With consumer confidence at record-low levels, only one thing’s for sure: you’ll have to pay more to acquire new customers. But how do you forecast this all-important metric heading into the uncertainty of 2023? This week, Richard and (mostly) Taylor talk with CTC Ecommerce Data Analyst Steve Rekuc about building — and outperforming — your 2023 CAC forecast.
- The Ecommerce Playbook mailbag is open — email us at firstname.lastname@example.org to ask us any questions you might have about the world of ecomm.
- Want more data? Subscribe to the DTC Index for weekly updates on 27+ metrics across 5 industry verticals
- Grab your tickets to the Ecommerce Marketing Summit from Agorapulse
- Get a $200 credit with Insense
Watch on YouTube
[00:00:00] Intro Ad: This episode of the E-Commerce Playbook is brought to you by Insense. We all know UGC video converts higher than most creative, but most of us don't have the time to find and manage collaborations with creators. That's why the Creator Marketplace from Incense does it all for you, and they're offering a $200 credit to all of our listeners. Just go to insense.pro or tap the link in the show notes to learn more.
[00:00:22] Richard: Hey everyone, Richard here. Just wanted to take some time to announce once more that we finally have an email address for the pod. That's email@example.com. So if you have any questions for us about the world of e-commerce, anything from any episode, make sure to send 'em our way and we may answer them on a future episode. Again, so that's firstname.lastname@example.org. That's commonthreadco.com. All right, let's get to the show.
Hey everyone, welcome to the E-Commerce Playbook Podcast. My name's Richard Gaffin I'm your host. I'm joined today, as always by the C.E.O of Common Thread Collective, Mr. Taylor Holliday. And first time for this season, so to speak, of the E-Commerce Playbook Podcast, we got a guest, and that guest is Mr. Steve Rekuc, who is our e-commerce data analyst here at CTC, say hi to the folks Steve?
[00:01:08] Steve: Hi, good to be on board.
[00:01:11] Richard: Yeah, we're glad to have you. So, we teased this a little bit in our previous episode where we talked about our Black Friday results and all of the data that we've referred to on this podcast as all of that originates with Steve.
So the data specifically that we spoke to last week was all stuff that Steve has been parsing through for a while and one of the takeaways from that conversation was, the gap between consumer behavior and the current state of the economy. So part of the reason we wanted to bring Steve on was inevitably there's going to be a lot of uncertainty in this upcoming year despite this past holiday weekend success.
And so your ability to forecast, amongst that uncertainty properly is more important than ever before. So since that, I'm not a statistician and I'm not as able to speak to this stuff as Taylor and Steve, I'll just throw it over to you, Taylor, to begin this conversation about how we think about CAC and forecasting and modeling CAC here at CTC and what work you and Steve have done together with that.
[00:02:09] Taylor: Yeah, thanks Richard, that's a great setup. So right now it's December 6th, we're recording this. And at CTC we have about 150 customers across all the various levels of our business, our different services, and I'd say 75% of them are on variable contracts, okay, why do I say this? Because I right now, as the CEO, am responsible for building the CTC 2023 budget.
I have a board meeting coming up in three weeks, in which case I have to present to them what I anticipate. Put my name on the line, put my job on the line around a number that CTC is going to go after and hit as an organization for our revenue. And I know that many of you listening are in a similar position, whether that's as a marketing manager or as a CEO or CFO, you have a responsibility to give somebody an expectation of yourself for the future of the performance of your business.
And so I have had many sleepless nights over this endeavor because if you've been responsible for this activity over the last two years, it has been incredibly volatile and difficult to do, and we've been wrong alot, and so what Steve and I, and Steve leads data science here at CTC and is great follow on Twitter and is constantly publishing a lot of the information.
He's responsible for the DDC index that we publish regularly and is the keeper of our aggregate data set. What we've been trying to ask is how could we be better about creating a close approximation of the future? And so in that we've gone after, what I really believe is the more volatile portion of that process.
So if we think about forecasting revenue in two parts, we call this the revenue layer cake at CTC. At the bottom layer, you have your existing customers, and we can build a cohort specific LTV model to predict the future value of every customer that we acquire today and yesterday in a fairly accurate fashion.
So even though we're just using historical performance to indicate future outcome, and there's limitations to that, and we've brought in some other partners to help us build some cooler models in the future. But it's really not the biggest problem here because it's the much more stable portion of your future revenues, the existing customers you already have.
