. We do this every single week. I We talk about it and you don't like the outputs? Change the inputs. Set some, right? We talked about inputs. It's a thing I coach that applies to marketing just as much. All right. So, here's like tier two. Now I added like the first layer of things and then I'm I'm adding in those additional inputs, right? Like the acquisition traffic, seasonality, RPE trends, acquisition stone by day, traffic per day for the last 180 days. These are all inputs and there's a little bit hard to see, okay?

14:37But now, the difference is it has a revenue goal or a total projected goal. So, now I'm actually telling it what it should what we should be hitting. Then it's projecting how much we're going to actually do and if we're going to be above goal or we're we're pacing to where we're not a great stone by day, but now we're actually inputting like if you're able to see this, there's like a you know, Memorial Day sale. So, now it's actually taking in further inputs and adjusting its projections to understand, okay, there's actually other things that I need to account for as this AI system in order for it to actually project more accurate numbers.

15:08Cool. This is actually defendable. Right? Like once you give it is all this input like tier one and tier two, they're actually very defendable num- numbers. I've used a version of these for my clients where it actually gets like within a variance of that's acceptable of what we can actually hit by day. Break it down by day for the month. And if I know I'm going to miss goal, we'll either, you know, throw on another campaign, run another sale, we'll present this to the client. We know that we're going to be off and we can actually tell them, "Hey, this is so much we think we can help you guys make based on your traffic, based on what we're spending, etc."

Feedback Loop Refinement

15:41Step six, we talked about it. I already mentioned it, right? It's refining through feedback. Oops. I'd say the gold's in the gap. Okay, so there's a gap from like projected to what it what you actually did, and there's like gold in there. That's like where you really get like the Pull that out. You basically go back and you tell it, "This is what went wrong. This is how much we actually made. This is why." We'll talk about like basically, you need to iterate on this.

16:08First, you have the projection, then you get the actual, and then there's this gap, and then you got to like input more things, right? You are the input now. The retroactive input that needs to feed it back in. And same thing. Coach Coaching Winery, failure isn't a flaw in the system, it's the system working. Failure is not problem, right? Like, I'm not saying you guys are going to nail projections for those of you who are actually going to go back and try to do this on your own.

16:33Like, you're probably not. Trust me, I sometimes don't, and then we meet with the client, and we beat up more information, and then we go back, and we're like, "Okay, that feels more accurate." Failure isn't a flaw in the system, it's the system working. So, this is care for you, right? This is actually like I did the mild version. Once you guys get to the slides, you'll see this is the most accurate version of a projection that I've yet to This is like, "Hey, this is what went wrong, and I we actually swapped the promo rate from 20% to 25% because we were underperforming. The brand knew what I had made that call.

16:57Paid acquisition, you know, met like meta spend was down by 55K for the month. That's just what the client decided to do. These are the things that affected our initial projection. This is why they're wrong. Can you account for these in the future?" Projections are not a deliverable. They're feedback.

Wrap Up Coaching Analogy

17:15So, to bring it back, right? You're all coaches. You game planned, you adjust at halftime. It's simple. Adjusting at halftime is what keeps you from getting whooped 46 to 6. Cuz it doesn't feel good. Find your baseline, you collect every input you can possibly get your hands on. If you do it once and you spend time doing it once, you can figure out how to do it again over and over again. You prompt your system intelligently, you analyze what comes back, if you use your own judgment, you don't let AI tell you whether you're right or wrong, you tell AI whether it's right or wrong.

17:45Then you assess what month what the month actually taught you, and then you feed all that in, right? You just tell it, "Yeah, you got it right, you got it wrong. Here's what you did wrong and why why you got it wrong." That's it. It's a system that you can repeat month over month and you can actually get projections for, in my opinion, a channel that's heavily undervalued in terms of your ability to determine or understand what you can actually drive for the brand that you're working with.

18:13Do you have any questions? >> You You cooked on this, Gus. That was fire. Fire puppies out here. Arab and Berzel Berzel would never touch the stage. >> See, it does like the connection of fight. A man that knows who you do. And then a man that knows how to educate in the proper way, and then also doesn't make us sitting in the stage feel like, "Oh, I'm really like that my job." Cuz all the stuff you're doing same it's like H to K for lazy. That was really really well done. what's even cooler is the fact that the talk right after you is like on super technical and so he play rights very. So, I'm going to lead this question with the group right now is their people will come back. They'll They'll They'll go back to the rooms that they'll find with somebody give me for game though. Yeah, epic. We can go

Audience Q&A Takeaways

19:05back and like and try and figure out, "Where do I start?" You What What is a piece of the information I could take outside of this deck in here? Can you give us the zero through 0.5 before I go to the one? >> Yeah, my biggest recommendation is like you actually don't know how many inputs you actually have access to. I guarantee most marketers don't understand. Like we have to try and go hunt it down, right?

