All of the Things! Custom MS Search Vertical of Serverless Azure DB and the new SQL Graph Connector!

Mike Homol

Mike Homol

Principal Consultant @ ThreeWill

Microsoft Graph Connectors are now in preview mode and appear to be showing up for most folks who have turned on targeted releases in their Dev tenants. So, naturally, I had to see what we could break... er, do.

Just how much can we throw in here?

Just to spice things up for this blog post, let's see if we can pull in as many different and new-ish things into this bad boy.

Throw it all in!

SQL or bust

One of the first things I wanted to try out was the SQL Server connector. This is big win for SharePoint and 365 to be able to include 3rd-party structured data without any custom development. So it felt like a great candidate to try out first. So here was problem 1: I had no "on-prem" SQL database anymore. And SQL is expensive. Or is it? Pete Skelly turned my eyes towards serverless Azure SQL - a consumption-based model for SQL. Seems pretty new and cool to me. Let's throw it in!

Gotta have a VM

We want to mimic an "on-prem" experience, so that means trying to use the Power BI gateway to connect to our SQL database. So let's make sure we have a VM to install the gateway onto that will sign in with a dev tenant login and can then connect to the SQL server. Things are looking good. Anything else?

Bump up those numbers

Some automation, perhaps?

What if we see how much of this we can build with ARM templates? Yeah, that sounds good. The ARM Tools extension for VSCode is pretty solid. Channel 9's DevOpsLab just started a video series on Demystifying ARM Templates, which shows the power of the MS team's JSON schema they've built. So yeah, why not throw that in. Plus, I've been wanting an opportunity to use the ARM Template Viewer extension in VSCode too, so now I have a reason and a way to visually represent almost everything we're creating for this little Graph Connector.

Wait, what are we doing again?

All part of the plan

So here's the plan:

  • Serverless Azure SQL database as a backend
  • The AdventureWorks database loaded up with Product data
  • Windows VM with a Power BI gateway to connect to the Azure SQL db
  • A new graph connector, which uses the gateway to crawl Products, to display in a new vertical in MS Search results
  • A custom adaptive card display (we gotta see if we can pipe in product pictures, right?)

Here's a high-level overview of the data flow, from SQL-to-VM-to-365: High-Level Data Flow

The VM's gateway will be controlled by the 365 Search service and graph connector API. Every 15 minutes, 365 will attempt an incremental crawl, by reaching out to the gateway on the VM, which will receive a query to execute agains the SQL DB on Azure. So let's get this party started.

ARMing ourselves (get it?)

Based on the above, here's what our ARM template needs to look like:

ARM Design

This diagram made possible by ARM Template Viewer.

Don't feel like making all of this? No problem! I've already set it all up and exported it to an ARM Template for you! I even went ahead and took the time to set it up a little better for automation. I've provided 2 files for you to use to get all of this provisioned automagically in your own tenant/subscription:

  • full_simple.json - this is the ARM template that represents the above design
  • runit.azcli - this is an Azure CLI script that is built for Bash.

Simply put the 2 files next to each other from wherever you plan on running them. If you want to run it from the cloud shell, like I did, you'll just have to upload them first, like so.

Optional: upload the files to Cloud Shell

The ARM Template JSON has a number of preset variables you should know about. If you want it to run successfully, you'll need to set them to be something unique for you. These variables are:

  • rg - A reprint of the resource group you plan on putting everything in. That will also get set in the runit.azcli. This name just gets used in some of the more generic naming of things like the NICs and NSGs that I'm hoping you don't have to even think about.
  • vmname - The name of the VM that will get made for you
  • location - This also gets set in runit.azcli so you don't necessarily have to set it here.
  • servers_serverlessserverpoc_name - The name of the database server that will be created
  • adventureworks_dbname - The name of the database (defaults to HomolWorks)
  • uniquedomainprefix - Should probably be the same as the name you pick for the VM - using this will make RDP'ing easier
  • my_ip - if you set this, then your IP will be automatically added to the firewall rules for the Database server
  • sqladmin_account - the SQL admin user name. Defaults to POCAdmin
  • vmadmin_name - the VM admin user name. Defaults to captainawesome

Run it!

The script only has 3 variables it tries to preset before deploying the ARM template. Naturally you could edit this all you want and feed in more parameters rather than setting them up in the .json file. Only 2 variables really need to be set: rg and location. This ARM template is scoped to a resource group, so the script creates that Resource Group first, then deploys the ARM template to the group. Note that you can optionally set the IP you're developing from in the my_ip variable.

