Chaos Structured

In this episode of FusionTalk, Anouck and Steve engage in a lively conversation about the intricacies of metadata management within SharePoint, drawing from their recent experiences in workshops and training sessions. They share insights into the importance of designing an effective metadata structure that not only enhances document searchability but also supports organizational workflows.

The hosts discuss their contrasting approaches to metadata, with Anouck focusing on the grassroots perspective of gathering input from end-users, while Steve advocates for a top-down corporate taxonomy. Their banter reveals the challenges of balancing user needs with overarching governance, as they explore the nuances of building a metadata strategy that works for all stakeholders.

Listeners will gain valuable tips on how to facilitate workshops, engage teams in meaningful discussions about document management, and the significance of accountability in establishing a successful metadata framework. They also touch on the role of AI in shaping the future of content management and the necessity of adapting to evolving technologies.

- Understanding the Importance of Metadata Structure
- Anouck's Grassroots Approach vs. Steve's Top-Down Strategy
- Engaging Teams in Metadata Discussions
- The Role of AI in Content Management
- Best Practices for Effective Metadata Management
- Conclusion: Balancing User Needs with Governance

Join Anouck and Steve for this insightful episode as they navigate the complexities of metadata management, offering practical advice for organizations looking to optimize their SharePoint environments. 

Creators and Guests

Host

Anouck Fierens

MVP | MCT | 🎙️M365 | Blogger | Book lover

Host

Steve Dalby

Podcaster "Office365Distilled" Driving Collaboration Business Goals, Speaking about Governance, Whiskey taster and imbiber all round father and good guy.

What is FusionTalk?

FusionTalk is a dynamic podcast where technology, collaboration, and innovation come to life! Hosted by Anouck Fierens, a Microsoft 365 & Power Platform expert, and Steve Dalby, a SharePoint & Teams specialist with a knack for humor, each episode delivers fresh insights, real-world stories, and engaging discussions with industry leaders.

🚀 Topics include:
✅ Modern workplace & collaboration strategies
✅ Practical case studies & expert insights

Join FusionTalk and discover how to work smarter, faster, and more efficiently in today’s tech-driven world! 🎧 #FusionTalk #TechPodcast #Collaboration

Welcome M to Fusion Talk with Anouk and Steve. So, you ready?

Always.

Always ready. Always, always ready for what?

For everything.

For everything?

No, Maybe almost everything.

So what are you not ready for?

If you ask me at the moment, I probably would not be ready for a Benji jump or something like this.

No. Bungee jumping. Good job. That wasn't on my list and menu. Oh, I like that. I'm almost ready. Hey, guys, and welcome to Fusion Talk. Steve Dolby here with the wonder Anuk, and she's obviously very effusive today. Effusive. Today we're going to get a lot of words out of her. Or maybe not. Are we going to get one or two words?

More than one or two. I didn't know what you were expecting of me this time.

You should be doing exactly what you want to do.

doing what I'm always doing, not knowing what I want.

It's true. Would you like to eat? Don't know.

So, yes, that's me as well.

Yeah. Yeah. Ah. Only unknown. So, do, any of your users ask you questions you can't answer?

Sometimes. and then I answer that. I will look it up for them, talk with people I know that maybe know more of them, and I will come back.

How much of your time is spent training nowadays?

at this moment in time? One day a week.

That's not too bad.

No, it's not too bad. It's actually quite a very nice change with doing some training and doing all of the rest of the work. Because I'm doing some interesting things.

Yeah, no, and that makes sense. I know you are. Yeah. I remember when I was a consultant.

Yeah.

I've had two weeks of contracts, not good contracts, sorting contracts, suing people, not suing people, sending nasty letters, refusing to sign things.

That's what you have when you want.

To be a manager, when you get to the point where you're leading it instead of delivering it. Yeah, yeah, no, that's okay. And to be fair, I do enjoy it.

I do need to say, I had one and a half day off this week and when on Wednesday morning when my day was off, I received a very nice letter that somebody asked me for an offer. They completely go with it and they want me to start planning to deliver. which is great fun because it's going to be all in the power app. power apps, power platform, all in there. And I do look forward to do it.

Nice. How sad. You really are a PowerApps nerd, aren't you?

No, no, no, no.

Yes, yes, yes, yes.

It's been a long time that I'm doing things like this.

You've done a bunch of SharePoint stuff.

I did a lot of SharePoint stuff and I think it's very nice. It's still SharePoint because all the data is being stored on SharePoint, but it's going to be some very fancy applications. I see a lot of nice upcoming sessions from experience coming there.

Nice. Good. That sounds like fun. I'm preparing for an AI hackathon and I've actually put a few, speaking sessions in to run live AI hackathons.

Oh, really?

In the community? Yes. None of them's come off yet, but, we'll see that. I think next year that will be quite exciting thing to do.

