Zbieramy dane jak punkty w Żabce. A decyzje? Na „wydaje mi się” | Maciej Lewiński

0:00I would say that this is the fault of

0:01gas stations and all the Żabka stores .

0:03All right , but if that went up by 20 % ,

0:05what would you do ? Every single social

0:08media platform today hates your links ,

0:10because when it sees your links , it

0:12cuts your reach . And what does Google

0:14do ? It does exactly the same thing . And

0:16what does that change ? No , because when

0:18you go to a Żabka , will you be able to

0:20buy your favorite drink or chewing gum

0:22with that , right ? Will the cashier ask

0:24you , " And how many users do you have on

0:26your website ? " So , to put it bluntly ,

0:28we have to start hooked on paid traffic

0:30to keep the business running , and

0:31that's why it's sometimes better to

0:33make decisions based on " I think so . "

0:42Hi , I welcome you to another episode of

0:44the Date with Data Talks series , a

0:46podcast from data practitioners to data

0:48utilization practitioners and vice

0:50versa . Today , my and your guest is

0:52Maciej Lewiński . Hi Maciek .

0:55>> Hi , good morning . I guess we'll be

0:57talking today like analyst to analyst ,

0:59right ?

0:59>> I hope so .

1:00>> Exactly . I would like to start a bit

1:02with what we are facing . You're

1:04probably facing it in analytics for

1:06sure . We are really struggling with the

1:07fact that we have a massive amount of

1:09data . Today , data isn't the problem in

1:12marketing . Rather

1:14>> the saying that " you can't have too

1:16much of a good thing " doesn't apply

1:17here . And I wanted to probe you a bit

1:21on how it looks from your perspective ,

1:23that despite everything , there is still

1:24this " I think so " approach to data in

1:26marketing , right ? That we have a lot of

1:29data , but we don't fully use it . Why

1:31does that happen ? Or perhaps you have

1:33this observation yourself ? How does it

1:35look on your end ?

1:36>> Oh Mariusz , thank you for the

1:37invitation . First of all , I actually

1:41have the same observation that we live

1:43in very interesting times where we

1:45aren't short on data . But it's a kind

1:49of paradox that concerns me , you , and I

1:51suspect the whole market — that we have

1:54so much data . Data storage costs are

1:58>> significantly cheaper than they used to

2:00be . The ways of processing this data ,

2:02even using artificial intelligence ,

2:04also have very low costs , yet people

2:06still like to make decisions based on ,

2:08you know , butterflies in the stomach ,

2:11what their brother-in-law said , or " I

2:13think so . "

2:15>> And the older I get , or rather the more

2:21experienced I am , the more convinced I

2:23am that it's not one factor that makes

2:25" I think so " reign supreme , but a sum

2:27of many different things . Right , that

2:31grams add up to kilograms , but if I had

2:34to point out and name that one decisive

2:36gram from my perspective , I would say

2:38it's the fault of gas stations and all

2:40the Żabka stores .

2:44>> That sounds very intriguing , what you

2:46said . Because I have the impression

2:48that they really conditioned us in

2:50terms of hoarding — not collecting

2:52points , but hoarding scraps — and we

2:54collect this data for later , thinking

2:56there will be some super catalog where

2:58we can exchange this data for a cool

3:00product that will be useful , let's say ,

3:02at home or in everyday life .

3:04>> Mhm .

3:05>> And I get the impression that marketers

3:07approach data , despite everything , on

3:09the principle of such hoarding . The

3:12only way they use this data is to

3:14describe what happened yesterday . Mhm .

3:16>> Uh , so they want to have a somewhat

3:18prosaic control over what happened in

3:20their marketing , because they want to

3:21know what happened on their website

3:23yesterday . It's a bit like parents

3:26picking up a child from some

3:28educational dump and asking the

3:29question : " So , how was school today ? "

3:32>> Yeah , that's kind of how it is .

3:34>> Well , what kind of answer are you

3:35expecting ? No ,

3:36>> good ,

3:37>> good . And that's it , period . And you

3:39have control over what happened , and

3:40it's similar in marketing . I have the

3:43feeling that the intention behind

3:45collecting data is a bit different than

3:47what we — meaning you and I — would

3:49think . Because it seems to me that

3:52>> it would be key if we changed the

3:54perspective of thinking about this data

3:56at all , even towards what I want to

3:58change in my business with this data ,

4:00right ? Why am I collecting this data ?

4:02>> So , start with Why , right ?

4:04>> Start with Yes . Simon Sinek would

4:06high-five us here and say : " Good job

4:08guys , maybe we'll add an affiliate link

4:11so we can collect something from the

4:13podcast . "

4:13>> He'd be surprised by a Polish podcast .

4:17>> So I bet that it's just a matter of the

4:20wrong — a wrong mindset when it comes

4:22to approaching data , right ? We don't

4:25think in terms of " what do I need this

4:27data for , "

4:27>> we just say , " I'll collect it , just in

4:29case someone asks me , to show something

4:31eventually , " and that is the key . That

4:33" something " and that " something " turns

4:35into something . That's why it's

4:37sometimes better to make decisions

4:38based on " it seems to me , "

4:40>> but that's my point of view .

4:41>> I really like your methodology , PMA for

4:44short . Question , Metric , Action . Yes , I

4:47spelled that out correctly . And it

4:50seems to me that this action is key

4:52here in how you work with clients .

4:56Maybe tell us how you implement this

4:57methodology , because I have the

4:59impression that this action is a bit of

5:01a solution to this only looking at the

5:03past , right ? Oh , you know what , again ,

5:08based on my experience

5:12>> It really irritated me how clients

5:14approached analysts in a purely

5:16tool-based way , like , " Are you the

5:18person for that Google Analytics tool ? "

5:20" Yes , I'd like an interesting report . "

5:22I ask , " But what kind of interesting

5:24report ? " " Well , one I can show to the

5:25board . " " And what will you be discussing

5:27at this board meeting ? " " Well , about the

5:29latest campaigns . " " And why did you run

5:30those campaigns ? " And that was the

5:32digging — just making a report for the

5:34sake of a report , art for art's sake .

