Contextualized Insights – Ep. 239

When someone says “the number is up”, the next question is always the same: up compared to what?

In Episode 239, Mike, Tommy, and Seth use a Brent Dykes / Effective Data Storytelling article on contextualized insights to unpack a deceptively hard skill for report authors: giving the audience the right frame of reference so the visual implies an action.

News & Announcements

• Contextualized insights: six ways to put your numbers in context — A clean checklist of context types (comparisons, scale, history, and more) you can apply to almost any KPI.

• Submit a topic idea — Have a scenario you want the team to argue about? Drop it here.

• Subscribe to the Explicit Measures podcast — Episode archive + links so you can share with your team.

• Tips+ Theme Generator — Build consistent, branded Power BI themes without hand-editing JSON.

• Mike Carlo on LinkedIn — Practical Power BI/Fabric patterns from the field.

• Seth Bauer on LinkedIn — Data engineering + analytics delivery perspective.

• Tommy Puglia on LinkedIn — BI leadership, adoption, and the ‘make it useful’ mindset.

Main Discussion

The main theme is simple: numbers don’t create insight—context does. A metric can be perfectly accurate and still be unusable if the reader can’t tell whether it’s good, bad, normal, or urgent.

Using the article as a starting point, the team talks through multiple ways to add that missing frame of reference. A few that come up repeatedly:

• Comparative context (vs. last month, vs. target, vs. peers)
• Historical context (how things have changed over time)
• Scaled context (percent, per-capita, normalized values)
• Confirmative context (does a second metric support the same story?)

They also connect this back to day-to-day Power BI work: context doesn’t start at the visual—it starts upstream with domain knowledge, definitions, and transformations. And on the report canvas, ‘more visuals’ rarely fixes confusion; the better move is to design around a single question and then add only the context required to answer it.

Key takeaways:

• Make comparisons unavoidable: pair every headline KPI with at least one reference (target, prior period, peer group, benchmark).
• Normalize early when it matters (rates, per-capita, per-unit) so scale doesn’t hide the real story.
• Don’t treat context as a formatting problem—get definitions, grain, and domain rules right before you build the semantic model.
• If a chart doesn’t point to a decision, add context via segmentation (small multiples), variance, or a short narrative so the next action is clear.
• Avoid ‘dashboard confetti’: start development with one insight per page, then earn each additional visual.
• Standard patterns beat hero-reporting: reusable measures + templates make contextual reporting repeatable, not artisanal.
• AI’s best role is automation + explanation (generate candidate visuals/measures and explain them), while humans still own the real skill: extracting what matters from stakeholders.

Looking Forward

In your next report review, pick one KPI and force yourself to add context until a first-time viewer can answer ‘So what?’ in 10 seconds.

Episode Transcript

0:01 [Music] foreign good morning and welcome back to the explosive measure podcast with

0:32 explosive measure podcast with Seth and Mike had some technical issues I believe this morning so we don’t have a timer yet we’re gonna see if that will appear here shortly but welcome and good morning everyone everyone good morning Mike and that’s weird yeah I hope it’s a happy Tuesday yeah this is this is the joys I guess of going live whenever we possibly can early in the mornings and we’re we’re a

1:02 early in the mornings and we’re we’re a man down at the moment hoping the internet works wherever Tommy is hopefully everything’s good to go there so we’ll see if Tony appears we know our audience is so fast and wide if when you could just stop by Tommy’s House knocking on the door or wake him up and tell him that he should be with us us as he as he gets online here or we hope it’s online we’ll pull him into the conversation sounds good so so some some things have developed in the in the Carlo household

1:32 in the Carlo household a very funny event occurred over this this weekend and so I I feel honored to share this one so you’ve made it when you’re a podcast when your children actually start making their own podcasts but they weren’t just making their own podcast they so my kids are gathered around the iPad in the living room and they had hit record on the little audio recorder I guess they have found the audio recorder on the iPad and so they hit record on the audio and they started their own explicit

2:03 and they started their own explicit measures podcast on our living floor living room floor so you didn’t tell me this part yeah I I you said that they had sat down and watched us well they have they have seen parts of our videos here and there but they’re like I guess bored on the afternoon and they were like we’ve gotta like we want to do a podcast and so they always they don’t like tease but they do tease so they sit down and my three kids start talking together as the podcast and my my middle child

2:34 podcast and my my middle child goes hi goes hi I’m Seth I know everything that was her that was her opener it was like okay that’s pretty funny and then my youngest who was interpreting Tommy basically said hmm I’m hungry I need to go get some meat so she would so she ran off to the kitchen to go get some food something to eat something that that is actually very Tommy honestly I would say that would be very much him and then my son who

3:04 very much him and then my son who taking after his dad wanted to be wanted to be me and so my son tells me he goes yeah it was just he’s talking in the into the iPad and he goes Dana data data data data data I love data data and I was like yeah I do say that word a lot so they you should he should have went in there Game Changer and then it would have been all over over it would have been over it was pretty funny so Jack Kirby’s asking when will this be released Mike

3:38 I think this might have to be a subscriber only interface foreign there was a lot of just random noise and then there’s being silly with together but they had like a five minute podcast and they then they insisted that I listened to it and they were very proud of like their intros that they were producing for each person so they they really enjoyed the the fun that it took this weekend so anyways a fun little yeah exactly data data data that they were they were loving Mike and Dad saying