So about, talking about sessions, I did something for your company. I did some workflow with it. I've changed that in a session as well, and it's being accepted.

I saw that. I saw that. You done it or we doing it?

I'm doing it.

Oh, I'm going to harass at the back.

Yes.

Yes, I saw that. I, saw it come up. Come. I can't remember where, but yeah. Oh, was that Tunisia? Tunisia, yeah, the one we're not going to mention tonight. No, that's true. All right, Nice intro. yeah. So we've got some speaking gigs that's worth doing if you want to come see us in action, both individually and together. We are in, England. Bletchley.

Gdinia first. But yes, we are in Bletchley.

Yeah. I don't care what order they come in. I'm an agile guy. I do them in priority order.

No, but I think so.

Yes, now that's true. So Poland and Gdynia first at the Baltic Summit.

Yeah. Because that's a full day workshop.

Full day workshop. Basics of SharePoint.

Yes. So we are taking people back to Basic for SharePoint and Power Automate flows.

Or flows, which has been really difficult to do in some respects because we go, yeah, and then we can cover this and then we can cover that. And then I had to sit there and go, guys, stop.

We go way too far. Yeah, but we also touching a little bit of the AI stuff.

We have to just basics because we.

See it also a little bit as preparation to get your co pilot. Right.

Yep, yep. And I'm doing the same thing, at Bletchley, so I'm doing sensitivity labels, I think.

Oh, you do two sessions at Bletchley?

No. Then maybe it's the other One I really don't know.

It's a.

No. I'm doing sensitive labors in Tunisia. I know that. And I thought I was doing it.

Somewhere else, but I'm probably thinking you're doing it in Gtinia as well, because in Bletchley, we are talking about, the 10 tips for governance for the platform.

Correct. That means we've got about four weeks to actually get the details down on that one. Yes, we do like to challenge.

And what do we do in Zagreb? I don't know.

I'm not sure either.

It's. It's October. We still have time.

Yeah, yeah, we got time, but that's the same with everybody. Marin and I often have a simple conversation. Oh, you speaking asp. What are you doing? I think I'm doing such and such essential with Moraine, of course. He's. He swaps around so often. Oh, I'm doing such and such a thing with so and so.

Yeah, but M. Gidenia, I know because I did some work on it yesterday evening.

Yeah, so did I.

So I had some time. So I start preparing a few circuses.

Come and listen to us and find out why SharePoint is like a circus.

Yes.

Actually, we should put some stuff on LinkedIn about that because, they'll be good to get the advertising.

Yes, but if we do that on LinkedIn, we need to make video like clowns or something like that.

we can do that. We can use my little video tool that I've just.

I will do your makeup.

You do my makeup? You would enjoy that as well.

Yes.

Yeah. part of the fun. Stop enjoying yourself. All right. We got a great subject to talk about tonight.

We hope so. it's just something that came into mind because I was busy with it today, and when we attended earlier, I mentioned it to you.

You did? Yes. I thought it was a great subject.

So. Yes.

So just. So today you did a half and half day. So you gave an organization some training, and then you sat down with them to do some design work and some thinking work and, architecture work around SharePoint. And specifically, you got into metadata. So here's the test. You described what you was doing today, really very coolly. So what did you tell me you were doing?

so I had a workshop with people, and I would say, take one of your documents. You have.

That's not what you described it as. You can go into that in a minute or two. But you said you were doing something around design, designing metadata structure, management.

Metadata structure and how they can deal with it in the company and how they can work around it and how they can make sure they have the right cycles, terms and cycles about it.

Okay, so explain, because I didn't agree with what you was doing to start off with.

That's the first time you don't agree, so. Really?

No, everybody knows we're always on the same page. It's fine. M. It's fine. but, but yes, the way you started it was quite interesting because I it from a different approach. So it's worth looking at both of those. So explain to people why I just confirmed we're all good technically.

Yes. So, I had a workshop with them and we were talking. I said to them it was basically management. If you have this, you have take one document type that you have in your organization, and if you think about it, what will be the thing that people need to know about that document directly from the moment they see it in a list without opening it in a library? Because they were trying to find a way that they find their documents much easier and quicker and quicker.

And so you, were able to also say, but if you get this right, guys, search becomes easier. And then I can build special pages.

Playing with views and all of that can be much easier for your end users. We can learn how they create their own personal view. That's not an issue. But if they know what information they need to fill in from that document, it will be getting more and more easier for them.

I get that up here. What we have is, some notes from, another podcast that I did with the infamous Moraine. And we was coming up with a model. I've still got some work to do on this, but it's kind of a reverse Harvard square. So, you know, you have four squares and half a square. So these are the outside edge. And what I did was I divided it up to do intranet collaboration workflow or system API, for example. And then we also did one for Whiskey, but we'll ignore the Whiskey one. But basically you can take those document types and start a line at intranet. So it's going to be an intranet, and if it is going to be driven by workflow, you can put a line in between that and workflow, and that line then represents. That's something you need to build in terms of your information architecture. So. And it was, quite cool. I got to build that model. I think, I think there's some fun in that to be able to allow people to actually easily work out whether something's going to be Done. But effectively this is the first step to deciding the content architecture for this organization.