The greater challenge, and the one that we're really trying to think about is the new customer portion of your future revenue. Predicting the cost to acquire the customers that you want in January and February and March that are not yet a part of your customer file. What is it going to cost to acquire them?
And then what is the relationship between how many of those customers that we get and the cost to acquire them at volume? And so this relationship between spending CAC, it's a question that our customers are asking all the time. If I spent this much, what would my CAC be?
Or if my CAC had to be this, how much could I spend? And those are very difficult questions to answer cause there's lots of variables in that equation. So, Steve, with that backdrop, maybe you can talk a little bit about how you've approached thinking about that problem.
[00:05:19] Steve: Yeah, thank you. I was approached actually by our growth guys to look at this relationship between spend and CAC. And doing the numbers of those you wind up running into this linear correlation that you see between as spend is increasing, you also see CAC increasing, not necessarily for all brands and not necessarily all the time, but there is that limitation that exists.
And then we also found that CAC was correlated really well to consumer confidence. So that's another component that could be extremely useful in building a model. And then I also looked into the previous month CAC relative to this next month.
So if you have the same assets running same campaigns, you're probably not gonna change your CAC that to the next month. It'll change some, but it's gonna be correlated real well to the prior month. So using those three variables. And then throwing in some seasonality, because obviously there's months where brands are performing a lot better, particularly November, and there's some brands that are not performing really well at other times of the year.
Selling, sunglasses in middle of winter is a little bit of a challenge, as a good example. So using those different components, I've been building models for our brands, and seeing how their CAC can be predicted using those variables.
[00:06:45] Taylor: Okay, so let's go over these cause I think this is great. So if we think about a model as a component comprised of different inputs. The inputs that Steve just talked about are one, a brand's individual data, okay? So every brand, bamboo Earth has its own set of data that has a relationship, sometimes tightly correlated, sometimes not between spend and CAC. And we could look at that data like in all time, like all of history.
We could look at it in segments, seasonal periods, we could look at it in the most recent period and look at the relationship between all these different pieces of a brand's individual data set. That's like one component of the model that's these outlining, and for some brands that's like a really tight relationship where it's very clear that spend and CAC move together, all the time, and seasonality is not that big a deal.
And then there's other brands where these things are seemingly like not related to each other at all. It's really all over the place, depending on the brand, right? So yeah, what that tells you is that like it's insufficient to just use an individual brand's data itself to predict this thing.
So the other elements that Steve's talking about is, one, we have a larger data set, we have an aggregate view, of spend and CAC across industries, across all of our portfolio that we could use as additional context for the anticipated relationship between spend and CAC in the way that if you had a brand with zero data, that was brand new, and you said, what can you anticipate from spend and CAC? Well, in that scenario, we might lean more on the aggregate data set and say, generally, here's what happens.
We have zero history for your brand, so we can't use your individual data, so we're gonna use the general data. And then the third item, which I think is really interesting, and one of the unique things that we're spending a lot of time on is in the bucket we'll call macroeconomic factors.
So you referenced, and you said it quickly as if everybody knows what it is, but you talked about the relationship between CAC, which for those, I don't know if we've done this yet, but customer acquisition costs, what is the cost to acquire one order? So it's basically ad spend divided by orders would be, customer acquisition cost and we can look at that at new customers.
We can look at that in total, we can look at it in lots of different ways, but in this case, we're talking about generally the cost to acquire a new costumer, which would be total spend divided by new orders, right? That's CAC in the definition that we're using. But that third element, this idea of macroeconomic factors, you said that there's a relationship between CAC and consumer confidence. Can you explain a little bit about what that data is and where it's coming from?
[00:09:14] Steve: Yeah, I was using data from the Office of Economic Development. It's an international organization. I was using specifically United States Consumer Confidence Index. And that is a survey that they take of consumers to say, how likely are you to spend money in the upcoming period?
It's essentially what they do through a series of questions, and they put a rating on that, with a hundred normally being the average that they try and adjust for and based upon the response to those questions, they have a number that they'll put on that and publish for a month and they publish that monthly, and it's an indicator of how well people perceive their spending to be in the future.
[00:09:54] Taylor: That's right. And this gets released, you said, once a month, right? And is comprised of a question that's intended to reflect individuals feeling about their financial future, right? Are they feeling good?
