Search
Log In
Subscribe
Search
Jul 27, 2026
Your hair, it's everywhere
Your hair, it's everywhere
00:00
19:04
Transcript
0:00
Your hair, it's everywhere, screaming infidelities, and it's taking its wear I love that line. It's a great one. And I mean, of course, we're Confessions of a Dashboard, right?
0:10
So we've got to start off with a little dashboard confessional as, as the first episode. On- only makes sense. Okay.
0:16
So what comes to my mind the moment that I hear this is the number of organizations that I've worked with that have, like, hundreds of data sets. I'm not, not even exaggerating there.
0:27
Hundreds of data sets that are absolutely everywhere. There's no source of truth.
0:32
Everything is tangled, and it's an absolute mess to get a correct answer out of, of anything, which then makes it really, really hard to understand what data you do have and who your... And then who your customers are.
0:48
Yeah. You know, the, the, the way you describe that to me, I think of, like, the rat's nest of microphone cables at a studio.
0:57
In a studio, the rat's nest of microphone cables is a problem because if something goes wrong, you have to chase it back and try and figure out where it came from.
1:03
So it's not just the endpoint, it's also where it's going. Yep. Does that relate? 100%. Yeah. For...
1:12
There was this one really large organization that I worked with and, and it was super interesting because they probably had between 25 to 30, like, true data sources, and they all represented something, like, really interesting for them, but in order to get them into, like, a united tool at the end so that it was a clear picture of everything that was happening in their business, we had to do so much processing.
1:34
At the end of the day, I think it probably took me over 100 transformers in order to get the data put together and as a source of truth that we could actually use in our tool. So when something broke,
1:49
having to go through, like, 100 transformers to figure out which data set did this actually come from and then where is it broken?
1:57
Is it broken in our pipeline and in our workflow, or is it broken on, on their side on the other end? Because it's not like that data was coming to us direct either, right?
2:05
Like, their team had to edit and process some of it before it came to us, too. This is a very extreme example, but it is... It really does shine a light on just exactly like what you were saying.
2:16
When something goes wrong, it makes it impossible to get the right answer. Yeah. So
2:23
i- in that situation where you, you've done all these transforms and you've cleaned this data to some extent, is there a way to permanently make that happen, like, you're...
2:31
So you're not always tracing it back through a transformer? Yeah. I mean, ideally, yes. That's kind of the... Ideally, yes, [laughs] that's the goal. And I...
2:39
And then that's comes down to, like, that idea of that one source of truth, right? So having things as simple as consistent store numbers or, like, donor numbers.
2:48
If you're mapping something and you wanna understand someone's volunteer experience and someone's donations and how long they've been with the organization, all that information, if you can have a unique identifier that that's your pair across all of the sets, it makes things so much easier because then you can see where the break is in it, and you have s- you have, like, a key, and the key to connect it all then makes the data flow through easier.
3:12
That's often what I recommend for folks when they're, when they're going through all of this.
3:16
Everyone's gonna have multiple different data sets, especially in the world that we're living in right now, where there's multiple different ways that folks can get information and different ways that they're processing it.
3:26
But what I try to go back to then is, okay, if we can't have one united source or we're working to getting one united source, what is, like, that unique identifier that we can use to map and match things across the board, and how can we get consistent then within our naming terms and our product IDs?
3:45
Or if we're adding a new color, what's the simple way that we can, like, add a new color to something?
3:49
If we are flagging someone as a monthly donor, what's a simple way that we can do that so that the term is consistent?
3:55
So if one development person is putting in monthly donor, they're tagging it the exact same way the other development person is, so that there's at least, like, consistency in identifying the person or the store or whatever it is, but then also within the products themselves because that often tends to be where things start to fall apart, is in the, the, the mislabeling of
4:17
information or just, like, you know, something's named, like, in-person participant and then something's named in dash person participant, and those aren't getting picked up to be the same things. Right. Okay.
4:32
So I feel like that's a good way to map, you know, a marketing lesson into, uh, or a data lesson into your hair, it's everywhere. But l- let's be real, screaming infidelities is what we're talking about.