4:08 they were loving Mike and Dad saying data all the time so it was just very fun good little interface that they were doing they’re just yeah it makes you smile so they’re listing it’s it’s also oddly strange your your kids are always watching they’re observing they know what’s happening so we’ve been on a run here recently where we’ve been doing a lot of extra recordings for podcasts so that like literally every morning I’m showing up to record things so this is very top of mind for them oh anyways I always had a kick out of it

4:38 oh anyways I always had a kick out of it oh I didn’t stop laughing when you told her the I’m Seth I know everything yeah yeah anyways I thought that was a really good little story today we’ll jump into our topic this has nothing to do with with my kids being silly and all that at home but but we are jumping into a talk topic today around effective data storytelling as is the website and there’s I think this is a blog post yeah I’m trying to figure out where who

5:08 yeah I’m trying to figure out where who oh it’s Sykes again and this is also from the data storytelling book series and we’re talking about the idea here around contextualizing insights six different ways to put your numbers into context which I think is context is very key when you’re talking about data data analytic things so let me put that in the chat window here I’m going to try to spell contextualized correctly correctly insights and then here’s the article

5:39 insights and then here’s the article that we’re going to be talking discussing today if you want to pull up the article on from the browser feel free to go ahead and hit that in the chat window it’ll also be in the description of the video if you want to follow along as well Seth I want to give us some initial points or a initial outline of this article figure out where we where we want to take it today yeah I I think we’ll we’ll spend a lot of time in this diagram and I just put together that walks through the different areas which context is is really important and

6:09 context is is really important and how we would present that obviously as this bi report authors and our job so comparative informational equivalent confirmative scaled and historical so we’ll be spending a lot of time in there but one of the the parts of his intro that kind of the the parts of his intro that stuck out to me that I really liked of stuck out to me that I really liked that we’ll talk about in the future as well well is is the importance of context when you’re connecting to the the data

6:40 you’re connecting to the the data sources themselves right and and pulling in data into the data systems and the context that’s most important is to have do like some domain knowledge around what it is you’re ingesting and we talk about that I think at length in a future episode but it resonates here once again where there’s a breakdown of old versus new ways in which I think business

7:10 ways in which I think business intelligence and analytics teams run or should be from an I. T team just being completely independent from the business right and just saying it stated it’s data data and you just pull it in right pull it all in exactly you’re gonna build this monolithic thing all the datas are the same way from The Source system it doesn’t matter how it’s being used right now and and then you start running it through the pipes without any context right or you miss a lot of the

7:40 context right or you miss a lot of the transformational things that a domain expert would know and right out of the gate that creates issues not from a data quality like like the data may be accurate you just don’t have the context of the business insights and you’re going to miss some things and down Downstream that’s just going to pollute your reports right so like from the very beginning I think this idea of understanding a lot more about data sets as opposed to just treating it as yeah

8:12 as opposed to just treating it as yeah it’s raw information right and we build the context in the report I think it points out that nope it’s right from the beginning we need to start understanding like what are some of the contextual factors or how do we apply certain transformational things to the data to make it meaningful from the get-go get-go when you when you do a little bit of studying around visuals or visual design you look you this is I think sometimes when you look at visual or building visuals on top of

8:42 visual or building visuals on top of data we think oh you just like know what to do and a lot of times I’ve interviewed people and said how do you interviewed people and said how do which visual to pick know which visual to pick what where have you learned this information from and a lot of people just say oh well it’s a Time series I just know that’s a line chart it’s it’s categories I know that’s just a bar chart but when you really think about it there’s

9:01 you really think about it there’s actually entire fields of study or people who have studied this for long periods of time thinking about like what does this look like how how do you represent the data or information on a page that produces a conversation around that information and I I think we’ve said this in the podcast in the past and I think this really resonates for me when you’re using visuals to somebody else else you’re actually communicating you’re talking to somebody else with the visual through the data

9:33 else with the visual through the data yeah speaking of through the data hey look who showed up I like I like your intro can’t get rid of me that easily all right welcome Tommy welcome back to the podcast your alarm should have been set for 7 15 this morning was there was there I was getting a little nervous I was thinking Tommy got run over by a car and a bike that’s literally what I

10:03 car and a bike that’s literally what I was thinking I was like he’s always here so sweet relief yeah there we go so we’ll we’ll keep going so what did you please we’re just starting off with the article Tommy we’re just jumping in here so yeah so I I felt like those are good parts but I liked at the end here he started talking as he leaves the intro and starts talking about these contextually insights inside things there’s a there’s a point here that I really like that was said Michael Ventura someone he’s quoting without context a piece of

10:34 he’s quoting without context a piece of data is just a DOT so it’s the other relative data points that make this one piece of data relevant and that makes a lot of sense a lot of times we’re doing year-over-year analysis this year compared to last year there’s two data points I can see is it up or is it down and this is something that has been beat in my head is when you’re thinking about data you’re talking about position what is the actual value you’re talking about and what is the direction right how is that that data point being compared to

11:05 that that data point being compared to something else what is the relative compare and then does that align to your goals are you expecting that number to go up are you expecting that number to go down go down I think talking about visualizations or communicating in those simple terms what what is the number and do we expect how do we expect it to change those two questions I think answer a lot of details around what’s occurring inside this context for data I think I I agree like the last couple