Once you've set rg and location, run the runit.azcli script from a terminal using Bash or Zsh. I ran mine from Azure Cloud Shell.

Running the Script

Not so bad so far, right? If you're feeling really adventurous, be sure to look over the ARM Template in full to see everything we've got going on. Lots of small stuff has been squirreled away so you don't have to care.

NOTE that I did actually get an error running the template, which was during the setup of one of the advisors for the SQL database, where I was told the resources were busy. But everything had actually executed fine, at least for the purposes of this Proof-of-Concept. So if you get that error, just make sure that the AdventureWorks database was provisioned in the new resource group and you should be good to go.

So, we're ARMed. What do we have to show for it?

Tis but a scratch

Review and Connect

Let's review everything we have now that the script and ARM template have finished. First let's just take a look at the finished Resource Group. Quite a few resources have been generated and most of them will cost next to nothing. And here's the best part: you can just delete the resource group whenever and be done with this.

Our new resource group

Click on the VM and notice the top right portion of the Overview tab. We have a static IP and a DNS. This will make for easy RDP, plus it has been beneficial to setting up our DB firewall.


Let's see where that IP is used by SQL. Go back to the resource group and click on the SQL server. Then select the Firewalls and Virtual Networks option. Note in the IP Address rules that we already have an item added - the same IP as our VM. If you hadn't setup your IP, now could be a good time to add that a Save.

SQL server firewall

Gateway Time

Remember that DNS? Well now's our chance to use it. Connect with your favorite RDP client, using the domain that was created and the VM admin account/password that you setup in the ARM template parameters.


Time to install the gateway. To start go to and select Download Standard Mode.

Gateway Install

Once it's installed, sign in using your dev tenant admin account.

Gateway Signin

Choose Register a new gateway on this computer.

New Gateway

Finally, setup a name for the gateway and a recovery key (which is a passphrase of your choosing).

Gateway name and key

Graph connector setup

Okay, we're almost there. Time to setup our sql graph connector. First, let's confirm that you can even do this. As we mentioned earlier, this is still in preview mode at the time of this writing, so if you want to have a shot at seeing the connectors, you need to set you organization settings to getting targeted releases. See the animation below for accessing that setting from the 365 admin center:

Targeted Release Setting

Here's where you go to access connectors. In the 365 admin center, select Settings->Microsoft Search and confirm that you have the Connectors tab.

Connectors tab

To connect to our SQL Server, click Add, then Microsoft SQL Server. Then supply a Name, a Connection ID (just a unique identifier of your choosing) and a description. Finally accept the terms to proceed.

After we click Next, it's time for us to setup our database connection. Finally! To get your connection string, head back over to Azure and select your database. Then select the Connection Strings option along the left. The first tab holds the information you'll need over in your 365 dev tenant.

DB Connection String

Note that in the Database Settings step for the graph connector, the On-premises gateway is selectable. You should see the name you provided for your gateway in there. Select it. Fill out the parts need for the connection and click Test Connection.

Test the connection

What's happening under the hood is that 365 is reaching out to the gateway agent running on your VM and then making the connection to the database, which has allowed your VM to connect through it's Firewall rules. Pretty neat, huh?

Aw yeah

What exactly is in this AdventureWorks, anyway?

Okay, so the last few steps of the Graph Connector are about setting up the data to be crawled. So what exactly do we have here? We're going to focus on Products. Here's a quick snapshot of the tables though from the Azure Data Studio:

AdventureWorks Product data

We intend to target Product and we will also include the Product Model through a join.

The next step in the Graph Connector setup is to provide our full crawl query. Note that we are expected to provide a date field that we will base the query off of, so that only items added after the previous crawl will be pulled to make things more efficient. This is the watermark field (@watermark). We have chosen CreateDate as that field.

Hold up. There is no CreateDate field. Well done, young padawan. Oddly enough, the MS folks didn't think to have one. So we will need to do it for them. Go back to Azure and select your AdventureWorks database. Click the Query Editor (preview) on the left hand side and log in with the database admin account you provisioned. Run this query:

ADD CreateDate datetime not null

This gets us the CreateDate field we need along with a default date in it. The first crawl is a NULL watermark anyway so everything is coming in.