Yes.

Even though you did do it wrong.

You think I did it wrong?

I, know you did it wrong.

You think I did it wrong.

Okay. Pride cooks again. No, no, no. So when I approach this, I do it differently. I think it's worth discussing both of them because it works quite well. For me, managed metadata is about corporate taxonomy. So from, And what you did is not wrong by a long, long way. And when you get a scenario where a particular department in a company says, look, we want to go to SharePoint but nobody else in the company does, then obviously you're basically looking at a site level. whereas of course, what I would be looking at is the whole tenant level and saying, hey, within this tenant, I want to search at the top level and pick these documents up from within all of these different divisions. so if I was looking for, roadmaps, hr, roadmap, finance, roadmap marketing, roadmap, product operation, roadmap, then I want to be able to have the appropriate terms around a roadmap. The type of roadmap, what the roadmap is going to be delivering in a managed metadata and then, ensure they choose that on their sub site. So for me, managed metadata always has to come from the top downwards.

Yes, but it comes from the top downwards. But you only try to find it on m a different kind of way.

Well, yes, technically, architecturally, we know that, but what happens if this particular team say, hey, we want these terms in our managed metadata, but they don't fit in the cumber architecture because they're actually a too lower level, Then you gotta, then you gotta fix something, basically. Or you have to have multiple term stores, which is of course is also possible. But that also means multiple term styles that have to be managed.

Yes. And that's why the audience in this workshop that I did was very important.

I do think, and I've always said that when you set up SharePoint, you have to start from the top downwards in the same proportion that you start from the bottom upwards.

Yep.

And you meet somewhere in the middle, because you can't start from the top downwards without the knowledge from the bottom upwards. and when I've had a complete organization to do, that's exactly what I do.

Yes. And it's not the easiest organization to do it for because they need to deliver a lot of proof that they have done something to keep some of their names and all of that.

Ah, you're Gonna need to give us some idea of what the business does.

Yes. So legally to be in their kind of situation, they need to approve that everyone in the organization is taking 80 hours of training every single year.

So it's like, it's like a legal company or an audit company or a consultancy company or some certified organization. And to keep them certification you have to prove that people are being continually updated and self assessed and all that. So you know, pharmacists have to do the same, prove that they've done so many hours. Oh, continual professional development. Cpd.

Yes. Yes, thank you.

You're very welcome.

I couldn't come to the name.

No, no, I just came to me as well. Yes. So that's fine. So they have to do CPD and the idea that you're building a SharePoint.

Is that they can get the proof of their, of some of their requirements. They have not all of the CPD because training you can't see who is doing what on check. You can have the reports on it, but not that easily. There are other tools for it. But yes, some of the CPD we are building in SharePoint.

So are you doing this, the content, or are you doing the process associated with monitoring what people learn? both. Nice. That's actually a cool application.

We are starting with content now because.

That'S got to be first. You can't build something if there's no content there.

If you don't, if I don't know what the content is or try to understand, try to understand it a little. What they are doing, I can't go into developing the process because for them it's completely new. Yes, they have it from this year so it's new for them and it's a learning curve we do together.

And this is going to be alongside normal collaboration and intranet ultimately I guess on the SharePoint environment we have the same thing. So as people know, I work for a company that sends trains around Europe, quite a lot of trains every day. which is a complicated business. More complicated than putting planes in the air. and there's also a tremendous amount of safety stuff associated with it. So we have a kind of safety wiki environment. So specific content types at each of the top level divisional sites where all of their safety records, information and everything else gets put into.

Yes.

And then it gets linked back to a special process application. not that we build it but it's out of the box that builds the processes so that if we have to do this it goes through the steps of the safety stuff and each step has got a link to a document and then that measures and tracks who's read what and done what. so that we know that we can audit it and say, look, you know, we had so many people go through the training and the documents are always updated and we can prove that the latest documents are there.

Yes.

So that's pretty cool.

It is cool. It's very nice doing this.

It is, yeah.

And, I think they had, first thought they had to me was like, why are we going back to Post Its? We are digital company. Yes, you are a digital company, but this, your computer needs to close. You don't need to look at your computer, you don't need to look at the documents. You just need to go with what you know.

You've jumped again, haven't you? So you're now building your brainstorming your terms and things you needed to do. And you were using Post Its?

No, I tried to give them some insights in what was possible.

Okay, so.

And they become to the conclusion like, there is so much we can do and we don't know so little about what we actually do. We need to have more information from the end users.