5:39So I decided that doing such repetitive

5:41work completely misses my calling ,

5:43because I always wanted to have an

5:45impact on something , you know , to help .

5:49Yes .

5:50>> I'm from Poznań , I'm not a spendthrift

5:52, I'm thrifty and I have a rather

5:54specific approach to spending money ,

5:56and that is why , among other things , my

5:58methodology was created , which is meant

6:00to guide people so they make some kind

6:03of decision based on data , right ? So I

6:05don't make a report for the sake of a

6:07report , but I make a report to answer a

6:09specific question .

6:10>> Even if I don't know the specific

6:11answer to that question , I have to form

6:13a hypothesis .

6:14>> Mhm . I have a point of reference , and

6:17then I perform a test , and having this

6:19test I can expect two solutions , A or

6:22B. And depending on whether it's A or B

6:24, I know what to do next . And that's

6:27more in the direction my methodology

6:29pushes , right ? So it wouldn't be just

6:31data collection , but rather they would

6:33be definitely more useful .

6:35>> Okay . So that question , " why , " that

6:38question of what the data is for at all

6:40, and you solve it , you solve it in

6:41such a way that ... ...

6:44>> someone says " I want a report " or " I

6:46want some data , " and you then , for them

6:48, move to conversion . I say , " No , I

6:50don't make those kinds of reports . "

6:52>> Okay , but I understand that you sort of

6:54keep asking " why , "

6:55>> right ? And then that " why " leads to

6:58that action later . Yes . Well , because a

7:01metric , I assume , is that report , so

7:03there is the " why " question , the metric

7:05, and the action is the deed . Then

7:11>> to make it even clearer for our

7:12listeners , it happens very often during

7:14consultations that clients come to me

7:16with some report , they show me a report

7:19or a dashboard , and I always ask about

7:21any metric that is in that report . A

7:25question . " All right , but if this grew

7:27by 20 % , what would you do ? "

7:29>> Super ,

7:30>> right ? And that is simply something

7:32that makes me see the other person

7:34suddenly pause and say ... ...

7:36>> start thinking , right ? And what this

7:38metric is even for .

7:39>> And I say , " Well , if you don't know the

7:41answer to that question , then why do

7:43you even have that metric in the report

7:44? " And that comes back to my roots in

7:46Poznań a bit , right ? To minimize the

7:48share of these numbers we are flooded

7:50with from everywhere , and only have

7:52those that are useful and move us

7:54forward .

7:54>> But isn't it the case that people

7:56prefer to know more rather than less ,

7:58in the sense of a report : " Oh , let's

7:59throw this in , and let's throw that in

8:01too . " And later , it works inversely ,

8:04proportionately to the knowledge that

8:06flows from it . I suspect that it is the

8:09case that they are flooded with a very

8:11large amount of information . They don't

8:13really know what to pay attention to in

8:15order to take a first step , because

8:17when they try to take that first step ,

8:18they lack certainty , because some other

8:20metric is screaming for attention and

8:22saying , " Hey , maybe you're thinking or

8:24maybe it's different than you think . "

8:26>> Okay . And when you lead trainings ,

8:28because you do a lot of trainings ,

8:30right , what do your clients ask about

8:32most often during these trainings ? Is

8:35it more about the metrics themselves ,

8:37like you see they don't understand them

8:38, or is it more about how to use that

8:40data ? How does that look ?

8:44>> Last week I led a two-day training , and

8:46I have a rule that I always ask my

8:48training participants what specific

8:49problem they would like to solve with

8:51this training , right ? What they

8:55struggle with in their daily work , in

8:57working with marketing data .

8:58>> Mhm . Exactly . And what are the problems

9:01here ?

9:02>> Well , if I were to generalize , I would

9:04say that there is one most frequent

9:06problem , which totally surprises me and

9:08also disrupts a certain peace of my

9:10internal analyst inside me .

9:13>> It is mainly a problem related to

9:15regaining control over marketing .

9:17>> Okay . No , I mean all the questions

9:19oscillate around this topic : what to do

9:21, which campaigns work , which don't ,

9:22which should we pause , what should we

9:24change , what should we optimize , what

9:26to increase , what to decrease , how one

9:28affects the other . Those are the types

9:30of questions . And notice that these are

9:33questions of quite significant nature .

9:35Because we need to have someone on our

9:36site so that the person buys from us ,

9:38because that's why our website was

9:40created .

9:40>> But on the other hand , there are , like ,

9:43super few questions . I don't know ,

9:45maybe even no more than 5 % regarding the

9:49website itself . So the focus of the

9:53questions is mainly on how we acquire

9:55traffic to the website , because we are

9:57super convinced that our site is , you

9:59know , that it surely has all the holes

10:04plugged , right ? So that is , like , the

10:06first thing that disturbs my peace when

10:09I listen to these questions , I think : "

10:11Darn , how can these people not think

10:13broadly , holistically , and only think

10:15in terms of one single slice ? " And

10:17perhaps at this stage it’s okay for

10:19them , because once they find out what

10:21that control looks like , they will come

10:23to the second point , which is patching

10:25up that website , which is very often ,

10:27well , let's be blunt , a leaky bucket

10:29from which people leak out like water ,

10:31right ? Mhm .

10:32>> Um , so so these are mainly questions

10:34regarding m -

10:36>> marketing , from what I understand ,

10:38>> marketing , traffic acquisition . But if

10:42I were to point to one dominant

10:44question that has been repeating itself

10:46for over a year and a half , it is the

10:48question of why this data doesn't match

10:50up , right ? Between Google Ads , Google

10:52Analytics , our bank account balance ,

10:54our CRM , ERP , or anything else the

10:57company happens to pay attention to .

10:59>> Okay . Okay . That is the main question .