With that little bonus step out of the way, here's our query for the full crawl:

SELECT p.[ProductID]
, p.[Name]
, p.[ProductNumber]
, p.[Color]
, p.[Size]
, p.[ProductCategoryID]
, p.[ProductModelID]
, p.[SellStartDate]
, p.[SellEndDate]
, p.[DiscontinuedDate]
, p.[ThumbnailPhotoFileName]
, p.[rowguid] as ProductGuid
, CAST(p.[CreateDate] as smalldatetime) as [ProductCreated]
, CAST(p.[ModifiedDate] as smalldatetime) as [ProductModified]
, pm.Name as [Model]
, pm.rowguid as ModelGuid
FROM [SalesLT].[Product] p
INNER JOIN [SalesLT].[ProductModel] pm on p.ProductModelID = pm.ProductModelID
WHERE p.[CreateDate] > @watermark

Choose DateTime as the watermark data type and press the Validate Query button. Select the watermark field, which we aliased as ProductCreated, then select the unique identifier field, which is ProductId. Notice we have a print out of the first 10 rows of data as well. Interesting side note here: currently money and float fields appear to not be supported by the graph connector. That's why ListPrice was left out of the query.

Full Crawl Step

Next we set the incremental crawl. This will append anything that has been modified since the last incremental crawl run. This step is optional but I recommend it. The crawl looks very similar to the full crawl, but instead our @watermark is based on ModifiedDate instead. Here's the query:

SELECT p.[ProductID]
, p.[Name]
, p.[ProductNumber]
, p.[Color]
, p.[Size]
, p.[ProductCategoryID]
, p.[ProductModelID]
, p.[SellStartDate]
, p.[SellEndDate]
, p.[DiscontinuedDate]
, p.[ThumbnailPhotoFileName]
, p.[rowguid] as ProductGuid
, CAST(p.[CreateDate] as smalldatetime) as [ProductCreated]
, CAST(p.[ModifiedDate] as smalldatetime) as [ProductModified]
, pm.Name as [Model]
, pm.rowguid as ModelGuid
FROM [SalesLT].[Product] p
INNER JOIN [SalesLT].[ProductModel] pm on p.ProductModelID = pm.ProductModelID
WHERE p.[ModifiedDate] > @watermark

Similar to the full crawl, choose DateTime as the watermark data type and press the Validate Query button. Select the watermark field, which we aliased as ProductModified, then select the unique identifier field, which is ProductId. I skipped the soft delete section for now. I also skipped leveraging any type of row-level security, which is actually supported and documented in Microsoft writeup on the SQL Graph Connectors. Essentially, you would need to include ACL columnsin the full and incremental crawls, named AllowedUsers, AllowedGroups, DeniedUsers, and DeniedGroups. Each column is expected to be comma or semicolon delimited and can include UPNs, AAD IDs or Security IDs. I just wanted to see if we could get this data coming back and looking good!

Incremental Crawl Step

The last big step is our Manage Schema step. We define what can be queried, searched and retrieved. If an item is found in search, you can only show it in the adaptive card layout if it's been marked as Retrievable, thus added to the schema. So select what you want. I went with anything text-based to be searchable and pulled almost all fields into the schema by marking them as Retrievable.

Manage Schema Step

The last few steps are a bit of a breeze, especially since I chose not to do ACL columns and row level security.

Final Steps

Results design and vertical

Our final step is to create our vertical and design the result set. After creating the connector, you'll see some callouts on your new connector in the Required Actions column. You can either click there to set things up or you can access Results Types and Verticals in the Customizations tab. As of the time of this writing, the best order of events is to setup the Result Type first then the Vertical. If you do it the other way, you will have 1 extra step to Enable the Vertical after you finish setting up your result type.

Required Actions

Time to Adapt

Let's start with the Result Type. Here are the basic steps, which the below GIF flies through:

  • Set a name for the result type
  • Tie it to a Content Source, which will be the new Graph Connector you made
  • Apply any display rules (I skipped this)
  • Design the layout
  • Review and Finish

The big thing here is our layout design. We are provided a link on this step to the Search Layout Designer. Select a design to start with and Click the Get Started button. This takes you to the designer where you can layer in the fields you want to replace the template with. We want to make some substantial changes, so let's click the Edit Layout button. This layout designer leverages the Adaptive Card schema to do it's magic. Also, any field we set as retrievable and is in the schema is now a field we can display in the layout. Here's what I built:

