Here’s a story from a friend…let’s call him Frank. Frank and his team had been building an analytics and reporting platform for hospitals for the past year, iterating with a small clinical practice. Frank knew they were on to something compelling: 100% of the physicians and administrators were using the platform daily and giving high feedback scores. Setting his sights on bigger fish, Frank hustled his way to demo the platform to the largest healthcare provider in California—but his team only had 1 week to prepare. They couldn’t just demo the instance at the small clinical practice to the big opportunity, so they scrambled to create an entire fictional practice with all the associated data, including fake medical records, clinical notes, provider employment data, patient surveys, etc.
Frank and his team finished late the night before the presentation. The next day, in the middle of showing a quality of care metric for the fictional provider group, the Chief Medical Officer grunts: “Nobody has a Case Mix Index greater than 4, furthermore, that group of patients you showed on the last screen should be dead with lab results like that. What you’re showing me is just wrong. How can I trust that you understand our business?” That was the moment Frank realized that he had lost the CMO’s attention. The rest of the meeting spiraled from there. Twenty minutes later, he left the board room with a friendly, but clear message from his sponsor: “We’ll be in touch.”
Product demos need data. good, relevant data.
Anyone who’s built, sold, or purchased an enterprise or SaaS product will be all too familiar with the sales demo. These come in many flavors these days, from free trial accounts with canned walkthroughs (Alteryx) to pre-recorded videos (Tableau), and from scheduled demos over a screen share (Mode) to vendor booths at conferences. Each of these demos needs compelling data to highlight their products’ features and make the customer feel like this could be something that works for their crazy, unique, extremely rare breed of business. Unfortunately, demo data is often an afterthought, much to the detriment of trying to deliver a compelling demo. Representative demo data is so important to enterprise sales, Microsoft is known to have created over 30 fictitious companies, largely for demo purposes.
There’s no lack of sales advice in circulation: “The goal of the demo is not to demo your product,” “Don’t sell past the close,” “Think like your prospect,” “Sell what they need”…. But talk to someone who sells with a product demo and one of the top pieces of advice you’ll get is “Customize the demo” (link, link, link). Real vignettes like these hammer the importance home: “Why are you showing me a demo using equities data? This doesn’t make sense to me, I just want to predict customer churn” or “Our Dutch clients will not understand a Bulgarian address format or a Bulgarian name – they cannot recognize which is the first and last name.”
Ok, so how do I get better data for my demos and bypass the dreaded uncanny valley?
Tip 1: Publicly available data
There are A TON of interesting datasets in the public domain, so many that it can be a little overwhelming if you go looking for a dataset with no restrictions. Take this list of Awesome Public Datasets that has over 30+ categories with links to 100s of sites, each with anywhere from 1 to 1000s of datasets. That’s a lot of options.
I recommend first thinking about the type of data will best show off your product, but be flexible because it’s unlikely that you’re going to find the exact dataset that you need for your product demo. For example, a friend’s startup was showing an early demo of their new customer support AI but they didn’t have access to a company’s internal customer support ticket system, so they improvised and used StackExchange and Wikipedia data. These datasets are in a similar question-answer format that can still highlight a product’s capabilities.
Sometimes you might get lucky and find a dataset that’s exactly what you need but it doesn’t cleanly fit into your data model. Like this anonymized B2B Sales dataset which would be perfect to show off your new Salesforce killer CRM but will take some work to load into your choice of backend, MongoDB. Additionally, it’s likely that the dataset you find will fall short and you’ll have to augment the data with more interesting trends that tell a story.
Tip 2: Create data from scratch
Just fake it until you make it, right? You can theoretically create any dataset you want but I have two strong words to the wise here:
- you need to have a deep understanding of the business and the data that you’re mocking, and
- for a complex dataset (read: many tables, multiple interdependencies), mocking is very hard, and it’ll take a lot of time to make it realistic.
We’ve personally spent weeks building a demo dataset to close one deal. Granted, it was an eight-figure multi-year enterprise deal with a large bank. We also know of a public company that invested heavily in building the perfect dataset to distribute with their software for training and tutorials. The individual who built this spent 3+ months and pored over excel to make it happen.
There are several resources to help you get started with creating data from scratch. If you’re a developer, make sure you use one of the Faker libraries (perl, ruby, and C#). Also, check out our post on using Faker to generate realistic test data. If you don’t want to code, start by using one of the many online fake data generation tools, such as Mockaroo, and add your finishing touches in Excel or Google Sheets.
Tip 3: Start with existing data
If you’re fortunate enough to have access to a dataset that would make an amazing demo but you can’t use it due to legal or privacy reasons, you might be able to use the dataset as a base to create a new dataset. If it’s a 3rd party’s dataset, you should first check any relevant contracts and speak with your lawyer to ensure that you have derived data rights.
There are many data anonymization approaches to create a demo dataset from a source. Most fall under one or several of the following techniques: permutate, redact, mask, generalize, perturb, encrypt, or synthesize. The pros, cons, and technical details of these approaches are beyond the scope of this article but don’t fret: we’ll be writing more on the subject soon, so keep tabs on this blog for updates.
In short, if you have a good dataset, derived data rights, and the right tools or technical knowhow, this is one of the quickest and most reliable ways to create compelling datasets for sales demos.
Sell more with better demo data
Don’t let bad or unrelatable demo data shut down an opportunity before you even get through a demo. Good demo data coupled with a strong pitch can help a customer visualize what’s possible with your tool and how it will solve their current pain.
It takes time to create good demo data but when you’re selling a six-, seven-, or eight-figure enterprise deal, it’s well worth putting the time in.