11:02>> All right . And what happens later

11:04during such a training ? I understand

11:08that , well , because these trainings are

11:10on Analytics , or do you combine tools

11:12during such a workshop ?

11:14>> Well , today it's impossible to

11:17>> without one

11:18>> talk about only one tool , so to be

11:20blunt , I lure them in by saying it's a

11:22Google Analytics training , but when

11:24they come to the training , they find

11:26out it's a much broader ,

11:27cross-sectional

11:29>> tool . This results from many different

11:31factors , for example , the fact that we

11:33don't have so much data , that this data

11:34is a kind of signal , that these signals

11:36need to be put together into one big

11:38whole , that attribution models , well ,

11:40it's nice that they exist . But the

11:41question is whether everyone , depending

11:43on who created the model , is a

11:45beneficiary of that model . So I show

11:49more simple solutions that don't

11:51require mega overthinking , but have a

11:53real impact on what will happen in

11:55their marketing , for example , starting

11:57Monday , right ? When they come back from

12:00the training .

12:01>> Super .

12:02>> In the context of acquisition , a lot of

12:03questions are certainly being asked

12:05today in the era of what we have been

12:06experiencing for a few years . And I , uh

12:09, zero-click marketing . I actually have

12:12the impression that you are the author

12:14of this term in Poland . At least

12:17that’s how it is in my mind . Uh ,

12:19>> that’s nice , thanks .

12:21>> And so , how is it , right , with this

12:23marketing ? Because here you talk about

12:25acquisition , but the traffic won't come

12:27, right ? Like , there is no point in

12:29talking in analytics about acquisition

12:31at the moment when someone answers

12:33their own need , or gains

12:36>> some knowledge even before they enter ,

12:38>> right ?

12:38>> And how do you approach this with

12:40clients ? Maybe within the framework of

12:41this PMA methodology ? Or

12:43>> PM ? Mhm . Well , maybe I'll start by

12:45explaining what Zero Click Marketing is

12:47.

12:47>> Very good .

12:49>> Zero Click Marketing is the impact of

12:51marketing activities without the need

12:53to visit our website . And that caught

12:56my attention 2.5 years ago , when I was

12:59analyzing many different breads coming

13:01out of different ovens , meaning

13:02different Analytics . And I always

13:05wondered why these breads always look

13:07and taste the same , because entering

13:09Google Analytics , 70 , if not 80 % of the

13:12traffic was traffic either from Google ,

13:14or from Google and Meta , and that's it .

13:17And nothing else was happening there .

13:18And it just kept growing more over the

13:20years . I wondered why other forms , you

13:23know , of marketing activities we know ,

13:25for example , posting on social media ,

13:27like organic ones

13:29>> or redirects from other websites , just ,

13:31I don't know , are becoming obsolete ,

13:33don't work at all , what happened ?

13:35>> Mhm . And when I started digging into

13:38this topic , it turned out that , you

13:40know , most companies are starting to

13:42have an allergy to blue links , because

13:44when you post them on the internet

13:45publicly , the algorithms cut the reach ,

13:49>> which makes these links not clickable

13:51and they don't send traffic

13:52>> to our website . You just need to do a

13:54simple test , right ? The same post

13:56posted to the same medium , once with a

13:58link , once without a link , and compare

14:01the reach , right ? That's that

14:02>> I experienced this on LinkedIn , I even

14:04tested it , and it's actually the case

14:06that those with outbound links sort of

14:08>> to make it more interesting . I

14:09experienced that today too , because I

14:11wrote probably one of the most

14:13important posts this year regarding

14:14measuring marketing activities ,

14:17>> because it's a bit of a digression . A

14:19lot is happening now in the context of

14:21AI and ChatGPT and advertising

14:22possibilities . Exactly . Yes . Mhm .

14:27>> So , I looked into this topic and it

14:29turns out that the vast majority of

14:30people in our industry are writing

14:32guides or already drawing conclusions

14:34on how to properly conduct these

14:36marketing activities . It's been maybe

14:40two weeks , right ? After two weeks , yes .

14:43>> To which I wrote a contrarian post

14:45stating that anyone getting into

14:47ChatGPT ads is making a mistake ,

14:49because probably no one has assumptions

14:51about when we will call this post a

14:53failure , because we don't have such

14:55assumptions , it's more like we launch

14:57these things just to launch them . And

14:59of course , nothing stands in the way of

15:01learning something by launching these

15:03activities . No , because you gain some

15:06experience ,

15:07>> but over time , that experience and

15:09knowledge starts to manifest as clicks ,

15:11impressions , and CTR in the ad panel .

15:14Ultimately , it ends up in Excel , and

15:15that experience in Excel starts being

15:17converted into performance , because

15:19someone will finally ask : " Alright , Mr.

15:21Maciej , what pays off ? " Yes ,

15:23>> that's right . We spent 3,000 PLN . What

15:26is this experience telling us here ,

15:27right ? And there is a fine line between

15:30experience and performance . And that's

15:33why I stirred the hornet's nest a bit

15:35by saying : " Hey , think about when you

15:38will consider this a failure , so you

15:40don't pointlessly spend 300,000 PLN ,

15:42only for someone from the finance

15:44department to tell you : ' Hold on , this

15:46isn't working at all ' . " No , because

15:49someone will have to take

15:50responsibility for that .

15:51>> And I had this gut feeling , I feel it

15:57so much in my stomach ,

15:58>> that there would be , you know , an

16:00avalanche of comments on the post . Yes .

16:02It will fly , so that when I leave your

16:04place , I'll pull up to the new Mercedes

16:07dealership and place an order for a new

16:09car . But as they say , pride comes

16:13before a fall . I decided that I would

16:17put a link to sign up for my newsletter

16:19in the post comment , where I want to

16:21delve deeper into this issue in the

16:22next edition .

16:25>> And what , it got cut ?

16:27>> Okay . I thought there used to be

16:30something like that on LinkedIn , where

16:32links in comments worked less

16:34effectively , then it changed , but

16:35either way , the trend is one , right ?