"type": "AdaptiveCard",
"version": "1.0",
"body": [
"type": "ColumnSet",
"columns": [
"type": "Column",
"width": "auto",
"items": [
"type": "Image",
"url": "{ThumbnailPhotoFileName}?raw=true",
"size": "Medium",
"horizontalAlignment": "Center",
"altText": "Not available"
"height": "stretch"
"type": "Column",
"width": 8,
"items": [
"type": "TextBlock",
"text": "[{Name}]({ProductID})",
"color": "Accent",
"size": "Medium",
"weight": "Bolder"
"type": "TextBlock",
"text": "{Model} ({ProductNumber})",
"weight": "Bolder",
"color": "Warning"
"type": "TextBlock",
"text": "First sold on {{DATE({SellStartDate}, LONG)}}",
"spacing": "Small"
"horizontalAlignment": "Center",
"spacing": "Medium"
"type": "Column",
"width": 3,
"items": [
"type": "FactSet",
"facts": [
"title": "Color",
"value": "{Color}"
"title": "Size",
"value": "{Size}"
"$schema": ""

Feel free to paste this into the Layout Payload Editor to see what it will look like. If you want to see a preview of it you'll need sample data to paste into the Sample Data Editor tab. Try this:

"Name": "Classic Vest, L",
"Model": "Classic Vest",
"ProductNumber": "VE-C304-L",
"ProductModified": "2008-03-11T10:01:36Z",
"SellStartDate": "2007-07-01T00:00:00Z",
"ProductId": "866",
"Color": "Red",
"Size": "M",
"ThumbnailPhotoFileName": "awc_jersey_male_small.gif"

What's the deal with the thumbnail?

Here's the last "special" thing we wanted to add to the heap of "All the Things". In the AdventureWorks Products table, there are Thumbnail binaries and Thumbnail names. Well, of course, I wanted to see these thumbnails come through in the results. Varbinary fields aren't supported by the crawler, so I had 1 of 2 options: either make an endpoint that would pull the item from the database for the Product ID on any call and return the byte array as the response or pull all of the binaries out of the database once and save them to files elsewhere. I chose the latter. Here's the source for it if you want to do something similar yourself. So now I had the files I needed in github, named by the ThumbnailPhotoFileName field value. So that's how I'm able to include that in my layout.

Here's a quick rundown of setting up the Result Type:

Result Type Setup

Last, but not least, we make our Vertical. It's even simpler.

  • Provide a name
  • Select the Content Source
  • Provide a KQL query (optional I skipped it)
  • Review and Finish
  • Click the button to Enable Vertical

Here's a quick rundown of setting up the Vertical:

Vertical Setup

Out of the oven

Finished Product

So what exactly have we cooked up here? Let's head over to a search box somewhere and type something in that we know is a word in the Products or Models. How about mountain?

Search up mountain

So there we have it! It was a lot to toss into the pot, but I think we brought it all together nicely. Hope you've learned a little something and that it gets you thinking about what you want to do next with the new Graph Connectors and what other structured data you may want to start piping into Microsoft Search for your customers and employees. Don't forget to delete this resource group when you're done messing around. Enjoy!

Mike-rosoft Learn Highlights for July

Mike Homol

Mike Homol

Principal Consultant @ ThreeWill

Like a good nerd, I'm a sucker for gamification and achievements in all of their forms. So I've taken to MS Learn badges and trophies like a duck to water. On the plus side, it's already single-handedly gotten me an Azure Developer Associate certification. So I'm giving back by doing a monthly series where I highlight some Learn badges that I felt were exceptional and deserve to be viewed.

This Month's Highlights

Publish an Angular, React, Svelte, or Vue JavaScript app and API with Azure Static Web Apps

Microsoft Ignite was all digital this year but was still pretty fun. It was like a giant Channel 9 event. In that event, John Papa did a great session unveiling Azure Static Web Apps. True to form, almost immediately after that unveiling, this little number showed up on MS Learn. It's great first look at a preview product, where Azure is throwing their hat into the ring with other JAMstack hosting services, like Render, GitHub pages, or GitLabs. I definitely recommend this walkthrough - it's quick, easy and free to try and you get a good glimpse of everything that's supported.

Enable automatic updates in a web application using Azure Functions and SignalR Service

SignalR is such a fun piece of technology. I had a blast building this one and watching things work. This is a methodology any developer should keep in their back pocket.

Expose multiple Azure Function apps as a consistent API by using Azure API Management

I have to imaging the Azure API Management can get pretty pricey so I haven't actually done any of this in a production sense, but the lessons surrounding Azure API Management are pretty neat and put a great bow on your API structure. But this particular lesson, which shows you how to pull disparate Azure functions into a single cohesive API, both structurally for you or your organization but also for consumers of said API, is really strong and shows you the true power of Azure API Management

Build a web app with Blazor WebAssembly and Visual Studio Code

I have high hopes for Blazor in the long run. The route of building WebAssembly is some pretty interesting stuff and this is a great introduction to the technology and nicely self-contained. Hopefully it will leave you wanting more.