19:25We have to try and go like, "Hey Rugable, like can we get access to your like paid acquisition spend?" And they're like, "I don't know, you're a retention agency, do you really need to see that?" Right? Go back and just sit down and figure out like, "Okay, what are all the potential things I can feed my system with?" And just write them down. You don't even have to like go pull You just have to go figure out where they live, right?

19:45That's like my know That's my zero step, like what are all the potential inputs and go figure out if you can actually go get the The answer is yes, like I promise you the prompt thing is actually a piece of cake once you nail it, right? Like now it's just here's all the inputs. Here's what I'm trying to do. So just the zero, which is actually the hardest thing, is figuring out how many how many inputs you can get your hands on and then going just writing down where they live.

20:07>> There was a couple of acquisition teams here or somebody that might be like a marketing general manager, like should I in-house this? Should I build it? I know the cloud plus the MCP is is valuable. Shoot. What you're maybe to like the opposite of what Cloud two questions in this line. I'm an acquisition team I will will clients I'm trying to extend LTVs to sell retention. Are they connecting the into the P and creating their own projects? Should I live with this iPod? Should I live with this iPod version? I see fire.

20:35>> I smiled because everyone's got a different method like nobody knows anything about AI. Like everybody here says like, "I 100% know I run this." It's like we started like three months ago, we had projects for every single client and then the moment I with the cloud I used cloud. And then the moment we I tried to run the same analysis for another brand, I was like, "Oh, it's stuck within a project and I got to re-prompt it from the very beginning."

20:55So I'm like, "Scrap all the projects, I'm not doing that anymore." Right? We had like 50 plus projects and every strategist had their own. And then the moment I'm like, "Go learn from that other thing." All that's to say, like don't don't broilers. just kidding. If it works for you, it works for you. But >> [laughter] >> Yeah, that's the thing about it. I have to see if it's how you do do this. Yeah, but like there's always a use case where it actually like does make sense, right? Like where it's super important that you you know make sure. So It's it's hard to answer like your question is like four acquisition teams.

21:27I actually question again, I want to make sure I saw it. Well, I'm I'm an acquisition, so it's really all of you. >> Right. I'm like, yo, I'm good at acquisition. >> Yeah. I don't want to get into retention. Yeah. >> I have five or six clients. I'm going to try to do your retention. I'm going to connect I'm going to get you access to the cloud you're collecting the MCP my one by one. >> Yeah, there there's a few people I don't know where else you actually have figured out how to bypass it and get all the MCPs connected all by one so they can like pull order.

21:49>> This topic theory you got that you got that bro. >> Okay, no. >> You got the other person. >> Yeah, so >> So there are like we haven't quite nailed it because of the volume applies and then like our biggest fear and I'll be honest with you like our biggest fear is because of the level of like brands that we have. The moment that it like spits out someone else's bread. Like if someone else gets like any level of like AG1 data like liability like huge issue, right? So like we do keep things separate by not connecting with what MCPs. I know a few people who have figured it out how to have them all connected at once.

22:22I still don't know that I'm confident in it cuz yet you know there's like a lot of people are running new things in Codex that like I just learned about like last week. I would keep them separate for now. And then like it's an annoying switch up, but I would still research.

Tools Data And Sanitation

22:34>> Are there any other supplementary BI tools and is there anything else you use like data Zappy? >> Yeah, there's definitely it just direct question. Sorry, he's asking if there's any supplementary BI tools or any other like >> data Zappy? >> Yeah, data sanitation tools that we're using. So it's all so dependent on the brand. Like I so much of it is like we are genuinely dependent on what the brand wants to use and is comfortable using. So then we'll have their own, right? Like we do have brands that just have never even touched our data, right? Like they've never even connected to anything, right? And they're like still reporting purely out of what like your your baby monitors and your ESP selling you. Then you have brands that have like their own Tableau dashboards that are like rented like Caraway.

23:13They are data geniuses. Like those guys are crushing it, right? And they have their own Marina who runs their accounts here. She's my counterpart on at New Standard. She will be the first to tell you like the moment she asks her something, it's connected. They run their own data sanitation. They have it all. So we're very dependent on that. Otherwise it becomes hyper resource intensive for us to figure out like how to run it for every >> Sorry. So then you're just here's my case base and exactly in-house the bogey. Yeah, I mean >> Yep, exactly.

23:40we do have like one or two people on our team who are like also like data people who will go through and like do like their own version sanitation, but we're not using any like crazy tools.

Final Thanks

23:53>> Anyone else? >> Thanks for listening to my like incredibly niche conversation.