Outdated distinction: customer support vs R&D

Most companies I work with are focused on R&D. As product development is viewed as the wellspring of all new products and, by extension, revenue, the function is often protected from external influences by putting gatekeepers in between the customer, suppliers and other ‘distractors’ and the folks in R&D. Generally, the thinking seems to be that most engineers are introverts who prefer to work on their thing with big headphones on, blocking out distractions, and are better left alone.

As long as we work in a relatively stable context where the primary differentiators in our offerings are driven by technology, the approach works fine. We can allow engineering and R&D to focus on converting new technologies into products and offerings that can do things we couldn’t do before. Customers will buy what we can create not because of our customer intimacy but rather because of our technology leadership.

One illustrative example is AI. The large language models (LLMs) and large multi-modal models (LMMs) are so impressive that we care less about anything else except their technical abilities. So, the leading companies in the space invest heavily in R&D and infrastructure as they know we’ll show up no matter what as long as they provide the technology.

The challenge is of course that technology leadership is extremely difficult to maintain as a strategy. In practice, new technology tends to become available to everyone in an industry at roughly the same time. If you build the technology yourself and have found a profitable niche to operate in, others will do their darndest to enter that profitable space as well with their own investment in technology.

Returning to AI, the recent release of Deepseek V3 sent shockwaves through the big US tech companies for exactly this reason. The belief was that by investing billions in scaling LLMs and LMMs to larger and larger models, the leading companies, such as OpenAI and Anthropic, were convinced they had a moat around their business through technology leadership. It now turned out that others could provide similar outcomes with smaller models and simpler hardware and that the perceived moat was non-existent or at least not as deep and wide as everyone had expected.

In my view, this is an incredibly positive development as humanity won’t benefit from technology that’s highly controlled by a very small group of companies but instead captures the true value of technology when this technology is democratized and is widely available at low cost. It’s only then that the most impactful and valuable innovations on top of that technology are created. The earlier everyone has access to cheap and abundant artificial intelligence, the better it is for us as a species.

When an industry shifts from being technology-driven to being customer-driven, which is where virtually every industry is today, we need to change the way R&D interacts with the rest of the organization and the business ecosystem around the company. One implication is the adoption of DevOps. The faster feedback loops between customers using our offerings and product development allow for significantly improved R&D effectiveness.

One interface and function also heavily affected by this is customer support. Especially in tech-heavy companies, this activity is often treated as the stepchild and regarded as an expense where call centers in low-wage countries are often viewed as the optimal approach.

However, in the companies we studied during the adoption of DevOps, we noticed that this traditional approach to customer support failed to support our customers in the way it did earlier. Due to the large gap and boundaries between R&D and customer support, the folks in customer support were always one or more releases behind. Hence, customers got support and feedback based on outdated knowledge, which didn’t work at all.

In the same way that DevOps removed the distinction between development and operations, we also need to remove the distinction between development and customer support. The continuous feedback loops between customers and R&D are vital for iterative development. They allow us to build functionality that actually delivers value to customers and iteratively improve it based on the qualitative and quantitative feedback from customers.

By being more exposed to customers, the engineers in R&D will naturally develop more empathy with them, understand their needs and realize functionality in ways that better suit the customer base. For this to work, customer support needs to be integrated with R&D. In this way, those most focused on customer support know what will be included in upcoming releases and can prepare their material and knowledge base accordingly.

A related topic is concerned with experimentation. As we adopt A/B testing to determine whether new functionality delivers value to customers, those working in customer support need to know which experiments are in progress and what customers are subject to them. Otherwise, unexplained behavior or perceived defects can easily be escalated even if they happen to be part of an A/B test that customer support is unaware of.

Rather than treating customer support as the stepchild in the organization that needs to be optimized for minimum cost, it should be treated as one of the most valuable sources of market feedback that should inform product development continuously. This requires R&D and customer support to be integrated, similar to how operations and development are integrated when adopting DevOps. To quote Derik Sivers: “Customer service is the new marketing.” In a world where customers can leave you at a moment’s notice, customer satisfaction and having customers tell others about you is the most powerful business strategy of all.