Better Sharing is Better Caring

Mike Homol

Mike Homol

Principal Consultant @ ThreeWill

As the PnP Weekly adage goes, "Sharing is Caring". So I posit the following: does this imply that "Better Sharing is Better Caring?" You be the judge. 😜

Current state of knowledge sharing - it's great!

The state of sharing in the development world, particularly thanks to the gains of open source over the years, has never been stronger. We can push code to GitHub almost instantaneously. We have lots of ways of describing our code to others using markdown files, typically at the root of our solution. Maybe we've even gone the distance and built out a set of GitHub pages. In it we have real writeups on specific features with example snippets of code or showing how to use our product.

But can it be better?

Have you ever felt like there's almost too much to learn and to unpack? Developers are expected to move faster than ever and know more than ever. How do you move quickly when you're jumping into a new technology or project without feeling overwhelmed? Is there more that we can add to our tool belt to assist with knowledge sharing, documentation or investigation? Allow me to throw at you 2 new technologies to assist in this endeavor: Jupyter Labs and CodeTour.

Jupyter Labs

Jupyter Labs comes from the Python world of data science. But it has moved far beyond just that. Essentially, it leverages Python to allow for different languages to have a runnable kernel against a jupyter notebook. What this gives you is something quite powerful and cool. It's a web application that allows you to create and share documents that contain live code, equations, visualizations and narrative text.

I once described it to my team as this: imagine having a wiki with runnable code snippets directly in the wiki. One of my teammates, upon seeing a demo for the first time, described it in a somewhat opposite way: it's like code with way better commenting. Either way you look at it, it's certainly more powerful that just code or just documentation.

Getting Jupyter Labs set up

Jupyter's site has instructions to install here. But just doing that isn't going to be as powerful, especially if you're a Microsoft developer. Jupyter's base install is a Python notebook that can run in the Kernel, We want to deal in things like C# and PowerShell. So lets add that to kernel using .Net Interactive. Personally, I think Scott Hanselman's instructions here may be your best bet, especially if you're on Windows. This means you'll need Anaconda installed first (remember all this is based on Python).

PowerShell Core

As you've probably already caught on, this is Python and cross-platform. This means we Microsoft folks need to stick with all things Core. .Net interactive's notebooks give us C#, F# and PowerShell Core though, so we have some fun things we can do. This does mean that the PnP PoSH folks are on the outside looking in, until it supports PowerShell Core. But hopefully that's coming very soon. So check out what we can do using PowerShell Core, in the examples below, and hopefully that will get your mind spinning about other things you could do, including when PnP gets added to PS Core.

Azure CLI example

So keeping in mind that we are sticking to PowerShell Core, I whipped up a few examples of utilizing other CLI's with PowerShell to do some computing. Let's start with Azure CLI. Below is something simple. I just copied the MS documentation for getting started with the CLI into a notebook.

It's a totally different way to imagine documentation. Allow readers to instantly see the results, in the context of their own data!

Office 365 CLI example

Let's look at another aspect of using these notebooks: helping your team get something done. In this example, I've crafted some instructions to give to someone to create a site with the same theme that I made inside my tenant. Check it out.

Code Tour

Let's end with a bang. I have absolutely fallen in love with this next one: CodeTour. It's pretty new extension for VS Code and allows for providing a tour of your solution. As someone who has a passion for learning and teaching, I can't think of a better way to handle the onboarding experience for coders than a guided tour. And there are many other applications too. Recently, the PnP team used Code Tour to assist with the SPFx Project Upgrade. I'm sure once you play around with it, you will also think of new applications for it.

Install the CodeTour VS Code extension

Get the extension here. I'm assuming that you already have VS Code. 😜 Also, a shout out to Jonathan Carter, the brains behind this. He's very receptive to feedback too so hit him up.

CodeTour example

I'll stay on point here and keep within the realm of PowerShell. Here's something I did recently for a PnP Provisioning Script for a client.

As you can see, it's a wonderful and powerful way to onboard or to simply amp up your documentation for a piece of code or for a script like this one.


Hopefully I've provided you some new thought-starters for better ways to share information. These technologies and others should become part of our best practices tool bag. They allow for easier explanation of code, faster results in collaboration, simpler paths to onboarding and so much more. Please take the time to consider how you might use these solutions on your next project.

