If you are a user of a Business Intelligence (BI) system like MicroStrategy or SAP, one thing clear is that BI users love their drilling.
The ability to start at a simple, high-level dataset and drill down into details has become an invaluable staple of data analysts.
Maps are also an essential part of business solutions so it also makes sense that users expect to drill on maps as well - combining the power of drilling and simple map navigation like zooming and panning creates an easy-to-use but extraordinarily powerful analytical experience.
However, there are some decisions when it comes to drilling on maps: How does a user activate a drill? How much data can a view handle? How does the user go back to a starting state?
Let’s get started with the requirements:
Why users want to do drilling?
- Simplify the map view and navigation
- Make large datasets more manageable.
- Compare two levels of data at a time.
What are the expectations of your users?
- Are they Dashboard or Mobile users who simply want to click and see the next level of data?
- Are they Analysts who want access to more data for analysis?
- Are the users willing wait for detailed data queries?
- How many levels and how quickly will they drill and reset the drill?
- Do they want full access to the entire data repository via a drill or will they follow a directed workflow to a more limited data set?
Each of these questions should help guide you on what your best options are for drilling. We will explore all types of drilling here and point out the pros and cons.
Let’s start with the most basic form of drilling which is what we will call ‘Standard Drilling’. This is what most will think of when we talk about drilling. The UI asks for more data and a powerful, intelligent server fetches new data for the client. Let’s first picture drilling without maps… you are on a grid of State level data. You click on an individual row of data (states) such a Texas and activate the drill mechanism. Most BI systems have the infrastructure do a new data fetch for the next level of data requested by the user. In this case a list of all Zip Codes in Texas will be returned. The data is filtered to a manageable size and exactly what the user asked for.
Now let’s look at this on a map, which works the same way. A well-integrated mapping system will always create a map of whatever data is returned by the BI system. Instead of showing a grid, we just show a map of the data in the system. If the system allows it, the map can be the starting point to activate the drill as well. Start with a map of all US States, select a state and activate the drill. The map system will tell the BI system to fetch new data and bring back Zip Codes in the selected area just as it did on a grid. The result is a Zip Code map nicely filtered and zoomed on the area requested.
Pros: Works out of the box in most cases, users have access to very large datasets that are managed by a server backend, users can drill in many different directions and even skip layers of data.
Cons: Complex. Knowing first that you can drill, selecting an entity to drill, choosing a drill target, triggering the drill, waiting for the system to fetch data and watching the UI reinitializes the data view (flashing UI). Oftentimes you have left your original report state and also need to know how to navigate back. This process is not quite natural to simple end users and is typically considered analyst-only.
Linked Report Drilling
Similar to Standard Drilling, Linked Drilling may appear to be the same to the end user but how it functions is different. In some systems, the drill hierarchy (relationship between high and low level data) is either not supported or just not possible.
Consider the following scenario. Your BI system has general, aggregated metrics about your stores but it doesn’t have all the details about the stores because that data is in the operational system managed by another system. That system will respond to URL-based web report requests but there is nothing automated about it. In this scenario, the map administrator could create a ‘link’ between the first report (map) and a second report (map) that can be called dynamically. The action from the user is to click on Texas, activate a linked drill and get back a Store details within Texas. Behind the scenes, the map system is taking the data about the selected state and passing it to the second report as a filter.
Pros: Administrators can link any two data sets as long as there is a matching data item.
Cons: All of the same Cons as standard drilling except it is not out-of-the-box and requires explicit configuration for every map.
This type of drilling is very popular among client-side dashboarding systems. In technical terms, this type of drilling is often little more than applying a filter and may not be considered by many as true drilling. However, for our conversation this is still drilling from the user perspective.
The user interacts with two or more sections on the dashboard, those sections can be two grids, grid-map, map-grid or map-map. One of those sections is the selector visualization and the others are the targets of the selector.
In our case above you can have two maps side by side: states and zip codes. Click on Texas in the left side and the Zip Codes in Texas appear on the right.