Yes, good. And that's part of the process is understanding that it is true. What's also interesting is that I said that I'm running, an AI hackathon. one of the things that we're doing is we want to be able to create training courses from that safety documentation and also the legislative legal documentation. So we're trying to get various AI tools. So we need to work them out. I know which ones we're going to be using. But to actually generate the course outlines, which you can do outside of your tools, but then you also need to be able to be inside your presentation application using different tools to create the presentation. So ChatGPT will be used to generate the course outlines from all of that content and information. And, Copilot, will be used to generate the slides from those outlines that are created. That's what I hope to be done anyway. But, we're going to hack it. So it's going to be a hackathon.

Can I join?

No. I'm the one that needs to look like God here on this one, not you. It's a long time.

I will be a participant. I will just join the course.

You just join the course? well, I've also, I mean, I know it's. We're digressing just a little bit. but I also have set up the AI sessions in the company. So we're also trying to find a way to identify where AI can be used. So generative AI. AI can be used to measure the size of brakes on cabins, cakes and it can be done, but that's obviously the machine learning side of AI and Azure side of AI, like we're using it for translation tools, but from an end user perspective where potentially you can make most, get most value from it. We don't really know where to start and people don't know where to start either. So you know, they've tried to do things that don't work.

So there are so many things based on AI that you can start doing. So, we are seen as experts most of the time.

Yes, that's true.

But sometimes we are struggling where to start. So. Yeah, I do understand that.

Yeah, no, it's true. I mean, but we're also experts on how to set up the content architecture for AI.

Yeah.

But not always about how to use the AI. But it's a little bit like, I can use Excel, but can I use Excel the same level that, a, data scientist would use it or a finance guy would use it? No, of course not. They're going to use it way level than me. But that's really the name of the game. So how do we get them into AI? But that's a different subject altogether. So content architecture. So you start from the bottom upwards. So what did you actually do? You got them to think about. You said you talked about their data types and then you did what?

Then I started to think about, to try to think like an end user that needs to search documents.

Okay.

What they need to know. Because, the people in my workshop today, they don't always know what's going on and how people are working. So they need to check what people want. They are coming from a file server.

Yeah.

So they don't have metadata and all of that.

They have folders, which is metadata.

that's something I try to explain as well. But even then they have one of their folder structure was 40 folders deep where the data was. So then I think how a new user, how does he know.

Where to put those things? Yeah, I know one of the things that I used to do a lot of when I was training and trying to identify where they're at to say, so you have these folders. So you're looking for a document, you go, oh, I know where it is, it's in folder A double Click, double click, double click, double click, double click, double click, double click, double click, double click, double click, double click. Oh, close, close, close, close, close, close. Double click, double click, double click. So, yeah, it's they. Yeah, especially if they. You work out that it's 25 documents and you just stand there going, double click, double click, double click, double click, double click, double click 25 times. And they laugh and joke and they go, yeah. Then you do it close, close, close, close, close. And they start to realize that, hey, yeah, ah, you know. So, yeah, it is an interesting one. I did, I did, I once did an experiment. M. I wanted to do a fast migration into SharePoint. And I do believe very, very strongly that the file structure is the only metadata that you have available when you actually migrate to File share into SharePoint. So plus, that's where they know their documents are.

Yes.

For them, that is this folder name here. So I wanted to try and bring some in. So what I did was I had a script written. Because I don't build any of my own scripts, I had a script written that basically took the file structure and made it a single metadata column so that they could search for finance colon reports and it would be part of the way through that text string. So they were able to identify where they were at. And that worked really well. If you collapse the groups, you're able to expand, and collapse them. But it did bring some horrendous metadata back.

I can only imagine that. But yes, that's sometimes the only way people try to understand it and know where to go to.

So what metadata did you create today?

We didn't create metadata.

What a waste of time.

No.

Yes.

No.

So when are you going to create metadata?

When they have talked to the right people. Ah.

so this is an IT or a project management. Ah, no. You missed an opportunity.

Why?

Who is accountable for the content within a department or a team or a division? Would it be a manager? You're trying not to say manager. I know you.

I think it, it is the management.

It's the management team. And so one of the opportunities you had was to actually say, guys, you're accountable for this. So in theory, you need to come up with the terms that will divide, design and define, tag your documents and take that to the, to your teams and check whether it's right or wrong. And how might they improve it? Because that does two things. One, I mean, you have to work this your own way. So don't take any of this the wrong way, but you've You've left them with the ability to just shake their shoulders for this responsibility and hand it onto their team instead of giving them the ownership for it. and it's true. I mean, you have, a. You a handbag, all right? I found out today you have lots of handbags, right? So those handbags are yours and you will make sure and put them away and do everything else. If you have two handbags left downstairs and they get taken upstairs by your cleaner, all right, and she throws them on the bed and. Or whatever she does with them, she's not going to have the same care as you would about that handbag because you have the history associated with it. And that ownership for this is really important because it gives you that caring. So next time you may still decide to do it your own way, of course, but think about where the ownership for it is. Because I believe managers need to understand their own things.