The trick to migrating list parts on a page with custom views with PnP

Mike Homol

Mike Homol

Principal Consultant @ ThreeWill

Adventures in PnP PowerShell provisioning templates

Has this ever happened to you? I had built a custom list with a custom view. To be more precise, I had basically lifted Chris Kent's sample here for a custom FAQ and dropped this on the page and the client was thrilled with it just as it is. Thanks Chris! Here's what the FAQ page looked like on the template site:

The way an FAQ should look
Not a bad looking FAQ list, right?

But this is just the beginning! This was my template. I need lots of sites to have this same FAQ page as a starting point, and it needs to look this good too.

So, onto provisioning with Powershell and PnP! At first I was running this:

Get-PnPProvisioningTemplate -Out $templateFull -Verbose -PersistBrandingFiles `
-PersistPublishingFiles -IncludeAllClientSidePages -Force
Apply-PnPProvisioningTemplate -Path $templateFull

Looks familiar right? Well pretty much everything was working great except for this:

No bueno FAQ
Not so good. Also don't focus on the color difference lol

What the heck was going on here? The view and the list were migrating just fine, but that view was not getting applied! Or, was it? I noticed this in the List part properties:

FAQ Properties
Something is amiss

See anything off? Nothing is selected in the view drop down, even though it is selected in the my template site.

Acting on a Hunch

So here was my hunch. Perhaps, the pages are getting deployed before the custom list and custom view, sooo when the page gets made, there's no view to select, which is why it looks like the above. I acted on this hunch, by doing the following:

I split out just the FAQ list portion from the full Get-PnPProvisioningTemplate - essentially doing 2 Gets: one for the list only and one for everything else. Here's what that looked like:

Get-PnPProvisioningTemplate -Out $templateFull -Verbose -PersistBrandingFiles `
-PersistPublishingFiles -IncludeAllClientSidePages -Force
Get-PnPProvisioningTemplate -Out $templateListOnly -Verbose -Handlers Lists `
-ListsToExtract "FAQ" -Force

Now you have 2 files. But there's 1 trick to this, if you want it to work in your favor. You need to open up the XML file for everything, and delete just the ListInstance node for the list (in my case, FAQ) from the XML file. So you can't easily do this all in one full script. You'd have to keep your pulls separate from your applies because of this manual intervention.

Then I applied my 2 files separately as well, starting with the lists first:

Apply-PnPProvisioningTemplate -Path $templateListOnly
Apply-PnPProvisioningTemplate -Path $templateFull

And, viola! My FAQ list was displaying as expected on the page, because the view was already found for the web part property because it already existed.

Coding with Nate

Mike Homol

Mike Homol

Principal Consultant @ ThreeWill

Coding with Nate

This is cross-posted here.

Adventures in getting your kids interested in coding

This is a series of posts I hope to do cataloging some of my progress with spreading the gospel of code to my kids. Recently I made some progress with my son, Nate, and hope to share what we build together in this series. 😊

Part One: Idea and Schema

For many years, I've struggling in fits and spurts to get my kids to embrace coding. I took to it like a duck to water and was coding on this when I was 9 years old:

Behold the Radioshack TRS-80
Behold the Radioshack TRS-80

I've had a few glimmers of hope, like when they were in robotics and were helping to write the programs, but eventually that died out. Then I had a brief period where my sons were doing lessons on That site is fantastic, by the way. But so far, nothing lasting. But I press on.

Something we both like

About a month ago, I challenged my sons to come up with an app idea. My thinking was that they might be more inclined to learn something if it would result in something that originated from their imagination. My son Nate started doing the Swift Playground lessons and things were looking up. But as usual, things sputtered out.

2 weekends ago, we were playing our favorite collector card game, Marvel Legendary. I have amassed pretty much the entire collection and we spent a fair bit of the weekend battling supervillains. There have been some apps made by the community already - namely one called Assemble. It randomizing matchups, which is really nice for a game that keeps growing with 3 or 4 expansions a year. But this app was starting to slow down in its updating. Additionally my son voiced his frustration with not being able to search up cards by the text on them or by their features. Suddenly, I saw the light go on and he turned to me and said, "Maybe this is the app we can build."

Eureka! We had a problem that we wanted to solve and it was surrounding a topic that we both enjoyed. It was the perfect storm.

Early Plan

So the basics for the idea are now there. We want a mobile app that will allow for in-depth searching of thousands of cards and that will also randomize setups of games.

Technically, here's the early plan: a database of cards built in Azure Data Storage, accessible from Azure Functions, via a mobile app built in React Native. We will open source the whole thing, so feel free to contribute or give feedback, especially if you happen to also play the game.