I think the question, I might be paraphrasing a bit here, but is like, how are you feeling about your future spending in some way, it's an indication of are they gonna save or spend in some way, is that right? Do you know the exact question they're asking?
[00:10:17] Steve: I don't, I think it's a series of questions that they'll use in suprise.
[00:10:22] Taylor: Okay. And so we could show this graph here. Cause you mentioned it's normalized to a hundred as being the outcome that they try and normalize for. And so the relationship to a hundred shows you increased or decreased confidence in this case.
And what we found, and it goes back like 60 years. Like they've been doing this a really long time, right? So, there's a meaningful amount of information here that we can look at. And so we started playing with this in June and July when we started hearing about the fact that this index hit an all time low, like in the history of measurement.
Back in, I think it was July, right? Where gas prices were through the roof and there was all this talk of war and all these other indicators, it was reached an all-time low. We noticed that there was also a meaningful tip in our portfolio brand's performance at that time, and that started this process.
So can you say a little bit about how you think about the performance of that index and how you actually predict that Index's performance in order to inform predictive CAC performance? Like how do you think about that?
[00:11:25] Steve: Yeah, I think, we found it was correlated, but if we're going to look in the future, we need to use the past month, so it's not even published yet for November or into December.
[00:11:40] Taylor: I think it did get published, I think it was down again.
[00:11:44] Steve: Yeah. I saw that for Conference board, I think published it. The Office of Economic Development has not published theirs yet.
[00:11:52] Taylor: Oh, okay. So two different indexes.
[00:11:53] Steve: Two different indexes, both measuring consumer confidence. And that lag creates a little bit of an issue in modeling because I can't correlate to November to November. And if I'm trying to predict January, I definitely don't have January's consumer confidence index yet. And I don't even have Decembers. So we're gonna have to use prior months consumer confidence index to utilize that in the future for any sort of modeling.
And it is one of the components that I use in creating a, because of that linear relationship between consumer confidence and CAC, we can use that to be a component to represent macroeconomic factors.
[00:12:34] Taylor: Yeah. And. In that, one of the things about the Consumer Confidence Index is, in terms of dramatic changes month to month. So as you think about why the previous month's data might be useful, in the same way that you're describing this idea that if I'm running the same campaigns that ended on, November 30th, on December 1st, the idea that, that's gonna represent a sudden dramatic change is, let's say unlikely.
It, can occur and there are things that drive it that way, but the Consumer Confidence index works in the same way where, if I'm looking right now, the largest single month drop off ever I think was like the month of Covid. And it was like, I think from 110 to 98, like it's a 10% decline in any single given period.
It's the largest discretionary change. And so it tends to move atleast directionally for a while in one direction. Now obviously that can change and there are errors, but this is why the recency effect, can be a potential indication of what we would expect in the future, is that fair?
[00:13:37] Steve: Yeah, absolutely. It normally doesn't change unless there's a massive worldwide or countrywide event.
[00:13:43] Taylor: That's right. The macroeconomic environment will move suddenly. There are things that do that. There are outliers, but other than that, it will move in a slower fashion, it's a big boat, right? To move that so to speak.
But okay, so we have these sort of three legs of this stool, okay so now, great, thanks guys you've given me all this context. So how does that help me think about forecasting next year's revenue and how are you bringing that to life specifically for brands?
[00:14:11] Steve: So I am building out based on the brand's data, trying to find the different correlations to each of those variables and then build different components of that into a linear model and use that data specifically looking to back test that against those prior values that we know are the measured CAC, try and model it and see how we would've performed using our model and how each of those components contributes.
And then try perhaps to adjust or even eliminate some of those components. or add a component of seasonality. We found there's some brands obviously that do extremely well at Christmas. Some they're trying to sell swimsuits in January and it doesn't quite work so well.
[00:14:57] Taylor: That's right.
[00:14:57] Steve: And CAC, might that time be affected by that.
[00:15:00] Midroll Ad: We talk a lot on this podcast about using paid social strategies to grow your business, but we don't talk about organic social. That's why we're excited about the e-Commerce Marketing Summit from a Agora Pulse happening December 15th. They're the go-to resource when it comes to organic social strategies for e-commerce brands, and they'll be joined by a lineup of 20 plus incredible speakers, not to mention interactive round tables panels and networking with peers in the industry.