Yes. Ah, but that's why I sent them home with an exercise, okay. That I need to have their metadata when I come back from holiday.

But they're going to get that from their teams. They're not going to do it themselves.

Well, if the team is there, because a lot of people are still on.

Holiday next week, it is an interesting way for. And I think lots of consultants do it in different ways. There's no doubt about it. I've seen, one other way that I've done it in the past that works. So from the bottom upwards, I always just give them a column called, topic and subject, or subject and topic comment, which way around to do it? and they're just straight text columns and everybody says, yeah, but we can have this and have misspellings and everything else. Yeah, that's already cool. But what if five months down the line you've uploaded your stuff, you've got things tabled. I can take all of those columns and start to build choice fields, metadata, manage metadata from them so that they still end up with the same capability. But now you get to. You get to that. That is the quickest way, actually.

Yes.

Have you ever done the, the card system for defining taxonomy?

The card sorting system? Yes, yes.

Oh, you done that?

not for taxonomy. For navigation.

no, it's great for taxonomy. It really is great for taxonomy.

Never thought about that, doing that for taxonomy. I did it for navigation, but that's.

You did it for navigation because you can visualize it by putting up on the wall or you. Or whatever you do. You can build it that way and say six people Think that there should be a top level navigation called X. And so you can do that. Do you resort the cards and do several rounds? Yeah, brilliant. Brilliant. Yeah. So, yeah, defining taxonomy that way is also really quite good.

I don't do it manually. I have a, website for it where I'm doing it.

I've used the same tools as well.

So. Yes. But it's a few rounds.

Cool. Yeah. But that's quite a nice way of getting it. But at some point you've got to, in my opinion, you have to go from the bottom to the top really quickly. You can't spend a lot of time with the people at the bottom because at some point the owner of the ownership of this, who owns the corporate taxonomy is really important.

Yes. And that's, often a little bit of a discussion.

A little bit of discussion. A discussion. Yes. At all.

Who is going to own it? is it going to be it or is it going to be some of the responsible in the business?

Yeah. And what have you done in the past?

I already have a combination of the both of them. Yeah. I don't like all of the responsible with IoT because it is very.

They should not own any of these.

No. They don't know what's. What's living.

No.

but sometimes it needs to be involved with it just to give them support about how they need to do everything.

I think it's like everything else, it owns the technology around that taxonomy and they. And I mean at some point your auditing team also needs to be involved to make sure the structure is covered and. Right. but, but ultimately somebody needs to own it and approve things. And it depends how you want to operate it. People can add to the taxonomy, but it has to be approved. What governance around it, you know, because the first thing most people do is go level one taxonomy finance and then report, m. calculations or audit. no, no, no, no. We're not going to put any departmental names in the taxonomy. It has to be descriptive terms of content, not ownership of content. And that's really difficult to do.

It is. One of the labels that came back a lot today was contract.

It's an object, it makes sense. But it's layer two or layer three, not layer one.

Yes, I was saying the same thing and they were like, yeah, but what do we need to put above?

True.

Yeah. Think what are your contracts?

Yes.

Who owns the contracts? Who needs to find the contracts?

Yep.

Can everybody in your organization see a contract?

Yeah, but that's also A different thing as well, isn't it? That's where labels come in rather than metadata.

But they need to try already to understand it for putting those labels.

So what was the conclusion about what goes above contract?

They came, they didn't came to a conclusion.

Ah. what would you have put there?

in their case, I'm not sure yet.

I would have put legal. So a legal contract, a legal habeas corpus, a, legal report. so I would have had legal. Probably because I like terms at the top that are easy to understand.

Yes. But to put legal in there, you need to know that the company is having an internal legal system. But they do also outsource people for the legal team.

That's okay, but the content is still a contract.

When they talk about legal, it's about the outsourcing system. So they said if we put in legal, people will think it's the outsource, sourcing systems.

It's not about who it is, it's about what the object is.

It doesn't matter something. I try to explain them as well, but they want to have a different kind of name or something that it's clear for everybody in the company.

Well, when I, whenever they argue with me, I tell them that they're paying me to tell them how to do it. No, I don't really. I'm joking. And we test it in some way. So we kind of send a survey out. Hey, look, this is a document that is being used to define, the contractual rights to the building that we're using as our hq. Which of the following terms best apply.

To the next step. I'm guessing that will be when they come back, everybody with their labels and we can combine them. Then we will do the card sorting things. Very good to keep in mind for the next step. And then we are testing it as well.

Nice, Nice. It's such a cool job to do this. And not many people want to do it properly. They just see it as a waste of time. They don't see the value. But I'm guessing from what you're talking about here, this is the legal aspect. So they're already fairly well structured with their documentation anyway if they're a legal organization. So same with scientific organizations. They also have a very good structure for the way they manage their data.