16:38These platforms

16:39>> well , you can't step into the same

16:41river twice , as they say , meaning

16:42everything is constantly changing and

16:44flowing . I suspect that these

16:46algorithms have also learned to look

16:48for links even in comments , and when

16:50they see them , they might not be as

16:52restrictive as in the main post , but I

16:54suspect they cut them a bit there too ,

16:56because I'm the best example of that ,

16:58at least on LinkedIn .

17:00>> But getting back to your main question ,

17:02zero-click marketing , right ? Well , more

17:06and more marketing activities are

17:08actually happening outside our website ,

17:10where we start to influence the user ,

17:12for instance by recording this podcast ,

17:15right ? It will be in video format ,

17:17where even if we put a link under this

17:19material , people won't click that link

17:21anyway , because the intention is rather

17:23to listen to what we have to say here

17:26>> in a given field , and not necessarily

17:28to visit our websites as a way of

17:30saying thanks . " Hey guys , thanks for

17:32the chat . " " We visited your site , right ?

17:34" Exactly .

17:36>> So , in the end , zero-click marketing

17:38causes us to build influence within our

17:41group . Trust , which is also quite

17:45crucial in making purchasing decisions

17:47or achieving business goals . So it is

17:50an essential part of marketing

17:52activities , isn't it ? That some

17:53marketing just happens , but it doesn't

17:55land on our website , and we need to be

17:57aware of that . Period . Every single

18:00social media platform today hates your

18:02links . Mhm .

18:03>> Because when it sees links , it cuts

18:05your reach , and what does Google do ? It

18:07does exactly the same thing . What’s

18:11more , this AI propaganda , I would say ,

18:14is very much in Google's favor because

18:16they show an AI overview , where links

18:18to sources are heavily minimized and

18:20don't really drive traffic to sites

18:22like Google Organic used to . Yes .

18:27>> So , in summary , the only channel that

18:30has an impact on your business today

18:32because it sends traffic to your

18:34website is , in theory , paid ads on any

18:36social platform or Google .

18:39>> In other words , we have to start

18:41feeding on paid traffic for the

18:42business to work . And that is a

18:45methodology that my Poznań - based self

18:48>> opposes . opposes . Exactly , because I'm

18:51not spendthrift , just thrifty , and I

18:53want to run marketing activities that

18:55have a very strong impact on the

18:56business . But you have to be aware that

18:59web analytics in today's world doesn't

19:01really mean that what isn't measurable

19:03has no impact . And there are actions we

19:09take , for example , on social media ,

19:11which cause someone to remember

19:12somewhere that we deal with a given

19:14field . When they have a problem in that

19:19field , they won't look around ; they'll

19:21just type in our names or our company

19:23names , knowing we handle these topics .

19:29Yes , besides , there's another important

19:31point to note : we don't live by clicks

19:33alone in marketing , because those

19:36clicks once promised — and we even

19:38contributed to this , Mariusz , in the

19:40early 2000 s — when we said , " We can

19:42tell which campaign brought us sales , "

19:45right ? But today , answering that

19:50question with 100 % certainty is

19:52practically impossible ; those days are

19:54long gone .

19:56>> No way . No way .

19:57>> No way . We have to be blunt about it .

20:01>> And now , if we stop expecting that

20:03ideal scenario from analytics and

20:05approach analytics in a way to check

20:07influences — that is , which actions

20:10impact something happening elsewhere —

20:12then I finally start to have control

20:14and direction , right ? A simple

20:18situation . I published a post on

20:21LinkedIn today ,

20:23>> so someone browsed the post , and if it

20:25reached someone , they definitely didn't

20:27click that link , right ? So someone saw

20:31my post . Then someone probably takes

20:33the issue I raised , like , hey , do you

20:35really need ChatGPT with ads for your

20:38business ? They'll type it into ChatGPT

20:40to consult an agent about what it

20:42thinks , because maybe I , you know ,

20:44planted a seed of doubt in them .

20:46ChatGPT will answer the question , but

20:49I'll still be on that person's mind .

20:51And then some recording will appear ,

20:53for example , like ours . Someone will

20:55watch it and say , " Look , Lwiński is

20:57talking about the same thing he just

20:59wrote about on LinkedIn again . " And

21:01when there's a board meeting , someone

21:03will type in my name , right ? They'll go

21:05to Google , type in Maciej Lwiński ,

21:07click on my website , go to my contact

21:09page and get in touch with me , and my

21:12Google Analytics will say :

21:13>> Google Search or Google Search

21:15conversion , right ?

21:18>> So , should I be doing SEO to get more

21:20search queries for the name Maciej

21:22Lwiński ?

21:23>> I don't necessarily have to have an

21:24influence on that . But how can I

21:26influence it by engaging in activities

21:28that affect it ? I have a feeling that

21:31the times of web analytics we have

21:32today are sort of coming full circle ,

21:34aren't they ? Because I have this

21:37reflection , back when I finished

21:38university years ago , 16 years ago , I

21:44worked as an econometrician , where I

21:46was looking for correlations between

21:48different parameters , right ? It was

21:53about outdoor advertising spending , TV ,

21:55individual channels , and I was

21:56forecasting the result , some conversion

21:58, I don't remember exactly what it was ,

22:00and we kind of went through those times

22:02where it was one-to-one . I see how much

22:06I spent , someone entered , there’s

22:07that click , I see what happened , and

22:09there was a conversion . But today , a

22:12lot of what happens leading up to the

22:13conversion isn't visible . That

22:16zero-click , meaning it happens outside

22:18of our view , in the minds of our

22:19audience . What we can do is collect

22:22data on the activities we undertake .

22:25Because you have data on , you know ,

22:27when you posted , what the reach is ,

22:29what the reactions are , not necessarily

22:31clicks , and then you correlate that

22:33with the final result . So I have a bit

22:36of a feeling that this analytics thing

22:38is coming full circle , right ?