4:48
Screaming infidelities. Well, I... But this is kind of where I love this line. It just reminds me of the garbage in, garbage out.
4:53
So it's like the i- [laughs] the infidelity at the end of it is just, like, the false lies that you're left with because the data is just hot garbage the whole way along.
5:05
And so it's less so about, like, the, the cheating on someone and all of that. It's more so to me about, like, you're being betrayed by your data because the whole pipeline is so dirty getting there.
5:16
So when your data is everywhere, what you get out is just... It's gonna be deceptive, and it's not gonna tell you the full truth. And so that's where I think about, like, the...
5:25
That's where the screaming infidelities actually made me really love this, love this line so much. I'm a musician. Sure, yeah. I, I don't know what I'm talking about here.
5:35
When I hear screaming infidelities, I mean, like, I, I hear, like, you, you can't miss it. Like, this is wrong.
5:45
I- is there a way to, when you're looking at that data on the output, hear the screaming infidelities from the beginning?
5:51
Like, what do you look for to know whether you need to check your pipeline and check all your sources again? To me, that comes down to, like, as you're building it,
5:59
I build in checks and balances at every single step of the way. I know you do this too in your studio. There's multiple spots that you can check, and you stop to go, like, "Okay, up to this point, is everything working?
6:11
Yes. Good. Okay."
6:12
And so for me- That's a big part of this process, is building in these, like, little stopgaps, these little checks where you can just go, "Yeah, okay, everything's still summing up to the right number," or, "We're still...
6:25
We haven't dropped anything yet." At the end, you're like, "Okay, here are all the numbers. Let's go back and check, like, the previous, the, the last QA. Is that last QA adding up? Yes. Okay."
6:36
And just, I go back all the way just to make sure, like, "Wait, where have we actually dropped something? Where, where has something potentially fallen apart?"
6:44
So a lot of those checks where you're, like, summing to zero, where you just, you want, like, this and this to equal zero because that means they're entirely balanced, that both sides of the equation has worked out.
6:54
Does that relate to sort of what you do in, in studio? Yeah.
6:59
I mean, there's of course, like, points in the signal chain, like the microphone is plugged into the cable, the cable is plugged into the input, the input goes to the piece of equipment, the volume knob's turned up,
7:10
so on and so forth, and you can... You just chase the problem down the chain. And then, yeah, like, let's say you're... You have
7:19
two microphones plugged into two cables into two channels, and you're recording the same thing. You can make sure that everything's working by inverting the polarity of one of them- Right...
7:31
and putting them back together, and now you should hear close to nothing if everything's working the way it's supposed to. You create a null.
7:39
So you're trying to take all the positives on one channel, all the negatives on another, and put them together, and you should hear nothing, and that means they're the same. Everything's working.
7:47
That's 100% the same thing. So that's exactly what we try to do, is do the... If we're running these two simultaneously, at any point where we check, can we make sure that we can get it to zero? Yeah.
7:59
So I, I wanna come back to the screaming infidelities thing one more time because I, I, you know, I agree. Like, yeah, walk back down the line, but let's say y- it's a new client, it's a new job,
8:13
you're new to the company, whatever, and they go, "Here's our data. Go." Like, are... Is your first thing to audit the whole pipeline? Or is- Yeah...
8:25
is there a way you can look at the data and know whether or not you should audit the pipeline? Well, I'm not trusting a pipeline that's handed to me- Okay... at all. Got it. And that's not on, like, anyone.
8:37
That's not on anyone. It's just, it's, like, so I can sleep at night. Yes.
8:41
Because it's more that in order for my brain to understand what the final product is that I'm working with, I need to know everything that went into it.
8:49
I need to understand, like, how all these different systems paired together and how they got what they did. And again, that's in, like, a very extreme case.
8:58
If it's just someone's Shopify store that we're looking at, that's very simple. If it's just one person's database, that's, that's not hard. I'm talking about, like, when we are looking at those 25 sources of truth,
9:10
I'm gonna wanna see what, what went into it so I can understand what data we're working with.
9:16
And also, and just to, like, add to that, this is where, like, where I get really excited on this stuff, just to add into it, like, sometimes data that's actually really cool and really useful doesn't seem cool and useful, and so it gets dropped along the way.