Yeah, they are.

So that kind of works.

So yes, that's a very fun thing to do. And I think, I said to them as well, what you are doing today because you are looking for AI in your company as well, this is a great start. If you have this in place, your AI will give you better results.

True.

And they were looking at me like, why? Because when your data is clear, you have a good labeling system and manage metadata, but also a document lifecycle management afterwards. When this is in place and you have all of the labels, the life cycle and everything, you will get better results because AI will understand your data better and you will only get one version back and not multiple versions.

Maybe Copilot will understand it. ChatGPT won't.

I'm not sure. Have you seen in ChatGPT that you can now upload from SharePoint?

Yes, you can, but it's only a document you're uploading. Whereas copilot goes via Graph API first, then it goes to API and then to AI, then it goes from AI back to Graph. So the metadata is in that Graph API portion of it. And so yes, it's part of the process. I don't know whether the ap. Yes, it would in some respects because in theory any document that's been uploaded to SharePoint will inherit the managed metadata as the properties of the document. So that's actually quite true. But I don't know whether Chat GPT would run that. It's a lot of so many unknowns.

I'm guessing they will not try it with ChatGPT. They are looking for Copilot because they want to stay in Microsoft system maybe. and then Copilot will give them that value.

Yeah. Then there's another problem with copilot. I'm finding some of the limitations. So the, there's a number of problems in the number of documents that it will look at at any one time. So you've got to got these set to set these up as agents so they can deal with those documents. And then the other part of the process is that ah, Copilot makes a summary of the document that it then uses as search. So it doesn't go into the intimate detail of the document, it just says, oh, I've read through that document, here's a summary. I'll make a note of that in case I need it later. And that's how it tends to work. So you have to be very careful about how you assess these things. But you are right, it's cool that they're able to see that metadata structure. Yeah, yeah, you're absolutely right. but it takes them on the first step. It's pretty cool.

Yes. And this is some kind of thing. We will have limitations in and do a little Bit in the workshop in Gdynia as well.

yes, but I don't think we'll go down to that level by a long, long way.

No, but it's also the, those first preparations that we are doing. I do like it.

Yeah. But I don't know whether we will. That's the debate we've been having a lot. How deep is deep or how shallow is shallow? Simple metadata. Yes. Text metadata. Manage metadata.

No, no. But even with simple metadata we help a lot of people in organizations.

We'll do the usual trick where you have a column, text column and you come up with a word and then you can search on that word and find your content back. just so they get some idea. And understanding your dog's name followed by your year of birth and the date that you made love to your first boyfriend, plus the number of nights you got drunk, combined together makes a unique word which you can then use to find any content anywhere on SharePoint. Be fun.

Yes.

Oh, it's going to be so much fun this workshop. Do you know how many people we've gotten here?

I haven't, didn't have luck but to go back to my use case today.

Oh, boring. Fed up with it already. It's metadata. It's boring. Yes, go on.

For the same company, I, ah, built something very funny, very nice. So, it's. They have a lot of things in their company going on. They use a lot of kind of software and they have news updates regularly. So people can fill in a list with news they are interested in and they want easy filtering that news. So I've built something in the PMP search. The first time they start searching, they go to the page that the filters report are being set by all of the items they included in the list. So they have the list and only the items that you filled in are visible in your filters.

Okay.

So that's something else you can do when you set it up correctly.

True. And the refinement tasks and all that kind of stuff, as you say, it gets them started. So.

Yeah.

Nice. All right, so what's your next step? What is the next stage of this? They're going to come back and they're going to, of they've, they've told their staff to find the keywords because that's what they're going to do. Their managers, they will send the instructions. Then what?

Then we will do the card sorting, I'm guessing. Yeah, I will set that one up for them. That people understand it better and even the managers Try to understand it better. With naming they need to use and all of that.

Yes. And synonyms and all that kind of stuff. Yeah. Trouble is, where do you stop with it? Eh? Well, I guess then it's the processes on how you keep it updated and how you modify it and what rules. I know that, when we were in Moraine and I were working Atlas, copco, they actually had a metadata approval team. So globally they had a small group of people that met every month that managed all of the managed metadata, columns so that they were able to sort of work out what terms and make sure there was no duplicates, etc, etc. It was a bit too formal really, I think.

Yeah.

But it is a way of at least understanding where ownership is and what it did.

It's an organization not only in Belgium, it's globally, so.

Exactly.

That's different than an organization only here in Belgium.

Yeah, yeah. Different rules, different playlists.

Yes.

So yeah. Cool. Well done. That was a nice little subject that you dealt with today.

It is.

And you had fun doing it, I can tell you. Enjoying it. That it are. We're done.

I think so.

Anything else to put into managed metadata? How many managed metadata? Oh, what's the top level managed metadata called? A.

Term set. Sorry, term group, term set, term group.