22:40>> Yes . And to be more statistical here ,

22:43we have two heads that have said this ,

22:45so let's listen to a third head that

22:47has already taken a step forward .

22:49That’s Google , which recently added

22:52these connectors to Google Analytics 4 ,

22:54meaning the ability to plug in any

22:56other advertising platform like TikTok

22:58or Criteo , to send more signals — even

23:00to Google , my company — to your Google

23:02Analytics , so you have as much

23:04information as possible regarding

23:06impressions , so you can correlate and

23:08grasp the context of what influences

23:11what . Exactly . Yes ,

23:13>> so yes . AI is changing our marketing

23:16and data a bit from the , let's call it ,

23:18marketing side .

23:20>> Just to be clear , I'm not entirely

23:22convinced that AI is changing it ,

23:24because it seems to me that this change

23:26happened significantly before AI . AI

23:31might be that extra gram tipping the

23:33scales toward what has changed .

23:37>> Actually , as you talk about it , I

23:38thought that the mobile revolution ,

23:40which lasted from 2012 to 2000 , I don't

23:42know , 19 , because every year was " the

23:44year of mobile , " those were probably

23:46the beginnings , right ? Because suddenly

23:49it turned out that from one device ,

23:51read : desktop , where we had everything

23:53connected almost certainly , another

23:55device was added . Indeed , there were

23:58some ways to connect those users , but

23:59it started to get increasingly

24:01difficult to do . Later , I think from

24:05'19 , toward the end of the 2010s ,

24:07privacy began ,

24:09>> right ?

24:10>> Consent mode and , like , harder tracking

24:13.

24:13>> But it wasn't just consent modes that

24:15changed the way data is collected or

24:17the quality of that data itself .

24:19Because remember also that technology

24:21has changed , right ? Restrictions

24:22started to appear on the browser side .

24:25DNT directives and the like appeared .

24:27And so on , which causes the quality of

24:30this data

24:30>> to start getting smaller ,

24:32>> to start getting smaller and smaller .

24:35>> And perhaps this is also the answer to

24:37your first question . Why we make

24:40decisions based on " it seems to me , "

24:43because

24:45>> our faith in this data is getting

24:46smaller , because there is less and less

24:49data , and this data , being in the

24:50minority , is also starting to be of

24:52worse quality than it used to be . Well ,

24:55it definitely consists of two things ,

24:57right ? Quantity and quality , because in

24:59terms of quantity , I always sort of

25:01refer to polling research , right ? When

25:05you think about what percentage such

25:07polls on the popularity of political

25:09parties are based on

25:12>> and then when we look , let's agree ,

25:14they are really close , right ? There are

25:17percentage point differences between

25:19what's in the survey and what comes out

25:21later in the actual , I don't know ,

25:23elections or exit polls , right ? Well

25:26yes , in the case of a website it's

25:28probably hard to translate it

25:29one-to-one , but the mechanism of

25:31operation certainly provides that

25:33direction .

25:33>> Exactly . Yes , it's the same as Google

25:36Analytics . It was Sylwester at my

25:38training last week who said that based

25:40on what I told him about Google

25:42Analytics , he concluded it isn’t like

25:44the GPS in our cars that leads us by

25:46the hand and lets us turn off

25:50>> our thinking , but rather a kind of

25:52compass that points north , so we can

25:54head in the direction where rescue

25:56surely awaits us .

25:57>> Definitely . Since we’ve started

26:02talking about AI — and it is one of the

26:04elements of this zero-click marketing

26:05— if we look at it from our

26:07perspective as analysts , because today

26:09it feels , especially to business people

26:10, like that’s where we’ll find most

26:12of the answers , right ? And now , a

26:17question for you from your perspective ,

26:19right ? In the age of AI , is an analyst

26:21— because it seems to me it used to be

26:23that way — still just someone who knows

26:25the tools and knows SQL ? Or is the

26:29analyst today shifting more toward that

26:30business-oriented role , in the context

26:32of that first question where you talked

26:34about knowing how to ask the right

26:36questions ?

26:37>> So , what will happen with these

26:39analysts ? Who is a good analyst today

26:41in the context of these two worlds

26:43I’ve described ?

26:46>> I’ll answer that question perhaps

26:48contrarily , because I believe a good

26:50analyst was never just a tool-based

26:52analyst . To me , a tool-based analyst is

26:54not an analyst at all . Mhm . What I

26:58expect — or would expect — from an

27:00analyst first and foremost is not that

27:03they know Google Analytics perfectly ,

27:05but that they know the business context

27:07perfectly .

27:13>> This was received in various ways when

27:15I appeared at conferences or trainings

27:17and said , " Listen , even if you hired me

27:19or anyone else from the market who’s

27:21been in it for years , they wouldn’t

27:23be as good an analyst or draw

27:24conclusions as well as you do , because

27:26you know the business context , which is

27:29hard for me to earn since I need time

27:30to learn it all , right ? " Besides , you

27:36have better internal company flow and

27:38know who to call , while I’m guessing

27:40and sending a generic email just to be

27:42on the safe side .

27:44>> That’s why , for me , a good analyst is

27:47primarily someone who has that

27:48contextual knowledge . That’s the

27:51first thing . The second thing in the

27:54age of AI , which is coming up more and

27:56more , is that they are someone very

27:58skeptical about things like data

28:00quality . They question the data , saying

28:02: " Does it really look like that , and

28:04what would happen if ? " " No , these are

28:06the questions in that direction . So ,

28:08this is someone who is skeptical and ,

28:10to some extent , inquisitive — meaning

28:11they definitely have critical thinking

28:13skills , which means they question the

28:15results they get from AI . Because these

28:17AIs are designed , well ,

28:19>> to flatter us , at the end of the day ,

28:21right ? question . Mariusz , thank you for

28:23asking by answering it and so on . Yes .