11:36 I think I I agree like the last couple paragraphs right before getting into the six ways to contextualize yeah are definitely interesting and have some key points because it brings to light this other theory that we’ve talked about I think which is your audience is pre-wired to understand visuals that they’ve seen before they know how to that’s true visualization yes correct your audience it’s the context of the visual that you’re choosing right that is going to provide them an

12:06 right that is going to provide them an idea of yeah I’m used to a line chart typically when I look at this this is going to be something changing over time right yes I’m looking at a bar chart I’m going to categorize and group things by These Bars it’s when we get fancy sometimes as business intelligence folks where we either change the norm of one of those visuals like trying to do categories in a line chart right which you’ve seen before which instantly makes somebody go wait wait what’s happening

12:36 somebody go wait wait what’s happening here I’m losing context like what I’m supposed to be looking at as well as when we try out new visualizations that’s true the importance that you probably need to add a lot more context because they don’t have that that like mental model of what goes on in that visual wow that’s really big Seth does know it all that that’s it right there [Laughter]

13:07 I’m just saying but that’s just very like I didn’t I didn’t exactly pull it out of the article but that is extremely profound profound and I think that’s only the first step too right like the mental model of the line chart or a bar chart that’s just to get that’s just to get them to the first step of I can at least comprehend what it’s doing there’s that whole other side of it talking about context of is this relevant to me is this important to me where yeah okay I know this is trending but now I have to make that decision on

13:37 but now I have to make that decision on what like what actual value am I going to get out of this what is good what what is bad that’s another thing we don’t have to be fancy to do that but it’s part of our job too whatever we’re displaying or whatever we’re communicating yeah that we’re all saying No this affects you in this way or this is what you need to be aware of especially when we launch a report like okay look here are all the numbers and guess what in a week you’re still going to be looking at the next

14:07 going to be looking at the next year and a half so what do I need to do as a user what what do I need to do that’s going to make me coming back yeah and I you’re bringing up an interesting point Tommy if I think about the different ways to add context to things or adding context a lot of this maybe this is why maybe this is why people ask for tables of data when you’re starting to look with reports to a new audience maybe those tables of data requests are coming

14:38 those tables of data requests are coming from a place of one I don’t understand a ton of visuals I’m used to just doodling on the data the way I want to but a table has it’s almost like a blank sheet here’s here’s all like if you think about like a wide table with lots of columns and lots of data in it in rows that that is something that has not been predetermined here’s the information you can you can design your own insights on top of that it’s very much of a blank canvas type approach when we start applying a comparative chart or

15:08 applying a comparative chart or historical line chart if we start putting the time on the x-axis right we start now making assumptions we start using that information to start communicating about a certain way and that may not be the same story the user wants to use therefore they’re asking for a table yeah I guess what’s what do you guys think about and maybe this is a question for the audience too right right one of the last points he made mixes like hey you just by throwing a bunch of annotations and we’ve talked about like information icons and pop-ups

15:39 information icons and pop-ups things to provide context to a specific visual visual he’s he he references that as like oh like you can’t just assume that you throwing that on to that visual is going to provide the appropriate context to a particular end user and I found that to be true especially as it relates to some of the reporting I’ve done in the past right where you have teams of people that’s been an inordinate amount of time yes especially when it’s customer facing right like you go through this battery

16:10 right like you go through this battery of like okay what are the questions we’re trying to answer for them right like what are the important key points we want to highlight and there’s so much work done before you actually build a report can you actually provide that level of context in just the report and Report visualizations or is does that also push this need to have a context page which I’d never see and I can’t say that I generate on a on a basis either yes is

16:41 I generate on a on a basis either yes is that something that we should be doing for a report that is not like just out of the box or like four domain area where they know exactly what they want they’ve requested it Etc because it it I think does the other important thing too which is Define the box that this this report yes thing solves or yes it should answer because the other challenge with this is instantly you have a lot of end users that start to try to stretch the report

17:11 that start to try to stretch the report into all areas yes right well you’re showing me this data it must mean I can look at it all the way through here and you’re like no oh purpose and intent of this report I. E the context by which we were providing your information was this box yes and if you want this thing and maybe that would help us like differentiate between what gets added in this report versus what’s a new one should that be part of our our deployment of certain reports is

17:41 deployment of certain reports is providing a contacts page that that gives more insights I don’t know what do you guys think I think I think again I’m gonna I’m gonna lean on the the ever Consulting question here it depends right the answer you mean the answer is it depends and I’ll get I’ll give you the more yeah

18:04 and I’ll get I’ll give you the more yeah well I’ll give you the more elaborate answer too in certain situations I would agree the extra context is very much needed and also I’ve even thought about or I’ve built some reports where here’s three screenshots of the way this chart could look this chart means this in this situation this chart means this in this situation so really really going to been a lot more deeper around okay here is the Insight here’s the here’s what I’m trying to convey with this Visual and I think a lot of times we put

18:34 Visual and I think a lot of times we put too many visuals on a page just there’s just too many and one it’s just over it’s overwhelming right if you had and when I talked to my team to build visuals or we’re working with clients to build visuals it’s literally I will write the Insight at the top of the page here’s the inset what we’re trying to build what is my sales over time and I’ll experiment with a couple visuals but maybe that page gets one visual on it when we’re talking the development phase and then once we work with the client and figure out how things are working