Term set, term group, term term. Okay. Yeah. So how many term sets should an organization on average look to have? Actually a different question. What's the maximum number they should actually have?

I would say two.

I think that's reasonable because you can of course put terms and term groups inside a term set. It also means that you can manage it easy. because even when you've got these terms, they have to fit into a term set or term group that is managed from the top downwards. Yeah. Two is about right. How many metadata columns did you say they should have on their libraries?

Not more than seven.

And you chose that number for some scientific reason?

because most of the time I've learned that customers end up between 4 and 6 or 4 and 7.

Okay. So basically they can have whatever number they like, but on average, you see. Do you have multiple libraries per site?

Yes.

Even if they're a team site?

Yes.

How do you access the document library from the team?

Oh, they don't use Microsoft Teams only for meetings, so they don't code their documents there. But generally, generally when it's in teams only.

No one library.

Yeah.

Because basically it's a short term collaboration thing, not a long term storage thing. are you allowed to have different multiple document Content types in a library. Not, not architecturally, but from an architectural design. So in finance, would you have a library that holds five different types of financial document?

It depends.

No, you can't do that. That's a crappy answer. It depends on what.

Based if it is a, project.

No, carry on.

You don't even let people speak. So if you are dealing with a project, you are dealing in the same project with an offer, with a contract, everything you want to keep in place before you decide yes or no. Maybe for that project. So then you have multiple content types. Otherwise no.

So you would. If you're a finance department, you would. Or a sales team, you would separate off customer contracts, customer quotes, customer, whatever as part, even though they're all associated with one customer. Now you want to change your answer, don't you?

No.

You want to say it depends again.

No, I would keep them in the same library, have different content types.

Yes, I would suggest that.

So I said it depends. When you working with a project, you.

Said it depends because you wanted to have a get out of jail free.

No, what I was saying about project is the same thing.

No, but the problem I have with a project is a project never ends up in a SharePoint site. Project should always be in team sites because they're not permanent.

Yes, but at the end some of the documents need to stay. Need to go to SharePoint.

Correct, but then you would keep them together in the same library. But the bottom line is, and I wanted to get back to the metad because it is true, you will have different types of documents in the same library because you want to pull them together. And let's be fair, you can stick a million documents in a library. You don't really want to have to work out which library do I upload this document to its customer information goes in the customer information library. and then you're able to choose which content type it is. But that's when you end up with more than seven columns of metadata. Because each of the content types may have different metadata and it can be.

Possible, but it's just giving them some kind of guidance that they don't need to go overboard.

No, no, no. Otherwise you end up like, training.

Session I did in the morning. They end up with 33 columns that they would like to fill in.

Of course they did. They always do.

That's not. People don't going to do that.

No. That's an interesting question. Six or seven columns. I don't think I go for that generally. Unless some of them are automated because generally most people won't fill out more than two or three, maybe four columns. But if your organization is a professional organization, legal, financial banking, then they will, because they tend to be used to dealing with all the data. but your general average Joe and a document library. You only need two things to be able to find the document easily. Usually they're not going to type in six keywords.

No.

And all of that, of course, is part of your content architecture governance. So, hey, we will have somewhere between three and seven columns on a document library for managed metadata. Two of those columns will be the corporate metadata called information type. And also maybe, sensitivity, but they do those by labels. But maybe some other kind of thing. Department. Department. I'm assuming you automatically fill out from the site name.

That's a strange one there.

Well, you got to have a department as a metadata item on a column, but you don't want people to fill it out. They don't need to, you don't.

Need to have a department.

Yeah, you do.

No.

Yes.

No.

Yes.

No.

Yes. Sales contracts, legal contracts, building contracts, facilities contracts, both are the same. They will all be on different sides.

They will.

And one of the easiest ways of saying what type of contractor who owns it is by having the department name. Now tell me why I'm wrong. I'm loving this. I can see you sitting there all cocky behind that pink microphone of yours.

In this case, they don't have any departments.

No, but that's not what I asked you and that's not what you was implying. You were saying you don't need it.

no, they do it differently. So, they would like to have. Would like to have. We didn't say yes or no that we were going to do that. Is that they would combine some kind of. Some of those documents together and not doing department separately. I'm working into destruction of how they.

Are working in some respects.

So they would prefer for those kind of documents to have some kind of centralized storage system. Yeah, the old type. Yeah.

Rather than ownership.

Nice.

I've only ever tried to do it once. No, that's not true. I've tried to do that two that I can remember, but I've got a feeling there was a third and only one of them really succeeded.

Yeah, and in this case, I haven't say yes or no to that, at this moment in time. But they are growing very fast. They are still in their growing area and they are buying up other firms and all of that. And they do it with A speed of one firm a month. So it's for them. I can, understand it that they want to do it, but I see the danger in it as well that it's not going to work. So I'm really trying to understand that process better, to understand what they want.