And it's completely free. Grab your ticket now at agorapulse.com/ecommerce or follow the link in the show notes.
[00:15:34] Taylor: So the end output, and I think if we could throw a screenshot up, at least on the video here of the spreadsheet that sort of an example of the working model as I would relate to it as a growth strategist, is what it gives me is the ability to input a spend number for a month and get a modeled or anticipated CAC, or vice versa to play the other game.
If my CAC is this, what's the max threshold of spend? And so you can think then about creating a set of inputs for a forecast to help you think about how many new customers can I anticipate in that month, and with some degree of an indication of potential accuracy, right?
So we can say how often the backdated model was wildly off, and you'll usually put 90% bounds on it and say, how often was it above or below the 90th percentile outcome on either end of the distribution curve? And we wanna see that happen very rarely, and then you could even look at that at like, how often is it within, the 60th percentile in both directions, and you can get a sense of the accuracy, of the forecast historically to help us think, okay, how confident should we be about this expectation?
And that's an important thing to represent as a forecaster, is one of the things we try and teach our people is that you need to represent to your customers the level of certainty you have about the outcomes you're presenting them because they're gonna make decisions based on it, right?
They're gonna say, if you tell me you're gonna spend this at that CAC, and you represent, I am very confident in this, I am 90% sure that this is gonna work. They're gonna behave very differently than if you say, look, the error bars on this data are very wide. we can't really, with much accuracy predict what's going to occur, and so therefore, we should be very cautious in making any monumental commitments of cash or otherwise against this outcome.
And that sort of holding the future with an appropriate level of certainty is a skill that I think this kind of modeling can help people to understand. It can give us ranges of expectation and confidence Is that fair?
[00:17:34] Steve: Yeah, absolutely. In fact, that was the other output of the model as well, like the third output was not just the predicted CAC, but if you give me a CAC that you would want for that month, given a spend, what is the likelihood that CAC will wind up higher and what's the likelihood it will wind up lower for that month?
And that's maybe...
[00:17:52] Taylor: That's great.
[00:17:53] Steve: Maybe the more important piece because then, you know, great, I've the model saying, $25 per CAC, I needed a $24, what's the probability I'm gonna be over that?
[00:18:05] Taylor: That's right. And so what this helps people to anchor themselves in, cause what I would say is a bad growth strategist takes a model and expresses it as the certainty of the future.
And a good growth strategist uses it and says, this is what's likely to occur, now let's talk about what we would like to occur and what we will need to accomplish in order to reach one of those upper bound outcomes to reach that upper quartile. How much better would we need to be and where and what are we gonna do to try and pursue that?
So we can anchor. With data ourselves into an appropriate expectation that's realistic and is thoughtful at the very least. And we could say, but we're not just gonna accept that because that premise might not actually be business viable in some case, that might actually be business unviable and we now know we have to actually go be better than that expectation.
So what are we gonna do to be different than the past? How are we gonna disassociate ourselves from that data and alter it going forward? But that initial sense of what does the information tell us is likely to occur is a really powerful starting point.
[00:19:07] Steve: Yeah, absolutely. And that also I've seen as well is that we've been able to improve CAC where we've taken over brands, where brands have kind of come on board and, they're beating the model more recently.
[00:19:22] Taylor: That's right. And that's of course the goal, right? That's what great marketing effort and appropriate ad structures and great creative and all those things are intended to help you do. But this starting point, one of the things that I've seen for brands is they think about setting 2023 goals, is this question of I've seen too many models, I've used this reference.
I'm working with a brand right now, and they have a former P&G C.F.O that's building their model and they've taken the CTC growth map and they went and applied it. And what he did is, a very common thing that I see in growth expectations, because it looks really great on a financial forecast, is you take spend, you ramp it up all throughout the year, and CAC is just constant.
As if those things have no relationship to each other, which would be great, we'd all love for that to be the case. But the reality is there is often some relationship between the two, and so you have to account for that to help you answer what happens as I attempt to grow. Because one of the things that I think brands have to do is they have to not only consider the rate at which they're able to increase spend and maintain profitability, but also the rate of growth of the existing customer file,
that sort of builds out the individual months, PNL, and as long as you can move those things together, you don't break what I call the velocity constraint. The velocity constraint is like the limitation of growth before you break that profitability threshold. And this model helps us to answer where does it break?