19:05 things are working then maybe we’ll distill that single visual down to another page that maybe has two or three other visuals so but I think if we add too many things it overwhelms people so I think less is more in this situation I think being to your point Seth I think it’s very important for us to be clear about what questions are we answering and if I look at other platforms that are doing a lot of analytical things one of them I look at the YouTube analytics it’s just what I do they’ve got a really good way of saying hey here’s a little table or graphic of here’s the videos that are growing your

19:37 here’s the videos that are growing your audience here’s the videos that it’s a very specific single like a point and explains that and now I’m actually seeing other companies now they’re they’re going into like like they’ll again YouTube’s another good example they’ll they’ll have videos from directly from the YouTube team that says hey here’s how to grow your video here’s how to grow this audience and then they’ll correlate that to okay here’s the visuals that help you or here’s the data points that we would use to help you grow that area and this happens

20:07 you grow that area and this happens everywhere in marketing it could be in sales but I think what they’re doing though is they’re trying to say here are the factors that we use and see in the data data that you should be considering as you build things yeah so I think that makes sense when we’re talking about customers that you’re building reports for internal customers in your company or self-service or Team level reporting I’m not sure that’s always needed okay so here’s so I don’t know if I would put as much effort towards it but you say

20:37 as much effort towards it but you say that but at the same time like how often are new people coming into those organizations it’s it’s not disagreeing it’s an access thing though too it’s like they just want to see it every which way so I have a bit of a hot take on this what you said and feel free to disagree but I really do believe in the future yeah a single line chart over on a monthly trending has never helped anyone ever by itself

21:07 itself so if I’m looking at just a line chart and I’m looking at this over the last two years that by itself has not saved anyone any money by itself by itself what could that possibly tell you because in terms of when we’re looking at it from a two-year trending not saying that there’s not any value in it at all but by itself and I think we overcompensate with the amount of visuals and trying to show this whole time frame because we have the access to it but if I’m looking at two years of data which is 24 data points what can I

21:40 data which is 24 data points what can I possibly make from that that I am going to spend less money on hire somebody on so to your point on do we need to show all these things or do do we always need to be a little more hyper focused on leaning more and more that way initially there’s always got to be some reason what they’re looking for I can give you the general data dumps I can give you the our whole two three years of data data but that’s not what you need right now it’s like an over it’s like the

22:11 it’s like an over it’s like the person who buys the Ferrari because they’re they’re overcompensating for I don’t know exactly what I’m trying to do in this last six months let me let me I’m gonna I agree with you to some level Tommy but I’m also disagreeing with you at some other levels as well I would so I agree with you depending on the size of the information you’re looking at the amount of data that you are provided right if you sell thousands of products and you’re only providing entire company

22:42 you’re only providing entire company sales with 24 data points I would I would agree like if that’s all that we’re looking at I would 100 agree a two-year line chart this year compared to last year does not help you very much at all other than it gives you the largest level of trend and aggregate around around our sales are better than last year by what percentage or our sales are lower than last year by what percentage that’s basically all you can get from that however that bit of insight when to your point mixed with hey I’m gonna I’m gonna

23:13 point mixed with hey I’m gonna I’m gonna let you select one category of products at a time right because now you want the line chart but you now want the line chart inside of like small multiples right so which which product line drove the least amount of sales or which one dropped off the most from this year to last year what caused that problem so while on one hand I agree with you the line chart by itself doesn’t make you don’t walk away with a very single pointed direct action I do think the line chart helps you start visualizing

23:45 line chart helps you start visualizing or or observe absorbing multiple year-of-year data and you combine that information with other filter or filter context because and this is the knowledge I use a lot with people when I teach a class is is you think of your data as one really large wide table right that entire table is not important of it in and of itself all of the fully aggregated amount of a massive table with millions of rows in it maybe doesn’t get you where you need to be however however there are a handful of rows ten thousand

24:17 there are a handful of rows ten thousand a thousand that are telling you part of that story and that’s what you’re trying to go find you’re trying to weed out the data that is not telling you anything and focusing on the information that is telling you something I guess I’m confused like lines don’t mean anything without context is that the point to a point where I think we use the line charts a lot more than we need to they’re a great filler and we but I

24:47 they’re a great filler and we but I think a lot of times they’re not really not telling a story I would disagree with you there on that same but at the same time right like I your your pick you’re picking a con you’re picking a visual under the context of looking at it in certain ways what you’re saying a a line chart is not is not valuable for planning if I have a planning and forecast lines on a line chart isn’t that hugely valuable to understand over

25:18 that hugely valuable to understand over the last two years like where the trend is going is going no and I would agree but we’re giving a business yes but that’s the content you’re you’re weak like this is where I think we have to be careful like you can’t broad strokes and say line charts on their own don’t provide us any information that we can action off of I’m like no you absolutely can okay so let me put it this way I think there are a lot more overly used than probably they need to be zero no you’re dead on there’s nothing oh there’s not a use case at all but

25:49 use case at all but I think they’ll use a lot if you had said pie charts I would agree with you at the same at the same time though like overuse of visualizations I don’t think is the problem because a line chart is a well recognized and well understood chart that like the vast majority of people understand like when you’re putting an x-axis on there that’s typically a Time range when you don’t do that is where you get in trouble because it conflicts with people but at the same