28:28That is the second trait that I think

28:30such an analyst should have . And the

28:34third thing an analyst should have , or

28:37female analyst , oh boy , this ages badly

28:39, this is crazy . Meaning , they have to

28:43be brave , right ? Meaning , they have to

28:45be able to make decisions , point out

28:47the direction , whether we go left or

28:49right . We said it at the beginning

28:51today , right , that when we collect this

28:53data , it's used so that we know what we

28:55want to change . So we have this data ,

28:57we form a hypothesis , we have a

28:59benchmark , and we run a test . We verify

29:02whether it works or doesn't work . If it

29:04works , we know what to do ; if it

29:05doesn't work , we also know what to do .

29:07And the person who says what we should

29:09do is , for me , the analyst , meaning

29:12>> and then they have to check it too ,

29:13right ? In the end , to say whether , once

29:15it's done , in which direction , right ?

29:17>> Exactly . So for me , an analyst is

29:19someone who is usually in the marketing

29:21department . That is an internal person .

29:24>> Mhm .

29:25>> And that is a real analyst to me . But

29:27if we talk about the tool that is AI ,

29:30then obviously AI is a great tool for

29:32quickly finding answers to questions

29:34that bother us . The only question is ,

29:38>> do we ask that question well ? Do we ask

29:40the question well , are we sometimes

29:42biased , or does the AI , for example ,

29:44take into account the fact that the

29:46data might not be of perfect quality or

29:49might not be complete , which we are

29:51able to infer using context .

29:53>> That's true . And this

29:54>> so this human factor , the protein

29:56mechanism when analyzing data ,

29:59>> well , you can't fake that , you can't

30:01>> as of today , right ? Because I always

30:05add a bit when I get asked about how it

30:07will be with AI in analytics , that at

30:09least today I don't believe that it's

30:11possible to digitize the entire

30:13business context , because

30:15>> surely someone will appear in the

30:17comments to write : " Mr. Mariusz , it's

30:19just a matter of the prompt . "

30:20>> Okay , let me give you an example . Let's

30:23say you're analyzing your campaigns and

30:25it turns out there's a sudden drop in a

30:27campaign that lasts for three days , and

30:29then it suddenly bounces back and . The

30:33reason why I think no one is digitizing

30:35this today — if someone is , I’d be

30:37willing to pay them a lot to show me

30:39they’re doing it — is that someone

30:40might have forgotten , for example , that

30:42the credit card they had linked was

30:44expiring . And now AI gets access to

30:49such data and correlates it with

30:51something that happened at that time , I

30:53don’t know , in other systems it has

30:54access to , and it will draw the wrong

30:56conclusion . And so I say that today

31:02it’s hard to digitize everything

31:03happening in a company so that AI has

31:05the full context , and humans will

31:07currently remain , in my opinion , hard

31:09to replace in analytics . That

31:13>> is true . As of today , that is certainly

31:15the case . I don’t know what the

31:17future will bring , because on the other

31:19hand , I think that the more I dig into

31:21this AI , I know it’s a matter of

31:23teaching this AI , right , these skills ,

31:25and simply the longer this AI is with

31:27us , the more ,

31:27>> let’s call it bulletproof it will

31:29become , but we will still need a dose

31:31of skepticism regardless , like asking

31:33ourselves if this is definitely the

31:34right decision . Because the thing I

31:38learned in marketing is , above all , to

31:40speak the language of benefits , to

31:42understand the other party I am

31:44communicating some thing or value to ,

31:46that I am the solution to the problem

31:48for which that person should come to me

31:50, right , to communicate with me . And

31:56what AI will certainly do is extract

31:58some algorithmic , averaged output from

32:01it without that whole beautiful

32:03foundation , that sense , feeling ,

32:05emotion , empathy , which a machine will

32:07certainly not have .

32:11>> That’s true . That’s true . So it’s

32:14certainly not the case that AI will

32:15completely exclude us from our

32:17profession . Yes .

32:18>> I understand . Actually , I once wrote an

32:21article about AI in analytics , that

32:23analytics today is about gathering data

32:25, analyzing it , drawing conclusions ,

32:27and making recommendations . And now it

32:31seems to me that in the context of data

32:33gathering , meaning preparing data for

32:35business needs , there is a lot of that "

32:37why " involved . And now

32:42>> AI , I can certainly imagine , could be

32:44taught to ask the right questions and

32:46map out business needs . However , for

32:52now , it seems to me that a human will

32:54still do this much better , meaning on

32:56the side of what to put into this data ,

32:58what to put into these systems , they

33:00will recognize that better . And now a

33:06question in the context of AI again ,

33:07because for a while on LinkedIn I saw

33:09many posts from business people saying

33:11they spent three days with Claude and

33:13built themselves a single source of

33:15truth , connecting various systems like

33:17ads , ads analytics , and internal

33:19systems to ensure this data collection

33:21element . And what do you think about

33:25that ? Does it help us as analysts , or

33:27does it harm us a bit , taking away our

33:29work ? How does that look from your

33:31perspective ? About the fact that in

33:33three evenings I created a dashboard as

33:35the only source .

33:36>> Yes , exactly .

33:37>> Well , that's like asking , " Maciej , do

33:39you like sweet and sour chicken ? " No ,

33:42because what AI gives us is the ability

33:45to create technological solutions in an

33:47absurdly short time , for which we

33:49previously needed gazillions of dollars

33:52just to

33:53>> create them until now , and that is

33:56>> undeniable , and what AI does is capital

33:59.

34:00>> The second thing that stops me , humanly

34:02speaking , is why someone is doing this

34:04in the evenings rather than at work .

34:13I'm sorry , generally speaking , but the

34:15thing that definitely raises a red flag

34:17is the term " single source of truth , "

34:19because obviously , a tool that collects

34:22data from many different APIs or places

34:24is something we can deliver quickly

34:26today ,

34:30>> that's true ,

34:31>> but what we cannot deliver quickly ,

34:34even in those three evenings ,

34:35especially when we do it alone , is

34:37defining what the truth actually is for

34:40us .

34:40>> Mhm .