9:28
I like to go back to understand every single piece of data that exists because then when we understand every single piece of data that exists, there might be some cool things that we can do with it that they, that just had not crossed their mind.
9:41
And so just to give an example, like, last week I was chatting with a, a client, and I had said to them, like,
9:48
"Well, what you have is awesome, and it would be so cool if you had, like, this one, this one piece of data," and it was a theme of shows that people were attending.
9:58
So not just looking at ticket sales, but actually looking at what type of shows those folks are attending, 'cause then we can use that to segment the audiences.
10:07
So I, I do like looking at the full pipeline to see where stuff is maybe getting dropped along the way that we could end up pulling in later to use. That makes a lot of sense to me.
10:19
I mean, a fresh perspective's always good. You might see a way to leverage something that a previous person didn't. That's awesome. Yeah.
10:29
Time of day is another big one that often gets dropped 'cause people are like, "Oh, it's just, it's just date and time. That doesn't really matter." But for me, what I wanna understand is what time is the person buying?
10:40
That explains so much. If you have people buying at 2:00 in the morning, and your product is baby-related,
10:48
you're going to need a mobile-friendly site that is easily browsable with one hand because that person is probably solving an emergency while holding a crying baby in the middle of the night or hopefully at that point a sleeping baby in the middle of the night, and what your site needs to do needs to be entirely different than a site that someone's browsing at noon on a desktop computer.
11:11
So even those pieces of information, that'll often get dropped. With donations too, we can also look at how many touch points did the person have before we had the donation? So don't throw out call logs.
11:24
I wanna see call logs also.
11:26
I don't need to see the content of them, but I just need to know touch points and number, and, and what was the cadence of the touch points before we got it and, and is there a pattern that actually works?
11:36
So that's another really good one that gets thrown out all the time and could really help paint a good picture when, when you're going to that next step of marketing and segmenting and, and targeting. Yeah.
11:50
I'm g- I'm gonna use some words that I might not know the meaning of, but I feel like that would be really useful. Li- like, it makes sense to me that people would just go, "Oh, a timestamp. I don't need that."
12:02
But I feel like that there's so many use cases for that. One that comes to mind is, like, if we send out a marketing email at...
12:09
And we, you know, we A/B with this time of day and this time of day, which time of day gets a higher click-through rate? Exactly Right?
12:17
Like, you, you wanna know that so you can time your emails to the time that gets the highest click through rate.
12:23
Well, and to challenge- And if that data's gone, it's gone Well, to challenge you more on that one too, while click through rate is great, I, I wanna understand is higher click through rate actually delivering a higher purchase rate, or higher- Yeah...
12:37
donation rate? What if this one, it's like 45% of people open, but you have like 10% of people buy?
12:43
What I wanna know is if this one only has 25% of people open, but 50% of them bought, that's actually more people, right? So I, you know, I...
12:52
That's where every piece along the funnel is important because we might throw out, we might throw out the click- Let's say we throw out the click through rate. We're like, "Oh, that's not important," right?
13:03
And we just look at the purchase rate. So that, that is important. When people wanna know the purchase rate, that makes sense. But the way you write those emails are different, and they might have different objectives.
13:15
One might be more of an awareness one, and so like you might think, "Oh, that one's kind of garbage," because it only deliver- it delivered a lower, a lower purchase rate, but it had a lot of people actually read the message, and that itself is data too.
13:29
So why did a lot of people read this message and not click through? Was the call to action not good enough? Like, what happened there? Okay. So yeah. So d- data doesn't necessarily scream infidelity.
13:41
It g- unle- well, I mean- Yeah... I guess it can, but if you just look at it without knowing the pipeline- Right... it's not necessarily gonna jump out to you.
13:51
Like, you're, you should probably just check the work anyway. Yeah. Well, it's like, it's like math, right? You gotta go back and... High school math, you gotta go back and check your work. Check your work.
14:01
Now, I th- this is fun. So it's taking its wear. I would love to know an example that you've run into of bad data taking its wear.