26:21 it conflicts with people but at the same time is it bad that we use a bunch of line charts if that should be representing the information over time now I would agree with you and I think he dives into it which we we probably can start doing as far as like yeah there’s six essential ways to add context and that’s one history four particular ones in historical is one of those right so what is how do we provide the most value or context in using certain visualizations

26:54 you care more about than the other stuff like maybe maybe I don’t maybe don’t go through it one by one I’m not sure if I have enough time to go through all of them is there one that you that really resonates with you you want to start with we can start with comparative which is number one

27:04 is number one I don’t know I’m just I’m just throwing it out there I do find this graphic very interesting the graphic that he illustrates there is nothing but bar charts line charts and there’s this one called equivalent which is more like a ranking chart or ranked chart like that so even in his infographics relatable yeah it’s it’s also a little bit more like like there’s no pie charts here ladies and gentlemen there’s there’s no tree map in

27:35 gentlemen there’s there’s no tree map in these visuals so so I’m just just pointing that out I think those are when I when I see reports that show up and I see a bunch of pie charts donut charts and or tree maps in them I immediately think to myself or not conveying a message someone wanted some colors on the screen we’re just trying to flash it up a bit like that’s a little that’s literally my thought I’m like we don’t we don’t know what we’re doing thinking about data and analytics pieces like how can we how can we fix this a little bit anyways

28:05 we fix this a little bit anyways I I vote we go for let’s start with comparative that’s the first one on the list here let’s start with that one so the comparative data observation mode mode let’s talk about that one how does how does this one add more perspective so they want to kick us off on an idea there or what that one what that talking point is going after do I want to oh I think we were gonna we were going down that route anyway I just wanted to make sure we were picking the right one and I I derailed here quite heavily so I’ll

28:35 I derailed here quite heavily so I’ll come back I if if we want to pick up from the beginning right sure comparatively speaking right he goes through three different aspects direct indirect and relative so so to add more perspective he just outlining here it’s very common to share an item with other similar items for example you would compare the monthly sales one product those other related products but this form of context you can either emphasize how similar dissimilar the items results are in

29:05 dissimilar the items results are in comparison to others and the items in addition you may be able to compare the result to the goal or Target or benchmark benchmark I I think it’s interesting that like we’re talking about this context specifically within the data sets and and the visual is like maybe he’s saying we’re choosing the appropriate visual for these data sets that we’re trying to compare right but I I get a little lost in in the fact that it’s not a

29:36 in the fact that it’s not a recommendation to use specific visualization types he’s talking about the the data and US providing context through a report without just throwing information on the page well there’s also another another one that I reference a lot is and there’s actually a number of Articles around this I think Dan is putting some other items here from other people who are Dan thanks for the links for some things there’s some PDFs here that are very good another standard that works really well in a lot of these situations

30:07 really well in a lot of these situations when you’re talking about comparative things right again in in this article for contextualizing insights we’re not focusing on a single chart one example of that would be is a bar chart where you’re comparing different categories that does make sense you’re looking relative size scale of the different bars but there actually is a whole language there’s a whole bunch of other visuals that could be comparatively focusing focusing as well as well but the the category of comparisons is like a known item the ibcs

30:37 like a known item the ibcs standard international business communication standard I’ve said this so many times I now know the acronym like off the top of my head now so so the ibcs standard also does some very interesting things and I really like that standard because inside that standard you may have a bar chart and depending on how the chart is built there are indicators or dotted lines that indicate here’s the largest bar compared to the smallest bar and here’s the percent or the number difference across those things another observation

31:09 across those things another observation I would say here for comparative is playing you’re comparing things you are trying to think about a consistent calculation that you run across the different categories the dimensions right in this example talking about the direct comparison or the indirect comparison right I’m trying to compare things of different sizes or different categories or maybe it’s relative right I want to see how this thing performed and how much more above a different category did or did not perform those are deeper level calculations and they when you

31:42 calculations and they when you have a bar chart with very large numbers in it in it the the bar chart tends to look like a flat line because the variation between each bar is so little so you may use something a different type of charting to show you the variation of that information over time yeah if you do a little bit of math and say I’m going to talk about variances or differences from a baseline you now actually have a chart that actually can tell you some information you can actually make some more decisions on it so I really like the comparative one I

32:14 so I really like the comparative one I think actually I use this quite frequently as far as different categories that I use use what’s hilarious I’m going through this article and again all these things are honestly cheat sheets on questions and conversations you have with stakeholders on what they’re trying to look for that’s a good point because I’m going through like oh yeah because honestly and what a great place to start where if let’s say they don’t have a Target they’re like well we don’t we’re not comparing our

32:44 well we don’t we’re not comparing our teams against each other it’s like well why don’t we at least start with an average and see who’s doing better or worse than the average and we’ll show that based on those rates so you can see just to start off with something this is a great exercise for me it’s my go-to exercise or approach when I’m dealing with someone was like we really don’t have a Target we really don’t have like you Target we really don’t have like these it has to stay Within These know these it has to stay Within These boundaries so rather than it being open-ended on the visual side I basically say a little let’s let’s