34:42>> Because from the company's point of

34:44view , when we have different

34:45departments , each department has its

34:47own goal and its own truth , right ? For

34:49some , it will be revenue , for others ,

34:52it will be profit . So the question is ,

34:54is it net or gross revenue , before or

34:56after returns ? When do we count those

34:58returns , who assesses lead quality , who

35:01influences it , how fast do we get that

35:03feedback , what is important to us ,

35:05right ? Whether they are new customers

35:07or returning customers . You know , a lot

35:09of questions start to arise , and a

35:11single person sitting in the comfort of

35:13their home in the evening , watching

35:15Netflix with one eye and coding in

35:17Claude with the other , simply isn't

35:19able to find the answers to those

35:21questions , so for me , it's more about

35:23introducing chaos into reality .

35:25>> So it actually adds to our work ,

35:27because that chaos , those strings will

35:29eventually have to be untangled . Yes ,

35:33but that's why , very often when I see

35:35such posts , I ask the question : are you

35:38sure ?

35:40>> Okay .

35:41>> In the sense that I still try not to

35:42say prematurely that this is definitely

35:44a good step . I have a concern with such

35:50an approach , that people who do these

35:52things will eventually start acting ,

35:53God willing , using this data , and start

35:55heading in a certain direction because

35:57the data points them there , only to

35:59find out that this direction wasn't

36:01actually good . And what , is it the

36:05data's fault ? No ,

36:07>> no , no , no . Wait , let's stop right here

36:10. Data is impersonal . At the end of the

36:13day , we must remember that the

36:15responsibility for a decision

36:16>> that's true .

36:17>> Someone has to take it , right ? So if

36:19that person created this single source

36:21of truth , they must take responsibility

36:23for the decision they make based on

36:25that single source .

36:26>> But in their mind , the cause or the

36:28reason is different . Do you think there

36:31will be reflection , where they think ,

36:32maybe the cloud didn't necessarily give

36:34me good advice , or maybe I didn't do

36:36this right ? I try to think that , after

36:39all , people are reflective , right ? And

36:41they are capable of saying , " Okay ,

36:43maybe I did something wrong here after

36:44all . "

36:47>> And I also have a second thought , that

36:49even if the data leads them in the

36:51wrong direction , that's okay , because

36:52it's already

36:55>> an answer to the question of what not

36:57to do anymore . So it is some kind of

36:59information that reduces our pool of

37:01potential , colloquially speaking ,

37:03fuck-ups that we shouldn't commit again

37:05.

37:05>> Mhm .

37:06>> So for me , doing

37:08>> is better than not doing .

37:10>> Yes . And it always has its purpose .

37:15>> That is actually a valuable observation

37:16, because as we were talking before the

37:18recording , today it's more comfortable

37:20for us to do what we've done so far ,

37:22because it will be easier to explain if

37:24it stops working . After all , we did as

37:28we had done until then , rather than

37:30stepping out of that so-called comfort

37:31zone and saying , " Okay , today I'm doing

37:33things differently and we'll see what

37:35comes of it . " No , like testing , but

37:39it's still about changes , like changes

37:41show exactly as you say , either it

37:42shows that I will learn something ,

37:46>> or I will win . I don't know , it was

37:48Nelson Mandela who said , I never lose .

37:51I either win or I learn . Yes .

37:53>> Only we still have a problem in our

37:55heads with the fact that this is

37:56learning , not a mistake . Yes , I feel

37:58that way , right ? With testing and

38:00entering some new areas .

38:02>> But this is probably cultural from the

38:04point of view of , it seems to me , Poles

38:06, right ? That we as Poles have this

38:07thing where it's either success or

38:09failure , and we don't think about

38:11everything positively , just in a binary

38:12way . Either it worked , or it didn't

38:14work .

38:15>> You always succeed , the question is

38:16just with what result ?

38:17>> Actually , I was talking to someone

38:19about Americans recently , that they

38:21don't have that , they have " I will . "

38:23You either win or learn , not lose . They

38:27don't have that L-word . For them , it's

38:29always a lesson . What is a generally

38:34interesting thing that has stayed with

38:36me lately is that the more mature I

38:43become , the more analog I get , in the

38:45sense that I'm starting to take notes

38:47not on a computer , but on a piece of

38:49paper

38:53>> and I have a small notebook where I

38:54write down quotes .

38:57>> Mhm . And these quotes are significant

39:01and important to me because , at first ,

39:03I interpret the quote very

39:04straightforwardly , just as I want to

39:06understand it . But the longer the quote

39:11stays and resonates with me , and the

39:13more I encounter it in my notebook , it

39:15sparks other reflections in me . And I

39:18was recently talking to someone about

39:20tattoos , and someone asked me if I

39:22would get one . Well , the answer to that

39:24question was no , but in principle the

39:28second question was a follow-up to keep

39:30the conversation going . " But if you

39:32were to get a tattoo , what would it be ?

39:35" So the latest quote I wrote down in my

39:37notebook was P of P is P ,

39:40>> meaning the price of progress is pain .

39:43>> Okay .

39:45>> I don't understand it in the context

39:47that it has to hurt for something to

39:49grow , but sometimes making that mistake

39:51is obviously not pleasant , not nice ,

39:53but it pushes you forward , right ?

39:56Because you learn something from it .

39:58>> Very inspiring , but even speaking in

40:00terms of self-development , it's true .

40:03>> I gave you quote for quote .

40:04>> Great . But I'm thinking , in a physical

40:07sense , right ? When you go to a workout ,

40:10the fact that you're progressing means

40:12you get sore , right ? Or just muscle

40:15pain . So indeed , it matters in mental

40:19or substantive development , and

40:21physical too .

40:23>> Definitely . For sure , where there is

40:25discomfort , where we feel discomfort ,

40:27we should primarily do that first ,

40:28because that's where growth lurks . Yeah

40:30.