It will work, but only if you have a solid corporate taxonomy, managed metadata, with a team of people willing to make sure it's looked after. But then your biggest problem, but I don't think it will be in a legal banking environment, will be people using it and not just storing it locally on their OneDrive or, well, on their teams.

That won't be the issue.

Yeah, no, but. Okay, go on then. Why not?

no, they are very clear about how they are working now.

That's legal finance. That is a professional organization. They tend to be better. That is true, yes. and that you don't have local ownership of contacts, so that would work. But yeah, the biggest fallover with doing that in my cases was that people didn't care enough.

Yeah.

So they just wanted it where they knew it would be, not where it should be placed.

So I also said to them, let's see what's coming up with, every metadata taxonomy you guys would like to have and if we can combine it. Yes or no?

No. This is fun. That sounds like you had fun and you've got it together, which is great.

Yes.

Nice, nice, nice. Best of luck with it.

I will.

Nice.

And if I need advice, I know where I need to go to.

Of course. Well, we brainstorm all these ideas anyway, so then we end up and every company is different, so. Oh, that's where it is. I've been looking for the remote control for the projector for ages, but I've been looking at this half of the room. But I do remember that that's its new home, that it's going to be when I move the equipment. So obviously I put it the other side of the room.

Did you say that a remote controller has a home in your house?

Well, that, that little remote control over there on that shelf is where its new home is going to be. But my mind yet is it hasn't. I've been looking for it for about three weeks.

You are saying that something has a home?

My ass. My ass has a home in this house. And it's, got one or two homes, actually. One is on the chair of my office desk. The second one is on the only toilet in the room in the house. But that might change in the next week or two. Yes, I know it's a bit chaotic here at the moment. I get that.

You are chaotic and then you give something a home.

Well, Anna's got a home.

I'm impressed. You're improving.

Yeah, but I didn't know where it was for the last three weeks. like a badly, A badly content structured organization often counters a lot of this stuff just, just to kind of round this up. So. Well, let's, let's. I'll come back to that in a minute because we've already been nowhere at 50 minutes already, which is quite cool. Whenever you go into a good subject, the time just runs away. So I, I think it's been a cool conversation. So we've talked about the importance of metadata structure within the organization, talked about some ways of collecting that data, and hopefully people appreciated that. They can follow up with us. They know how to get hold of us. If the card system is something you haven't no idea what it is, we can find some links. There's plenty of things on the Internet to do that. the bottom up, top down approach to manage metadata, if you want us to expand on that a little bit, you can do one is really driven by the people, the other one is driven by the organization and you need a structure to be able to do that. But, and so you then can build that managed metadata and the larger it gets, the more difficult is to handle. So if you can do it in small chunks and in small areas, that would work quite well. But Moraine often says nowadays it's an important. You don't need to do it it. Because search is getting so good, it doesn't always need the metadata to be able to allow you to find your content or sort your content. Plus, AI and auto labeling and auto tagging of content is also getting good. But in all honesty, we're not there yet.

No, we aren't. And it really depends on the type of company you are that you needed to have it or not.

That is true actually.

And because of their core businesses, they are so well organized that for them it's good to have.

It's also a little bit like my old man used to say to me, God bless him. Listen, if you want to do it, son, do it yourself. You want to do it well, do it yourself. And I think there's a lot of that. That said, look, if we're going to make sure this content management is right, then we're going to give you the tool so that you can label this with the terms you need to use that are, shared around the organization and so that we can group and find and can this content as and when we need to. So there is a, there is a value in that rather than trusting it. And with AI, of course the amount of frustrations you get because it only does 80% of anything that you want to do.

True. And the 8020 rule is still important also in all of this. but even then I think for the company they are the metadata needs to be right.

Yeah, I agree. But that include. I mean it's going to the 100% which means you have to do the difficult bit which is the 8, the last 20% which is 80% of the effort.

But I do like the difficult bits.

True, but you've also got to think about the processes afterwards. That's going to be key to be able to maintain it. Well, great subject. Yes, thank you very very much. You now have the responsibility of the next subject. You know, you've done it so well this time.

Oh, next time that we are recording, I do also a workshop that day. So maybe I find something interesting there.

Okay. No, that would be fun. We'll see where we go. All right, Steve Dolby here. Really hoping you've enjoyed this. it's been a fun podcast. Touching on this. I haven't done this for a while. I do have to do it but it'll have to wait a little bit. so it's nice to have this conversation so that in debt. So yes, that's Steve Dobby saying goodbye.

And ah, goodbye from me as well.

Do a real goodbye. Come on. Make them feel wanted. Make them feel like you care. The fact they're listening to you.

I do. And I really hope they still enjoy holidays and that they start listening to all of the episodes we did over summer.

Catch up on the backlogs. Yes, no, that'd be good. All right guys, thanks and bye. Sam.