33:15 basically say a little let’s let’s calibrate ourselves and let’s create an anchor on just trying to understand what is good and what is bad for you’re looking at looking at number of calls handled or you’re looking at the your success rate whatever the things are let’s just try to understand where do things stand and then you can understand hey well why are these teams doing so much better and then and why is these two so much worse so it’s this is one of my best exercises to immediately start getting

33:45 exercises to immediately start getting the someone to start thinking about things in context in a little subset of time but it is the most powerful not just from visual but honestly from the conversation View and understanding what will trigger someone yeah I I think I’m grounded now like as I’m like quick perusing through here the thought the thoughts came back it is this it is the precursor to to visual selection almost right yes it is and I like I like the as you read through the six essential ways right

34:15 through the six essential ways right it’s the thinking through the meaning of the data the data and what it is you’re trying to convey to an end user and there will be certain visuals that do that better than others right and I instantly went into like yeah we would be choosing the certain visuals that we’re all familiar with but if you’re not right I think this is a good way to get into or at least add contacts to visual selection is is at like understanding

34:46 at like understanding what it is you want to convey and then you can check choose which which visuals best represent a comparative analysis or a historical right so diving further into these I think he represents certain scenarios in each one of these that are pushing you in a certain direction or allowing and end user to lock into the the particular visual that best represents that context of information I think the ones that stick

35:17 information I think the ones that stick out to me the most and maybe it’s just because we talked about tool tips and like where we would want to go in the last episode was the informational or comp con confirmative because you what sticks out to me is like you have these data points right that are could be perceived as outliers or like and I think a lot of people like think about outliers as like the highs but what about the dips like the severe

35:47 but what about the dips like the severe dips of like okay well we if you think about systems system Health all of a sudden you’re you’re traveling at this 98 percentile whatever and then one day boom dips down and then back up more often than not an end user may just be like well that’s an anomaly that doesn’t make any sense yep and and that’s where

36:06 make any sense yep and and that’s where I I was thinking like man it would be great to have that like side tool tip helper where I could correlate a point and maybe even push out a color right of a different like I’m looking at things over time boom it drops hover that and have it have an informational context pain on the side that provides information that says hey this this system down effect happened blah blah blah blah there’s no recovering this data communicate to the customers that XYZ

36:36 customers that XYZ so that would be extremely helpful right it’s interesting you mentioned that because because there’s there’s two things in desktop that get you close but I think what you’re looking for is something like a combination of the two features like there’s one feature right now you can do anomaly detection on a visual so it’ll one It’ll point out the boom or the high rise or the lower it’ll figure out based on the time series of that data where the anomalies are occurring okay good great it’s doing that half of what you want but the analysis part of

37:06 what you want but the analysis part of the why the why that’s not very well articulated but now it also it also wouldn’t be something that you would want like float like having this huge hover thing floating around all the time and that’s yeah I love Tommy’s idea of like on the service side where it represents that panel like man yeah give me the fly out and whatever I can add that context of paragraph of information correct it is the only way you’re going to understand like what is going on here and then the other thing I was thinking here when you were talking about that Seth was this is

37:37 were talking about that Seth was this is I think potentially where goals or power bi metrics come into play a little bit because you can take a Time series from a data chart you can represent that on a Time series and because there is this ability to be able to make like notes and or comments against it right you can have some investigation and there’s to your point Seth there is another layer of human based context that’s being applied to that time series oh I can say

38:08 that time series oh I can say our number was really low because of XYZ thing right here here’s here’s the reason why we dug into this and this is why it went low or went high and maybe this is another area where you could use some reporting inside teams right have that conversation that that chat thread you’re embedding the report into a know you’re embedding the report into a team’s team’s environment when you’re talking about hey look I found this thing on this page here’s the issue if I click these options it appears to be the problem comes from here but to your point that conversation that chat thread that text

38:39 conversation that chat thread that text the the Deep dive into why that point went up or down means something and there’s not in normal reporting unless you’re doing all that work up front there’s not a lot you can do to add that context after the fact like does that make sense though like the only tools that lets you like store that data against an arbitrary time series or data point in general is the goals or metric I keep calling goals they’re metrics that’s actually that’s a really good

39:09 that’s actually that’s a really good point too because as much as the functionality out of the box with power bi if you want to do provide that little areas of con context you are have to create a lot of measurements up front and you have to be very direct if anyone just proved that it was data goblins with the bar chart online chart I don’t see if it looks to the model tons of visuals tons of measures to do that but to do all the things that he

39:39 that but to do all the things that he did yes but the big point there is and you downloaded his let’s go back to data real quick before we go further so data goblins did a in a two-part series basically hey here’s a bunch of way you can build line charts or actually think it was bar charts initially right the first article was again here’s a bunch of way to build bar charts and he had I don’t know it was like a three by three grid nine twelve I don’t get my four by four grid there’s like a four by four grid of just 12 different ways to use a bar chart and then you download the file right is that what you’re referring to and then in that downloaded file there’s

40:10 and then in that downloaded file there’s a ton of additional measures that he used to do maths to get to those bar charts and make him do what he wanted right and we’re not going to do that on our own for any official just in case it’s usually because I to set point if that outlier is something that’s going to that is already a part of the conversation then I’m going to be very direct in creating those types of types of highlights and this is where I’ve been LED myself to oh come on