My homework for Nate so far is to start going through the JavaScript lessons on Send me thoughts if you have some better ideas. My thinking is for him to try and pick up some basics of JavaScript while we're building out the backend. While he's learning, he's volunteered to help with data entry. We'll see how long that lasts.

Looking at Cards

Sooo many cards!

Our first order of business was getting the data figured out. My first real teachable moment. We examined various cards and started making notes, first in bullet lists of the various types of cards and the properties we were seeing on them. Here are some examples.

Dark Phoenix MastermindDeadpoolShe-Hulk Villain

As you can see, there are many various qualities and this is just a small sample. In fact, just at the top-most level, there are 10 types of cards: Wound, Token, Officer, Sidekick, Bystander, Scheme, Enemy Leader, Enemy Support, Enemy Group, Playable. Pretty quickly in, we switched to a JSON document and started going over samples of each card type, appending properties to the same single JSON as we discovered new things on a card. This actually sunk in with him. Seeing the JSON get updated live as we were reviewing information on a card was like seeing the objects and properties come to life real-time on the screen. Here's what we ended up with:

"Name": "",
"StartingDeck": "Hero | Villain | Officer | Sidekick | Wound | Bystander | Mastermind | Token",
"Type": "Wound | Token | Officer | Sidekick | Bystander | Scheme | Enemy Leader | Enemy Support | Enemy Group | Playable",
"Classification": "Wound | Grievous Wound | Bindings | Villain | Officer | Madame Hydra | Hero | Special Sidekick | Bystander | Recruitable Bystander | Scheme | Plot | Mastermind | Commander | Henchmen | Backup Adversaries | Villain | Adversary",
"Team": "Shield | Avengers | X-Men | Champions | Fantastic Four | Heroes of Asgard | Sinister 6 | Hydra | Foes of Asgard | Crime Syndicate | Brotherhood | Cabal | Guardians of the Galaxy | Illuminati | Marvel Knights | Mercs for Money | New Warriors | Spider Friends | Venomverse | Warbound | X-Force",
"Class": "Instinct | Ranged | Strength | Covert | Tech | Basic",
"Color": "Yellow | Blue | Green | Red | Silver | Grey",
"Ability": "",
"Keywords": [
"Human Shield",
"Fight or Fail",
"X Uru-Enchanted Weapon"
"CombinationSymbols": [],
"CombinationEffect": "",
"Expansion": "Core | Villains | Civil War | X-Men | S.H.I.E.L.D.",
"AlwaysLeads": "",
"CommandMasterStrike": "",
"VictoryPoints": 6,
"Attack": 4,
"Recruit": 2,
"RecruitCost": 3,
"IsRecruitable": true,
"Setup": "",
"Twist": "",
"NumTwists": {
"1": 9,
"2": 9,
"3": 9,
"4": 6,
"5": 6
"NumHeros": {
"1": 4,
"2": 5,
"3": 5,
"4": 5,
"5": 6
"NumVillains": {
"1": 1,
"2": 2,
"3": 3,
"4": 3,
"5": 4
"SchemeWins": "",
"SpecialRules": ""

Seeing as we will have to start simple, we will begin by manually creating JSON docs for cards, until we get a UI going. So to try and make it easier on us, especially Nate, I set out to make a schema for this sample doc. This route would give us intellisense and document validation in a good editor, like VS Code.

JSON Schema

This was my first foray into JSON Schema. I started at the source and read through this walkthrough: Getting Started Step-by-step. That gave me most of the tools I needed to understand what was going on. Essentially, you're setting up type definitions and other rules as you identify them.

That said, it was still a pretty big doc for me to try and render out the schema manually, so I started with a schema generator - found here - and applied changes from there, namely titles and descriptions, as well as identifying things we missed as I was validating fields. The generator helps a lot with following along, as you can compare your node to the schema generated. The only big thing I had to learn how to add was the enum property. This is a great option for string properties. It allows you to provide an array of options for editors to leverage. Here's an example:

"enum": ["Yellow", "Blue", "Green", "Red", "Silver", "Grey"]

And here's what happens in VS Code when you reference the schema and attempt to add a color field to the document:

End result in VS Code

Pretty cool, huh? Well I thought so at least lol.

'Til next time

So far, the schema is all we have and we're starting to create the data for Azure Storage. I'll share more as we make progress. Hopefully he stays interested and we actually produce a working product in the App Store. We shall see.

The full schema is published here:

Tee of the Week

Mike Homol

Mike Homol

Principal Consultant @ ThreeWill

Tee of the Week

Behold my newest edition! Took forever to arrive, thanks to good 'ole 'Rona, but still worth it. Seriously, where the hell is Don, Carole?!