14:16
Like, you know, I, I'm sure most people are familiar with the con- uh, the idea of introduced errors. Yeah. And that's really what we're talking about here. Yes.
14:26
So yeah, have you ever had any situations where like, of, you know, without naming names, where, you know, uh, maybe a decision's been made or something based on added, added up bad data over time? 100%. 100%.
14:42
And by very intelligent, well-meaning people. This happens to the best of us. Like, this is...
14:47
You know, sometimes you're too close to it or you're in it every day, and you just, you don't have that perspective of like, of... Or the time, honestly.
14:56
Like, and a lot of the folks I work with, it's they don't have the time to go back and, and check and audit every single thing, which is part of the reason that, that I get brought in.
15:04
One organization, it was like a really big difficulty in actually forecasting what their live revenue was. Wow. Not understanding what your product mix is, what your product, product lines are, right?
15:15
So when you, when you are working with something and, and the naming is inconsistent across your products, you are unable to actually find out what you're selling, and so you're not gonna know what to stock.
15:27
You're gonna have no idea on how to do that forecast planning, and that's another one that I've run into, is just the struggle with forecast planning when there's not a clear picture there of, uh,
15:36
of, of, of what it is you're selling. It might be logical to you, but when you think going back like two years, three years, four years later, are you gonna remember what you were thinking in that moment?
15:46
Sometimes it's also, Sam, like not even just a bad decision being made. I find what often actually happens is it's a lack of decision being made- Okay...
15:55
which sometimes can be just as bad as a bad decision, because you're not able to make a decision that would move you forward.
16:02
And so you're stuck in this limbo because you can't get access to the data that you, that you need because it's so messy, because it's, it's been this very difficult, very difficult thing to organize. Yeah.
16:17
That makes a lot of sense. I try. Yeah. Okay. So we've talked about a few things here. Let, can we sum this up into like what's the thing someone can do? Like- Yeah, that's a great question...
16:30
h- like I want people to hear this, uh, to, to, what, for what, you know, every time they're singing Dashboard Confessional in their head, which we, they're all doing- All doing. All the time...
16:40
I want them to think of the line, you know, "Your hair, it's everywhere screaming infidelities, and it's taking its wear," and I want them to go, "I need to remember to do this," or, "I should do this." What is that?
16:52
That's a great question. Get to that single source of truth. That's what it is, and I know that like maybe that sounds a little too arbitrary. So if I had to say like pick one thing, streamline your product mix.
17:04
Whether you are a nonprofit, whether you are a small business, whatever it is that you are selling, have consistent naming across what you're selling. If you sell sweaters, have consistent naming for SKUs.
17:18
If you're a nonprofit, not only having consistent naming across touchpoints and, and donations and all of that, but ensuring that you have good donor IDs, 'cause I know we're all guilty of this, having five duplicates of the same person in the database.
17:33
That's something that happens a lot, because they get added as a volunteer, and then they get added as a major gift donor.
17:38
And so coming back to the how can I simplify this down to like the smallest, cleanest amount, because that will make anything you do afterwards easier.
17:49
And, and to just add on, and I know this is like not a tasty little nugget, I'll try and sum it up a little bit better, or maybe you will, 'cause that's what you're really good at. You might have to start from scratch.
18:00
Don't be afraid from going like, "From today forward, it'll be good." If it feels too overwhelming to go back, yes, it sucks to lose historic data, absolutely,
18:12
but starting today is better than just not doing anything at all. That's gonna be so much better for you than if you don't start at all. Can you sum that up for us in a tasty, [laughs] tasty little nugget? All right.
18:25
I mean, like you said b- at the beginning, garbage in, garbage out. We've been talking about the pipeline. This is before the pipeline. This is, this is T equals zero. Yes.
18:36
Make sure that you understand the labeling of all of your data, because that will make the pipeline easier to follow. 100%. So clean your, clean your, clean your data. Clean your data. Pick up your hair.
18:54
Don't let it go everywhere. Exactly. Keep that in a nice, tight bun. Yep. You don't need a fourth Roomba. You don't need a fourth Roomba. [laughs] No
Confessions of a Dashboard
Recent episodes
No results found
