Creating an iterative software product means adapting to new information and feedback from your customer. If a particular feature does well with your users, you’ll want to explore it more. If a feature doesn’t perform well, you need to understand why so you can avoid that mistake in the future.
The key to iterative development, really, is learning. The quicker you can learn about your customer and what they want out of a product, the quicker you’ll be able to make something that appeases the market.
In our post on the Lean Startup and your SaaS, we talked about validated learning and the build-measure-learn loop. In order to inspire learning, you have to measure whatever you’ve built. So if our goal is to learn as much as possible, we need tools in place to measure our product increments.
In a perfect world, we would have the resources to call every customer and ask them what they think about new versions of the product. If you’re operating a low-touch model with only a handful of customers, this may be an option for you.
But if you’re like most SaaS products, you have (or are aiming to have) more customers than you can call in an afternoon. You need specialty tools in place ready to capture report data from your SaaS.
1. Usage analytics
Image: Wesley Fryer / Flickr
Analytics are your hard, cold numbers that indicate how customers are using your product. It will tell you which pages they visit the most, which buttons they click, and how they are interacting with each element.
You should collect as much data as possible, even if you don’t have an immediate use for it. This way you’ll always have a history to pull from if you decide to look at a new metric. For instance, if the number of people who use a particular feature suddenly becomes relevant, you would want to be able to look back in time, rather than have to wait a few months for new data to arrive.
Your analytics tool should offer a customizable reporting feature. Management, stakeholders, or other departments might need regular reports. If you need a unique piece of data for some reason, you’ll want to be able to generate a custom report. The best tools have customizable dashboards for quick referencing.
Our recommended usage analytics tools: For small to midsize user bases, use Google Analytics. Stop using it once you’re getting 10 million page views per month. Large apps should use Adobe Analytics or Adaptive Insights. You should also supplement your data with a tool that helps you define analytics in terms of your business objectives. We recommend Kissmetrics.
2. Heat maps and click maps
While there’s no way to tell exactly where a user looks on the screen, mouse movements are surprisingly similar to eye focus.
Heat maps and click maps are visual representations of a page of your website or a screen of your application that show how users use their mice.
On a heat map, mouse motion is indicated by color. Red indicates where the most action occurred and blue indicates no action. (Depending on the tool you use, no action might be marked by no color.)
We can use a heat map to understand where users are looking for information. For instance, you would probably notice that your users search for information in the traditional F-pattern, starting at the top left. In this case, you should put the most important information where they are inclined to look for it.
Heat maps aren’t perfect. Sometimes users read and search without using their mouse. Some people are proficient with their keyboard. Sometimes we flick the cursor far to the side so it’s out of our way. But once you begin to aggregate data from lots of users, clear patterns emerge.
Click maps track the locations where users click on a particular page or screen of your application. Data is aggregated with dots. More dots on a particular point equate to more clicks.
A click map can tell you which areas of your site/app users prefer to click. Do users expect something to be interactive that isn’t? Are they failing to click things they should?
3. Video session replays
A video replay of a user session can’t be quantified or aggregated with other pieces of information, but they are still useful ways to get into your users’ heads. You can actually watch how a user used your website or application.
For example, you might watch a user abandon a form halfway when prompted to submit a phone number. Without ultra-sophisticated analytics, you would have never known whether your customers are comfortable giving that information. Perhaps you could encourage engagement by removing that requirement.
Sometimes just being able to witness the user’s hesitation is enough to give insights on how to improve your product. Did they look all over the page for something? Did they scroll back and forth a lot? Did they spend an unusual amount of time reading a section of copy? It’s hard to quantify what you’ll learn from a session replay, but they are worth your time.
4. Customer feedback surveys
Surveying your customers is an excellent way to turn qualitative data (their feedback) into quantitative data. They can be distributed to customers in bulk (as a one-off email to your subscribers, for instance), based on triggers you set, or within your application.
The most effective type of customer survey uses the Net Promoter Score. This format works well because it’s short, easy for customers to complete, and collects a quantitative metric as well as some qualitative information.
An NPS survey asks two questions:
- How likely are you to refer this product to a friend? Select 1 through 10.
- Why did you choose that number?
The first question gives you something to aggregate across all of your customers. If your average score is 3, it’s safe to assume your users don’t enjoy your product very much and changes are necessary.
The second question gives you feedback regarding what to change. Yes, going through each response can be tedious. We recommend grouping these replies into buckets that correlate with ways to fix or improve your website or application.
For instance, you might create buckets called Design, Performance, Reporting, and Integrations. If a user complains that your reports don’t offer much detail, add a tick to your Reporting bucket. If a user complains that he couldn’t attach his Facebook account, add a tick to Integrations. Once you’ve gone through each complaint, tally each bucket’s score. If Design had the most ticks, that’s where you should direct your developers to iterate.
Ideally, you should make your buckets as specific as possible. Instead of reporting, you might have Make Reporting Simpler or Wants Exporting Feature or Wants Reporting Dashboard.
Longform surveys can be useful, as well. They can help you gather a wealth of qualitative data. The drawback is that response rates are generally low because you’re asking for a lot of the user’s time. They’re also too big to be delivered within your application. You have to send them via email.
(Disclosure: Ask Inline is our product, but we’ve built it to be a better tool than other customer survey applications.)
Get creative with data collection
This list was by no means exhaustive. Your SaaS product is unique. Depending on your customer, your industry, and the technology you use, there might be other ways to capture data about your users. Feel free to use whatever method helps you achieve the most learning so you can build a better product.