Where does cat get too high at what spend? And that becomes your velocity constraint.
[00:20:50] Steve: Yeah, absolutely. And then you'll have the opportunity with those additional acquired customers to provide that more reliable returning customer revenue.
[00:20:59] Taylor: That's right. So Steve, maybe what have been some of the challenges that you've run into? Because I know at first what in an ideal world, there's like a generic model that works for everybody, right? And that would be great, and we could throw it up on the website and you could plug in your thing and we'd spit out the numbers. But maybe talk through some of the intricacies of each individual data set and what makes this exercise really difficult to do across so many different business.
[00:21:26] Steve: Yeah, it certainly becomes difficult. One of the examples I, in fact had mentioned is that when a brand comes on to us as an agency and we've had data prior to them having us as an agency and then us being their agent, it doesn't necessarily have the same sort of performance. All of a sudden we're increasing spend, and getting a lower CAC.
So that looks like not necessarily the closest relationship at that point because there's this data set that occurred before where they were on a different path with respect to CAC versus spend a different curve. And then we took over kind of their ads improve their campaigns, and then you see the not exactly the same slope exists.
So that's one of the challenges, one of the other challenges I found was the seasonality, and I had to add a variable for that. Particularly November, we see a lot of CAC drop, as we've been talking about, a big opportunity for brands to acquire more customers at greater volume at keeping their CAC lower.
And then as well as seasonality for brands that might not be conducive to selling their item at that time. Some summertime stuff doesn't necessarily sell that great January, February, March.
[00:22:39] Taylor: That's right. And that's why I think having a partner or having somebody like Steve inside of your organization or a partner like CTC is, really important because, I'd love to be able to just, like we said, just feed you guys the inputs to the spreadsheet and send you on your way. But the reality is every business is, has a novel set of information and it has a unique sets of attributes. And it's really important that you take into consideration as many of those human understanding elements that inform the model as possible.
If you know that next March you have a big product drop coming, and last year you didn't, and you anticipate that it's going to alter the AOV of your brand, well then CAC is likely gonna change dramatically too, right? As the composition of what you're selling changes. And these are the kinds of things that it's very difficult for any model to account for is when the future is wildly different than the past.
And so that's where us as growth marketers and partners have to think, in a very human way about what we know to be true about what we're attempting to do in the future and how that might inform the way we think about our model.
[00:23:46] Steve: Yeah, absolutely. And I think there's gonna be a lot of different components. I'm gonna be able to add onto these as we go. I'm going to keep improving these models and pull in more.
[00:23:55] Taylor: And that's right, and I really believe when we think about what we're trying to design at CTC, we call it an operating system for profit.
And what we want to do, and the components of that, I really believe are critical for every business is to start with a clear objective outcome that you're trying to accomplish, and it should be a financial measure, and not to rehash the hierarchy of metrics, but we would like to forecast it down to the contribution margin level.
Have a sense of the inputs that generate that output, have a model or view of how those inputs change, to set a goal then for each of these inputs around that generate that collective output, have clear ability to measure your data against those inputs every single day, and break those goals down into daily expectations of those metrics so that you can quickly identify where you are inevitably wrong and course correct.
And that's that when we talk about a system for managing, that's it. Use data to inform a model that creates expectations, break those expectations down into daily inputs, be able to monitor those daily inputs and make adjustments in real time as fast as possible in order to get to where you're trying to go.
And that's what we're trying to build and inform it as best we can, and so the work that Steve's doing as well as Qua and the team on the status side, are all about generating the environment in which we have the best information to make the best decisions as fast as possible to get where we're trying to go.
[00:25:16] Richard: Cool. So actually Taylor, do you feel like what you just talked about there, what's the takeaway for a brand that does not have a Steve on staff? Those of us and I say us because I'm among them who do not have this sort of inherent or learned understanding of statistics, what are like the key basic takeaways that you can do so you don't make mistakes, for instance having the same CAC every month in your model?
[00:25:43] Taylor: Yeah, so I'll answer then Steve, I'd love your perspective as well is that I would understand that in general there is a relationship between the increase in spend and the increase in CAC. And that is an obvious statement, I'm sure, but I would ask yourself, do you really feel like, one, you're building that anticipation into your forecast, that every month if you want to increase spend there, there's gonna be some increase in the cost of the new customers that you acquire.
