How to get paid for building new features

Innovation Fatigue

“I wish you guys would stop releasing new features.”

It’s 2015 and a typical sunny day on the Florida coast. I am showcasing the OTAS application to traders at our largest client. Its European office is full of power users, but adoption elsewhere is pedestrian. These traders haven’t bedded in one innovation before we release another, and as a result they are tuning out.

I recalled this cautionary tale while watching the response to Jenson Huang’s big announcements at last week’s NVIDIA GTC25. Huang envisages a 900-fold leap in GPU performance between today’s most common architecture and the next but one iteration.

Core customers who have yet to realise an adequate return on investment, are being asked to start making next generation commitments. As well as the cost of the units, data centres will need significant upgrades to move from air to liquid cooled servers. Upfront costs are high, meaning even the promised major reductions in running costs will take time to recoup investment.

In sales there is always a temptation to believe that additional features will accelerate revenue growth. When core clients work with you to develop those features, it is easy to believe that this will increase adoption by new clients. The opposite may be true.

Cheerleaders for Change

When OTAS was bought by Liquidnet in 2017, I moved to New York to oversee the roll out of the product. The US is where financial software is made or broken, and selling the company to an established presence was a logical progression. There were, however, two new issues to consider.

Liquidnet’s software is the market leading way to trade a lot of shares in one go. This avoids leaving a footprint that automated trading algorithms can track. Clients pay for this ability to trade a lot of shares and do not need analytics to help them time transactions. We had to make changes to the application, to downplay the charts and elevate the predictive cost features.

Like NVIDIA’s customers, OTAS’ new owners had spent a large sum on the technology and were not minded to spend a lot more. They had their own ideas for improving the customer experience, which were separate to anything OTAS had planned. Some of these ideas came from traders, but many more from the burgeoning product team.

Scaling a business requires careful consideration of return on investment. This means both yours and that of your clients. The time it takes them to realise returns dictates in large part the pace of your development pipeline.

Product and technology teams always want to work on innovations, and inexperienced sales people can become cheerleaders for change. Managing their expectations while making difficult decisions about what to build, is the hallmark of leadership.

Differentiate by Use Cases

I expect Huang knows what he is doing. His vision is to transform industries by enabling robotics, quantum computing and brain-computer interfaces. Accurate real-time decision making requires the scale of capacity expansion his teams are planning. When your business model is innovation, speed to market is everything. While NVIDIA is the dominant AI company today, Huang knows his future customers will be bigger than their supplier.

For most of us still scrabbling for AI use cases, the ability to do much more, far faster, feels unnecessary. We will not want to pay for the latest technology until our own return on investment from AI is clear. Some of NVIDIA’s customers will target us, while others push the technology frontier.

The accelerating pace at which NVIDIA innovates segregates its client base. Those who intend to supply the emerging industries of the next few decades, will make whatever investments are necessary. New GPUs will continue to be eye-wateringly expensive, while a second-hand market may develop to serve the rest of us.

Segregate Clients by Price

The lesson for business owners is to use price to segregate clients when product offerings are differentiated. Even companies selling cans of soda at a single price point, experiment with higher priced flavour varieties. Companies supplying services must price differentiate between clients, or else margins will collapse as clients’ demands increase.

Do not hesitate to charge more for those who demand new features. As with OTAS, those features become increasingly tailored to specific company workflows and not what mass market clients want. By ensuring there is a path to upgrade and downgrade services, just as NVIDIA’s modular products are forwards and backwards compatible, we can develop both a broad market and a deep niche at higher prices.

Questions to Ask and Answer

1. Am I selling a single product or a range of differentiated services?

2. Do I charging more demanding clients an appropriate premium?

3. What new products can I launch to increase my market at higher prices?

3 ways I can help:

1. Book a consultation to talk about all things AI.

2. Read our deep dive into data science.

3. Explore use cases using accelerated computing.