Where's Don, Carole?

Tee of the Week

Mike Homol

Mike Homol

Principal Consultant @ ThreeWill

Tee of the Week

Dept of Redundancy Department

I've gotten lots of mileage with this one and it's still one of my all-time favorites.

Thoughts on Customer-Centric Nav

Mike Homol

Mike Homol

Principal Consultant @ ThreeWill


Recently, I was tasked with discovering the needs of a customer in order to assess their current state in SharePoint Online and needs for the future. This was for a collection of technology groups that performed a wide array of services for other employees within the company. Early on, we knew that much of what we were evaluating was a lack of access to these services via SharePoint, so much would be created for the first time. Additionally, for the content that did exist, there was a lack of understanding in where the items could be found. Another interesting issue was the lack of understanding of the functions of these groups. Basically, a number of interesting problems, but largely centering around the employees being serviced, or their customers, if you will.

To the Hub!

Based on the feedback, particularly around wanting consistency and knowing that there would be a hurdle with getting folks to know where to go, we decided early on to use a Hub for the job. It solves so many issues out of the box: consistent look-and-feel, consistent navigation, the ability to roll up content from spoke sites and search scoped at the hub level, among other benefits. But just because we now have a spot for folks to come and a place where maybe things look better and more consistent, it doesn't mean that they will suddenly know what to do. Part of the art of helping folks with SharePoint intranets or portals has very little to do with the tooling that is or isn't available in the 365 platform. Much of it is a User Experience and an Information Architecture problem. This may be my heavy marketing background at play, but it felt critical to me to deal with the hub needing to be customer-centric from the get-go, particularly the navigation.

Ways to Solve Navigation

There were many thoughts on the navigation, both from stakeholders and discovery. The default thinking around the Hub navigation was to break it up by the various domains or spheres of teams that were solving problems or serving customers. One of my first thoughts on navigation was to look into taking it towards a task-based navigation, as it seemed like customers may only know what thing they may need to do. In my UX research, I encounter the following 2 articles that I wanted to highlight:

Object-focused vs Task-focused Design

Audience-Based Navigation: 5 Reasons to Avoid It

These are great resources for gaining insights into user habits and trends. So what is actually the best way to have a navigation that keeps the customer at the forefront?

Topic-based = Customer-centric

The bottom-line is that typically task-based navigation can be very confusing to most users. So that was a wash. What becomes clear from UX research is that there isn't always a silver bullet. But UX is an iterative process. So let's find a solid starting point. Based on those articles and knowing that we still needed information to be somewhat grouped by communications site spokes - we still need to be thinking about the authoring challenges too - it seemed like our best starting point would be to go with topic-based navigation. This would allow the sub-options to still be domain or sphere-based, but should provide a customer with the terminology that they can relate to. The customer needs to see words that represent the things that have brought them out to the hub.


Let's look at a quick example of one of these domains: a group that we will call PMG. They were responsible for presenting resources pertinent to PMs and the PMI certifications including in-house governance resources. Originally, we had a top-level navigation called "PMG" with sub-options. But instead we landed here:

As you can see, the topic of "Project Support" aims to be a topic that should gravitate these PM users' eyes to the item that matters most to them. Is it fool-proof? No. But it's a solid start.

Final thoughts

IA and UX is 2 parts science and 1 part art, in my humble opinion. I have worked on countless websites (namely at Brown Bag Marketing) over the past 10 years with Creative and UX resources. I've seen the sheer amount of different opinions mixed with a bevy of different facts, which also tend to change. But some things have remained true: the customer needs to be at the forefront and the calls to action need to be clear and take precedence. These are just some random thoughts in that arena. Hope you found it useful.

Tee of the Week

Mike Homol

Mike Homol

Principal Consultant @ ThreeWill

Tee of the Week

I love the colorblind?

This week’s tee is off to a great start today. Got some compliments and laughs at the doctors office. 🤪

New Space, Who Dis?

Mike Homol

Mike Homol

Principal Consultant @ ThreeWill

My journey with blogging has been a complete mess so far. But I keep trying, so I guess that's something. Like so many people, the busier I get, the less I think about doing this. But I want to make this a better habit because it helps me reenforce the types of creativity I espouse.

For a long time, I've used Tumblr, but that just seemed like I was taking it too easy. My alternative has also been to piggy-back on my company's blog - again, also too easy, and just doesn't feel like "me."

So I set out to finally make my own site, which could start to be a little more "me" and start to allow me to combine having my own presence and be built on the things that give me joy.