You could do a very simple correlation analysis, export your revenue or your spend by week for the last year, you could export your CAC by week for the last year, go into Excel, hit equals C O R R E L parenthesis, pull in one dataset, comma the other, and get a sense of, okay, in my individual dataset, is there some relationship that I can obviously see,
if so, okay, you've got a starting point to consider the relationship and that it exists and that it might. Fairly linear. One goes up, the other goes up, right? Like that could be ideal scenario, okay, you have some sense of it. If there's none at all, you've got a more complex situation. And that's the case for a lot of brands with small data or where things have wildly changed over time in some way.
And then I might enlist the help of somebody to help you think about this because getting it right is business critical because you're gonna make decisions about inventory purchasing, and you're gonna make decisions about hiring on the basis of whatever this forecast is. And so I think at least having some idea of this in your forecast, you know, we have a great entry point is we have the enterprise scaling guide that walks through our cohort specific LTV model that's like, okay, for a thousand bucks, you can walk in a step-by-step basis through exactly how to do this.
We have a service on the simple side that's just, we'll build you a forecast with this data-driven model for next year. That's a one-off project or on an ongoing basis, this is our service. So I think you've gotta have a point of view. So the question I would ask yourself right now is, do you have a point of view on the spend on the relationship between your spend and CAC?
If not, how do you get your hands on one? So that would be my first thought. Steve, what would you say?
[00:27:37] Steve: Yeah, I think that's fantastic. Way to begin particularly, and if you find issues in the correlation, I would also consider throwing out outliers. For example, a lot of Q2 of 2020 for a lot of brands as an outlier.
What happened there with your CAC is not indicative of normal operations. That was a worldwide pandemic that was going on, it drastically changed e-commerce. So that I would consider in looking at the correlation, consider throwing out those data points. I think you also want to consider macroeconomic conditions.
we know this intuitively, somewhat, that when consumers are less confident, they're less likely to buy, and that means your CAC is gonna go up, typically. There's some brands that buck that trend, but in general we see that rise. So anticipate that, don't just look at last January and say, we're gonna do the same exact one as last January. That's probably not gonna happen at the very least, costs have inflated.
[00:28:33] Taylor: Yeah. Here's an underrated growth strategy, okay? Because generally speaking, one of the number one things that when I meet a business that they have to accept is slowing down.
And what I would say is, if you could hold, assuming you have even a modicum of LTV, like even 50% in the year, let's say, which is half of what we would call the sort of golden ratio of 30%, 60 days, a hundred percent in a year, even if you got to 50% in a year, this would hold true for your business.
If you could hold spend and CAC constant every month, your business would grow and your margin would increase over time, okay, and this is not intuitive to people necessarily. Hold, spend hold CAC constant. What would happen is each month the percentage of your revenue that was comprised of existing customers versus new customers would grow towards the existing customers.
And assuming we're gonna assign $0 in spend to that base, your margin would widen over time and you get to the state that the biggest, best businesses in the world are where 90% of their revenue comes from their existing customer base, and 10% comes from existing customers or even less, that's where real profit comes from.
When you're in the hypergrowth stage and most of your revenue is paid for, meaning it's new customers that you're acquiring via paid acquisition, this becomes a very difficult exercise and often your cack will deteriorate faster than you anticipated, and it causes real business problem. So one of the ways to de-risk this profile is to slow down.
So that's another recommendation. If the data is uncertain in that uncertainty, you are more likely to die overspending against an unprofitable C than you are to underspending against the profitable one. So if one path is death and one path is like, I'm just going slower than I would've liked and maybe making not as much money in uncertainty, don't walk into two paths, if one is death and one is slow with uncertainty. If those are the choices, you better be really certain when you make the decision, and so in general, in this next period, slow down.
[00:30:35] Richard: All right, let's end it there. We can't all have Steve's on our team, and if you want access to somebody like Steve or Steve himself, as Taylor mentioned before, we are putting together 2023 forecasting packages right now, go ahead, go to commonthreadco.com, fill out a form.
We would love to talk to you about helping you out with that right now, but anyway, Steve, thank you for joining us, appreciate having you.
[00:30:59] Steve: Thank you, Richard and Taylor.