40:30>> That's true . So , heading toward the end

40:34, I'd like to ask you , uh , because you

40:36have in your LinkedIn headline that by ,

40:38I don't know if I'll quote it right ,

40:40correct me , that by the end of 2028 you

40:42will train 50,000 uh

40:45>> 50,000 people , do I remember correctly ?

40:47>> Well , maybe not train , but teach them

40:49how to analyze ,

40:50>> how to analyze . Tell us , since many

40:53people are listening , can you reveal

40:55what level you're at now with that KPI ,

40:57that metric , or whatever ?

40:59>> Just under 42,000 .

41:01>> Oh , awesome . So you'll get there faster

41:04.

41:04>> Well , maybe . We'll see what the

41:06algorithms have to say , because let's

41:07be honest ,

41:08>> a large part of my mission is , after

41:10all , based on social media . I have my

41:12own newsletter .

41:12>> Mhm . But , you know , I also had a

41:16thought recently that a strong digital

41:18currency today is someone's email

41:20address , right ? Because it is totally ,

41:24you know , untainted by the algorithm . I

41:26send an email and it arrives when I

41:28want it to arrive , and with what I want

41:30it to arrive with .

41:31>> Whereas in social media , it's the case

41:33that these algorithms sort of dictate

41:35what to show to whom . On the other hand

41:39, we have to learn to write those

41:41headlines , leads , and lead-ins that

41:43will be a bit clickbaity because they

41:45need to interest you , because your task

41:47is to rack up comments inside , because

41:49if you rack up more of those comments ,

41:51the post will reach further . If it

41:54reaches further ,

41:55>> then we're dependent , right ? I balance

41:58a bit in my communication on a thin

42:00line between generalizing and being

42:02super precise , but that only stems from

42:04wanting those posts to reach as many

42:06people as possible . So perhaps , if

42:10nothing changes — though I suspect the

42:12algorithms will unfortunately change —

42:15then indeed by 2028 I will reach those

42:1750,000 people who will know what to do

42:19in their marketing activities .

42:22>> I think if you keep acting the way you

42:24have been , you'll be fine .

42:25>> Thank you . But I'd like to ask you ,

42:27from your observations , right ? Because

42:29you've been doing this for over a

42:30decade now , right ? How is it changing ?

42:34I asked you what they are asking about .

42:38But how is the need for this training

42:40changing ? Not in terms of a specific

42:43operational question , but what do you

42:45observe that people are missing , and

42:47what are you aiming for in these

42:48trainings , what are you trying to teach

42:50them ? Not to answer the question of

42:52what to invest in , but some kind of

42:53general concept . I don't know if

42:55>> contrary to appearances , I don't teach

42:57them the tool . They come for a Google

42:58Analytics 4 training , and I try not to

43:00teach them the tool , because I say

43:02today the tool is this , tomorrow it

43:03will be that . We had Google Universal

43:05Analytics , we had Urchin , right ? Now we

43:07have Google Analytics 4 .

43:10>> You could say even three different

43:12tools ,

43:13>> but the principle and the way of making

43:15decisions or drawing conclusions is

43:17exactly the same all the time and it

43:19doesn't change . That's why my success

43:22after training is when someone goes to

43:24a meeting with anyone , gets some number

43:26,

43:28>> and asks themselves the question : " What

43:30does this change ? " How does this impact

43:32my business ? The point is to

43:34>> ask that question regarding this number

43:36. Is this a meeting I want to be at , if

43:38we are discussing a number that changes

43:40nothing and contributes nothing ?

43:42>> Okay , so coming full circle from PMA ,

43:44you have a metric , you ask yourself the

43:46question , and then you wonder ,

43:48>> why do I need this metric ? Exactly

43:49right .

43:54>> This is a bit biased , because somewhere

43:56in my life I have this idea of

43:58minimalism , which I’ve been observing

44:00more and more in myself lately , so that

44:02our dashboards don't consist of 56

44:04metrics , but only those that are most

44:07important . And the ones that are most

44:13important are those where you can

44:14actually answer the question of what it

44:16changes , right ? If something goes up or

44:19down , how will my company act ?

44:21>> Okay . Could we leave our listeners with

44:25this in the context of marketing , to

44:27start with a small number of data

44:28points and treat them based on the

44:30question ? We have the number , we take

44:34action , and we look at potentially

44:36expanding that set . What is one thing

44:38we could leave our listeners with ,

44:41>> the most important one ? Well , picking

44:45just one would probably be hard , but

44:46thinking in that context , it was cool

44:48that after trainings , you’re happy if

44:50training participants go to a meeting

44:52and think about that number . So , what

44:56is one piece of advice to leave our

44:58listeners with here in the context of

45:00data in marketing , from your

45:01perspective and from what you’ve been

45:03doing for over a dozen years ,

45:06>> so they start asking themselves that

45:07question . Why ?

45:08>> Why , why ?

45:10>> Great .

45:10>> What is this number for , right ? What

45:12does it change ? Why am I collecting it ?

45:15If it goes up , what will I do ? If it

45:17goes down , what will happen ? That is

45:20the kind of thinking I mean , right ?

45:22That when someone throws a metric in

45:24your face and says , " The number of

45:26users on our site is this much , " you

45:28ask yourself : " So what does that change

45:30? " No , because when you go to the

45:32corner store , will you be able to buy

45:34your favorite drink or some chewing gum

45:36with it , right ? Will the cashier ask

45:38you , " And how many users do you have on

45:40your site ? " That kind of thinking is

45:41enough for me . If that thinking is

45:44there , then I believe I’ve achieved

45:46success , because I’ve sown a certain

45:48kind of seed that will surely take root

45:50in fertile soil , because the people who

45:52come to these trainings are often very

45:54inquisitive , and it will simply pay off

45:56, it will grow .

45:58>> Super . So we could summarize : " Let's be

46:01curious , let's not leave it without

46:03reflection , just inquire , dig deep , and

46:05good things will come from it . "

46:08>> Most certainly .

46:09>> Great . Thanks . for the talk .

46:11>> I also thank you for the invitation .