In most systems, the data you need is right there in your working dashboard instance, so the response to the click is immediate and the users can explore quickly.
Pros: Very attractive, Easy to navigate, very responsive with fast data delivery, see both levels of data side by side.
Cons: Dataset sizes are limited, need large portions of screen real estate to accommodate the different views, typically not mobile friendly and may mean smaller maps (who wants that?)
One type of drilling unique to the map world is Zoom Drilling. This type of drilling is what you will most commonly see from map vendors, but as we discuss it you will see that it has benefits and problems.
The administrator can set up two levels of data so that different pieces of data appear as the user zooms into the map.
Each ‘Layer’ of data is locked to a specific Min/Max scale so that it will appear and disappear as the user navigates higher and lower.
Here is the configuration window for locking layers of data to specific zoom levels.
Now picture a map user looking at Texas data and zooming into Texas for a closer look.
When the user reaches a specific map zoom level, the state level view of Texas disappears and the Zip Codes in Texas appear.
In fact, all Zip Codes appear and there is no filter applied. The system simply shows all data in the bounding box that is the user’s screen.
One specific issue that we see with this type of drilling is what we call the Big Shape-Little Shape issue. Start with locking zip codes to a specific level.
You zoom into Texas and tell the system that you want Zip Codes to appear at scale: 8,000,000. (and of course, states disappear at that level) You test it by zooming in and out to watch Texas disappear and Zip Codes appear.
Perfect! Now you pan over to the northeast the test it on Rhode Island. As soon as you zoom in on the northeast, all Zip Codes for that entire region appear.
You weren’t expecting Zip Codes until you got much further in and would prefer to see States at this level.
Then you realize that Texas is almost as big as all of the northeast states combined. O
ne scale for the entire layer may not work for you…
You would prefer a scale of 2,000,000 for these smaller states but you cannot have different scales for different regions of the same layer.
You will probably have to split the difference and compromise the ideal view.
One other issue that we have seen with Zoom Drilling over the last 15 years of delivering map solutions is that end users are often more simple than expected. There are still many users out there that find a task as simple as zooming to be foreign to them. Drag-pan and mouse-wheel zoom actions which map analysts use instinctively are not used by some end users. They will still click on the plus/minus icons or arrows to pan and zoom a map. This makes Zoom Drilling quite painful and non-intuitive when you are using buttons to pan and zoom. Mobile and touch technologies are changing this mindset but on web browser mapping this is still a reality.
Cons: Still not simple enough for all users, no filtering of data, big-shape/little-shape issue.
This type of drilling is new to Visual Crossing and we have mostly seen this only in custom mobile apps. It was designed to make Zoom Drilling better by solving the big-shape/little-shape issue, adding filtering and making navigation single click/touch which any user will do instinctively.
The scenario is simple… you see a map of your Regions, you touch on Southwest and the data is filtered and the layer changes to states that are ONLY in the Southwest Region. If we continue and click on Texas we get Zip Codes in Texas. A ‘Back’ or ‘Reset’ button appears on the map view to allow the user to go back up one level or entirely reset to their starting map. One click drill, one click reset… that’s it. The working mechanism behind this map utilizes Zoom Locking but it does it dynamically based upon the bounds of the entity selected. A single UI click is telling the system to do many things for the user which gives them the view they want.
Our scenario below was only two clicks to zoom from Regions to States to Zip Codes in Texas. One more click would reset this map back to the original state:
Pros: Easiest drill for end-users, works extremely well on touch devices as well as web, only one map view required, easy to navigate many levels and compare levels, fast displays of data.
Cons: Configuration is required, fixed path drilling without branching, not designed for advanced analyst types, data set sizes are limited as with all client side data drills.
Try it yourself!
Now it is your turn to provide the correct type of drilling to your users. The good news is whether your are using Microsoft Excel, MicroStrategy or SAP, you have many options to deliver a map drilling solution that fits your user base for both ease of use, performance and overall capability.