Y Combinator veterans Dalton Caldwell and Michael Seibel are warning startup founders about a problem that can hide behind seemingly good news: the customer who brings in revenue but gradually takes control of the entire company.
In a recent episode of their startup advice series, Caldwell and Seibel use the deliberately provocative term “terrorist customer” for customers who effectively hijack a startup’s time, resources and product roadmap. The discussion argues that founders sometimes need to walk away from revenue when a customer relationship is actively damaging the business.
The terminology is metaphorical. The hosts are describing business relationships that create a persistent negative cost-benefit equation—not actual terrorism.
When a valuable customer becomes a threat
According to the discussion, the problem often starts innocently.
A young startup lands a large customer, perhaps a major brand that becomes its biggest source of revenue. For a company struggling to gain traction, the customer can appear to be a lifeline.
But that dependence can give the customer enormous leverage.
The startup may begin changing its roadmap around that one account, building specialized features, providing excessive customer support or accepting projects that are difficult to monetize.
Eventually, the founder may discover that the company is no longer building a scalable product. It is effectively building a business around one customer’s demands.
Caldwell’s central warning is that founders should not confuse revenue with healthy revenue.
A customer can generate substantial sales while simultaneously making the company less scalable, less focused and less profitable.
Not every customer request deserves a “yes”
The hosts also distinguish between a problematic customer and what they call a “terrorist request.”
A good customer can make a bad request.
For example, a customer might ask a startup to build a feature that will not meaningfully improve its business, develop software outside the company’s core product or undertake a project that will cost more to deliver than it is worth.
The instinctive response from a young startup is often to comply.
After all, the customer is paying.
But Caldwell and Seibel argue that this creates a dangerous mental model: the customer asks, the startup builds, and everybody wins.
That assumption is not necessarily true.
Founders need to understand what actually makes their customer’s business better rather than simply implementing every requested feature.
Customers don’t always know what they need
The conversation expands into a broader product-development lesson: customer feedback should be interpreted, not followed literally.
The hosts point to Facebook as an example. Users might request individual features, but the most important thing Facebook needed was not necessarily another feature. It was growth—getting more of a user’s friends onto the platform.
The same principle appeared at Twitch.
Streamers asked for various improvements, but they did not necessarily ask for ways to make more money because many of them did not believe monetization was possible.
The company therefore had to look beyond explicit requests and examine the circumstances of its users.
One revealing observation was that some highly successful streamers were still working ordinary jobs. That suggested an opportunity the users themselves were not explicitly articulating: they had valuable skills and audiences but lacked a functioning economic model around them.
For founders, the lesson is straightforward: the loudest feature request is not always the most important problem to solve.
The bigger problem may be inside the startup
Caldwell and Seibel extend the same metaphor beyond customers.
They argue that startups can encounter similar problems with employees, recruits and investors.
A particularly risky employee situation can begin during recruitment.
A founder desperate to hire may encounter an exceptionally impressive candidate and become so focused on securing the person that they overlook an escalating list of demands.
The candidate’s reputation, previous employer or specialized expertise can make the founder rationalize every warning sign.
But if those demands continue after the person joins, the original problem has not disappeared. It may have only become more expensive.
The hosts make an important distinction: a difficult employee is not automatically a bad employee.
Some difficult people create extraordinary value. In those cases, the company’s cost-benefit calculation can still be strongly positive.
The problem is the employee whose cost-benefit equation remains negative despite repeated attempts to make the relationship work.
Investors can create the same problem
Investors can also consume disproportionate amounts of founder attention.
An investor does not necessarily need to be hostile or unreasonable. Repeated questions, requests for detailed updates and long exchanges can gradually turn into a major time commitment.
A founder can end up spending hours managing an investor’s concerns instead of building the company.
The hosts also discuss a more subtle problem: an investor who has lost confidence in the founder.
Because investors can occupy an unusually influential role in a founder’s life, persistent pessimism can have an outsized psychological effect.
That leads to another theme running through the episode: good startup advice requires both candor and genuine belief.
An advisor should be willing to identify the company’s biggest problems, but should also genuinely believe that the founder can solve them.
The sunk-cost trap
Perhaps the most important lesson is psychological.
Founders often know a relationship is damaging but struggle to end it because they have already invested so much in it.
They tell themselves:
Maybe the customer will get better.
Maybe the employee will change.
Maybe the investor will become more supportive.
Maybe one more concession will solve the problem.
But every additional concession can create another demand.
That is why the hosts repeatedly return to the idea of walking away rather than endlessly negotiating.
The startup’s limited resources—especially founder time—are too valuable to spend indefinitely trying to repair relationships that consistently damage the business.
Revenue isn’t the only metric that matters
For early-stage founders, the message challenges one of the simplest startup instincts: grow revenue at all costs.
Revenue matters enormously, but its quality matters too.
A customer generating $1 million in revenue may be less valuable than several smaller customers if that single account requires extraordinary customization, support and engineering resources.
Likewise, a highly prestigious employee is not necessarily a great hire if their ongoing cost to the organization exceeds their contribution.
And an investor’s reputation does not automatically make the relationship valuable if managing that investor consumes a disproportionate amount of the founder’s time.
The broader principle is that startups need to evaluate relationships based on whether they strengthen the company—not merely on how impressive they look on paper.
“Dodge them” rather than negotiate forever
The hosts’ final message is intentionally blunt: founders should learn to identify these relationships early and avoid them.
The goal is not to label every demanding customer, difficult employee or skeptical investor as a problem.
Instead, founders should recognize the much narrower category of relationship that repeatedly consumes resources, creates new crises and produces little or no corresponding value.
For startups operating with limited money, people and time, those relationships can become particularly expensive.
The answer may not be another negotiation.
It may be knowing when to cut the relationship, accept the sunk cost and move on.
As Caldwell and Seibel argue, these people and situations do not necessarily determine whether a startup succeeds or fails. They are obstacles on the road.
The founder’s job is to recognize them early enough to steer around them.