40:40 on I’m finished there’s an idea here you you went too fast I want to slow down here just because I think what you have is a really impactful idea I know right data goblins did this amazing visual build the stuff that we’re bringing to that visual is a category and a number a bunch of categories a bunch of numbers that’s what we did and the fact that you have to write so much extra Dax to get to these different visual Styles in my opinion is a Miss by

41:11 visual Styles in my opinion is a Miss by Microsoft Microsoft like I should literally be able to say here’s 10 different types of bar charts if I just click on one of these things I should be able to drop in any one of those 12 items for stylized bar charts and have the Dax like hey power bi I’m gonna give you this category in this number it’s a table of data it’s a two column table produced to me the the lollipop chart the bar chart with the indicators like

41:41 chart with the indicators like I should be able to give power bi less information about it and it should Auto generate me more of the visual side that’s just a huge Miss on micro if Microsoft if you’re listening all three of you of you maybe maybe you could help us out there like this is what the community is doing and and there’s a larger barrier for that beginner business user they know what they want they may want to get that type of chart but they can’t do it easily because there is no easy button to build it and there’s a larger hurdle to get people to generate that type of

42:11 to get people to generate that type of visual that’s that’s the point I wanted to make Tommy I think you made a really good point there and I feel like this is a Miss on the product side to some degree is we don’t have that Auto building Auto creation side of things inside the visual space and you you are a thousand percent correct but I I do want to say I’m not it’s not because I’m saying that because I’m frustration it’s that that more what a lot of people don’t realize is okay we you maybe created your base measures well you shouldn’t stop there because again that context is usually

42:43 because again that context is usually something that we also need to create for too and if we understand like oh so now that we’ve been looking at this you really don’t like things when they’re when they fall between this 20 80 percentile okay so that’s where we’re going to focus a lot of from creating a visual side of the measures and creating our measures to say okay let me just see things between these like these pseudo anchors we have I will say though with doing and Dax them it’s so Dynamic

43:13 with doing and Dax them it’s so Dynamic and you can reuse it rather than if I had to set this for every visual yes but the the biggest Point here is we we generally just stop with okay I’ve created the number the measurement here so now let me just focus on the visuals well if I have a really good idea of someone’s anchors and where they’re calibrated for a certain metric then I should be able to create a few measures that are going to support that from a visual point of view and I’m just my mind right now is

43:44 and I’m just my mind right now is melting right now Tommy I’m I’m looking at these six essential ways comparative historical scaled confirmative equivalent and informational and I’m now in my mind trying to say okay if I was going to build these visuals what data do I need and what report visual measures would I need to produce to create those and where my mind goes is where do those measures live do I’m already thinking like I have this model with your point Tommy

44:14 I have this model with your point Tommy here’s some basic calculations this is the basic stuff I’m going to build a report on top of that so I want measures experimented with or built maybe in the thin report so those thin report measures are going to be but more specific to the visuals that are represented on that page but when do we make the decision to like reuse them do we want to give those customized measures to my gosh so many thoughts are going on in my head right now how do you

44:44 in my head right now how do you distribute how do you communicate this measure is for this bar chart built this certain way it’s in my report do I push it back to the centralized model does it make sense to go Upstream would someone else even know how to use that visual with that measure in it they may not and so so now we start talking to like you point there to talk about reusability of measures and things I I would love them all to be reusable

45:09 I I would love them all to be reusable but there’s more there’s even more context that you need to produce around the the individual measure formula the the the the Dax expression just to communicate to someone okay here’s this is literally only used on this one visual or this measure is only used on the header the dynamic header of this visual visual where should that live and when do I want to reuse it my friend if that is my worry because I’ve gotten the stakeholders to be that targeted on their their success and failures of a

45:41 their their success and failures of a measure and that’s honestly the least of my problems because I know that I’m gonna have a great report because I have someone who’s really invested and also helps me out from the build where I know exactly where their their their their pressure points are if I know the pressure points I will create those report level measures and have a ball with it but that’s the hardest thing and okay sorry I went off on a total random tangent on that one but that was

46:11 random tangent on that one but that was just just I had a lot of light bulbs clicking off at the same time there measure context not not just filter contacts for the measure but literally why does the measure get used how do you appropriately build that and and communicate that to people either in the thin report or in the central model and how does that support a confirmative or a scaled up and down visual comparison that you’re doing how does that work that’s impressive it would be that’s a deeper discussion

46:42 it would be that’s a deeper discussion for another time I think maybe another episode I I think so I think so because it’d be worth exploring from a mental model exercise like hey could could we actually build something that had standards and practices around measures and context or like to scale out on this idea but in a developer way I want to tease the idea what like what if what if I could give power bi what if there was a tool out there and I’m thinking that potentially

47:12 there and I’m thinking that potentially also AI based things right so take Kurtz 12 bar charts take Curts 12 line charts that he did another a week later right those two Graphics right here’s all these patterns and for each of these patterns here’s a list of standard measures you would need to support that Visual and that visual pattern what if you just said to an AI machine said hey I care about this category I care about this measure in that context just represent me seven or eight visuals I put I click on

47:43 seven or eight visuals I put I click on the bar chart type it just produces all the visuals for me automatically with the appropriate measures arbitrarily created created and then I can say do I like it and then you hit yes and then boom it just shoves it into your report drops the visual on the page creates the measures that you would need to produce to make that visual work boom like that would be a huge Time Saver and then you can use AI to explain the visual like throw the image back at the AI and say hey look here’s the produced

48:13 AI and say hey look here’s the produced visual what is the inside of this what does this visual tell me and the AI should be able to I’ve got to believe that at some point someone’s going to throw images at Ai and say explain this chart and then the AI just says well it’s here’s how I would read this chart writes out a little text thing so now you have the the AI of stuffs and this is where I think AI should be more prevalent inside power bi is it should do all the stuff I don’t want to do I don’t want to write how this measure works I just want to write the Dax move on a I should explain it because it communicates to the next

48:44 because it communicates to the next person here’s what this is trying to do but here’s what it’s trying to accomplish that stuff should just be automated because that’s just it’s not a waste of my time but I don’t want I don’t want to spend mental effort there I want to spend time on thinking like which visuals to pick and what’s the story yeah let me I’m getting very passionate and being and being visual right like if rather than I I think how much better would reporting be if if we were presented with the best options of visuals because each one of those right now takes time to build and you’re

49:15 now takes time to build and you’re probably not exploring all the options so rather than that like could our reports actually be better if it was like hey this is what your data would look like in these contexts of like here’s your 10 10 to select from exactly even as report authors going like oh wow oh yeah that is the best way to present this data yeah boom but I think you’re right AI like in general is it’s funny how it’s going after some of the hard technical engineering jobs it relates like hey I want to do this thing make it easy for me it’s like okay here boom it’s easy because it’s just code yeah

49:46 it’s easy because it’s just code yeah I think ultimately like wrapping up I think my final thought is like I think this brings to light that we should always be intentional about the choices we’re making when we’re developing a report whether that’s working from with the data from the very beginning working with choosing which visualization and how to adjust it or what properties and themes we’re setting and overall like the layout all of this matters when we’re engaging end users in consuming data and and context in all of

50:17 consuming data and and context in all of these make matters I agree with that any final thoughts Tommy if you were wrap here no I this is I think way more something we can explore but I think this goes back to even if we have ai that can run through the model there’s still the skill that we need because everything we’ve been talking about is still the skill of being able to extract from somebody what’s important to them and then driving the focus because I wouldn’t know what to ask Ai and I don’t know what to build myself as a human

50:48 know what to build myself as a human in the first place either so the skill that I have my ability to extract now that Target and calibrate someone that’s where where to me this the whole this whole thing’s going going so I have to give this definitely comment very much homage here in the end here the episode so Enterprise AR says really those most serious question we have to address here is when the AI is being used in your power B reports should the AI explain it in the voice of Mike Tommy or Seth like that that’s

51:21 Mike Tommy or Seth like that that’s really what you should if you want the Italian spin if you want the word cannoli in your explanation that would be Tommy like let’s imagine I’ve got three cannolis and then you take away one here’s what this is telling you but it was a really high on Wednesday wow hey man wrestle so you’ll get a lot of Italian food references I believe if you had Tommy do it what were you doing last night I voted for Seth’s voice just because Seth does know everything so like my kids already acknowledge this part of this so it it Donald

51:52 this part of this so it it Donald wants the AI to argue with itself well if it if we train the if we train the AI on the podcast it’ll give you no answers just lots of words that just run around in circles all the time because that seems to be what we do anyways very good thoughts I really like this article Brent I’ve been really enjoying the content you’ve been putting out very nice job really appreciate the content very good food for thought and helping me conversate around data data things and what visuals we’re picking so really

52:23 and what visuals we’re picking so really like that as well so for anyone who’s been watching the blog or what are we doing now we’re watching the podcast geez Louise I’m I’m I needed more coffee apparently so everyone’s yes the word conversate did come out of my mouth I guess maybe that’s a word I’m not sure if that’s a real word Wikipedia that one later thank you for listening today we really appreciate your ears we know your time is valuable we hope you got a couple laughs and enjoyed some some food for thought here as well if you like this we really would love to have your

52:54 this we really would love to have your help we leave other people who know about what we’re talking about here it helps grow the community and honestly I love watching the chat here I’ve been doing a couple recordings having people communicate in the chat it makes me laugh you guys are hilarious this is so funny so I really enjoy the chat team here helping us out and adding some more conversational thread there Tommy where else can you find the podcast you can find it on or you can find us on Apple Spotify wherever you get your podcast make sure to subscribe and leave a rating and it helps us out a

53:24 and leave a rating and it helps us out a ton if you have a question an idea or a topic that you want us to talk about in maybe a future episode we’ll go to powerbi. tips slash podcast and you can leave a question and to your heart’s desire finally join us live every Tuesday and Thursday every 7 30 a. m every morning except when my computer doesn’t work there we go well thank you all very much and hopefully we’ll clean this up for the next one we were a little bit rough this morning Tommy didn’t even show for the first

53:54 Tommy didn’t even show for the first part so we’re gonna definitely Razz Tommy by the time we get to episode 300 we’re definitely have a big major episode I believe Tommy now owes Tom Seth and I Stakes just for the heck of it after that one so yeah anyways we’ll see you all later appreciate everyone see you next time

Thank You

Thanks for listening to the Explicit Measures Podcast. If you enjoyed this episode, please share it with a colleague and leave a rating or review—it helps more people find the show.