“A relationship, I think, is like a shark. You know? It has to constantly move forward or it dies. And I think what we got on our hands is a dead shark.” — from the movie “Annie Hall”

Most high-growth businesses stare down periods when growth unexpectedly slows down or stops altogether. At some point, that stomach-churning moment has visited the leadership of most of the companies I’ve advised. We also faced it at OpenTable, eBay and Reel.com when I was managing them. And it has reputedly happened to a number of today’s mightiest Internet businesses as well, including the likes of Amazon and Facebook.

CEOs at high-tech businesses work hard to keep as much growth going for as long as possible. Investors — both public and private — tend to value growth over everything else during most investment cycles, and the recent cycle has been no exception. Growth is where all the action is, and to where all the money flows. The reason for this is that the majority of returns are driven by a handful of companies that “break out.” And a business can’t get big enough to break out if it isn’t growing.

Of course, there are a number of key performance indicators for managing a business well and toward profitability — which we’ve written about here and here — but there are still times in most growth companies where that growth trajectory was interrupted. Sometimes it happens gradually, but in a surprising number of cases it happens suddenly: The business is growing one day, then it’s not growing the next.

I’ve taken to calling these growth hiccups “Oh, shit!” moments for CEOs. So what do you do when the growth rocket judders?

Entrepreneurs’ reactions to these moments fall into one of two extremes: There is the preternaturally calm CEO who resolves to watch the metrics closely to see if they change; he or she is concerned that a hint of panic will cause the rest of the team to panic, so they “Zen it out” as best they can. Then there is the CEO who freaks out, setting off alarm bells that reverberate throughout the entire building and continue ringing in every single employee’s ears.

Which approach do I advocate? The freak-out, of course!

Why? Because the unexpected slowing of growth in a “growth” business presents an existential risk to the company. Growth rates over a company’s history tend to move only one way over time (down); even in hypergrowth companies, growth rates tend to fall to earth … which is why I’ve referred to this effect as “gravity.”

Once gravity takes hold, it’s very hard to reaccelerate the growth of the business. Slowing growth portends a strong possibility that the company will never again experience prior levels of growth going forward. There are precious few examples of this happening on a sustained basis: Amazon accomplished it during 2010-2011, contributing to its value today. We accomplished this during my tenure at OpenTable, leading to a fourfold growth in market cap over six quarters.

But growth will never resume magically.

If CEOs can figure out what exactly happened to cause that slowdown, it presents them with the opportunity to try to correct what went wrong. Yes, they’re still freaking out — but with a plan to fix things. Here’s the process behind getting to that plan:

Drop everything else and focus exclusively on figuring out why.

At eBay, we took to calling these “911s,” basically declaring a state of emergency and clearly establishing that nothing was more important than diagnosing the cause of the slowdown. We got all hands on deck: Meetings and travel were canceled, a war room was set up, copious amounts of food and caffeine were brought in.

Involve all the key people whose skills/expertise is needed to diagnose the underlying cause.

Trying to keep the situation secret from the rest of the company so they don’t “lose faith” is futile. First, they’re going to know sooner or later (and frankly, nothing will cause an organization to lose faith more quickly than observing that its CEO appears disconnected from the reality of the situation). Second, you are far better off enlisting them than falsely shielding them. A well-coordinated group of people can cover more surface area more quickly than just a few people.

Search maniacally for underlying causes.

Growth doesn’t screech to a halt for no reason. It’s usually because something changed, even broader forces outside your company. Maybe a key upstream source of traffic made a change, such as Google shifting its SEO algorithm, or Facebook changing its NewsFeed algorithm. Or maybe critical partners impacted site performance, such as an issue at your payment-processing provider. Your job is to try to reverse it if within your control, or address it another way if not.

Assume the situation is self-inflicted until definitely proven otherwise.

I can’t tell you how many times a sudden dip in growth was caused by something fully in control of the business. Frankly, this is the best-case scenario, since you have full control over things in-house. In my previous companies, it was typically caused by our breaking or degrading key functionality on our own website when rolling out new code. Good news, as that’s more straightforward to fix quickly. (But I’d always expect a post-mortem from my team the next day to assess how we missed the issue.)​ If you know the cause, don’t let it happen again.

Make the search for these underlying causes systematic.

At eBay, we used to call this process “unpeeling the onion.” Start at the highest levels to find evidence about what changed, and then use those clues to work your way down into the details. When did the issue emerge? Did it emerge suddenly or more gradually? Was it experienced in all geographies? Was it experienced across all platforms (website, native apps)? Was it experienced in both organic and partner traffic, was it experienced consistently across partners? Was it a top-of-the-funnel issue, or an issue with conversion in the funnel? What did competitors do, and when did they do it? Is your customer-support organization hearing anything out of the ordinary from customers?

Divide and conquer so that you can get to answers as fast as possible.

Once you’ve set up a war room, do meetings every couple of hours. Start by having the core group brainstorm potential causes, then delegate across team members to investigate potential culprits. Reconvene the group periodically to share what they have learned and reprioritize the next areas to investigate as you unpeel the onion. Do this until you are able to find the root cause.

Consider a Plan B if necessary.

If the startup gods are smiling, you will be able to figure out the cause of the issue and correct it. But if the startup gods aren’t smiling, and you can’t either figure out the cause and/or figure out how to correct it, it’s time to start working on a Plan B for the business. Plan B often includes kicking off a strategic process that ends up in the sale of the company before it becomes as obvious to others as it is to you that you’ve got a dying shark on your hands. This is what happened at Reel.com when it quickly became obvious that Amazon’s launch of the DVD/video category in November 1999 represented an existential threat. (Unfortunately, the shark died before the business could be sold; Reel.com shuttered operations in June 2000.)

A closely related situation to rapidly slowing growth is when a competitor starts to outpace your business. In most consumer segments, one company tends to emerge over time as the winner. That could be because they establish a network effect, or they have access to better employees and more capital at better terms. But falling into the rearview mirror of your key competitor can potentially be a strong existential threat to your company, so it also merits running the “Oh, shit!” game plan. This doesn’t mean you can’t still build a great business. But if you’re going for winner-takes-all, you might consider changing your business altogether as another Plan B.

The best CEOs take immediate ownership over the situation and leap into action. They don’t make excuses, and they don’t accept excuses. They assume that growth is under the control of the company — even if it isn’t — and they work like crazy to get it back. Companies run by CEOs with these instincts are highly correlated with the most successful companies.

So, when growth slows or stops, feel free to freak out. In fact, feel free to freak out even sooner than that. You’re going to want to anyway, and you certainly have my encouragement! Hopefully, you now also have a plan.

Jeff Jordan is a partner at Andreessen Horowitz and serves on the board of Airbnb, Belly, Tilt, Instacart, Lookout, Pinterest and Walker & Company, among others. Prior to a16z, Jeff was president and CEO of OpenTable, which he took public in 2009. Before OpenTable, he was president of PayPal, and he was previously SVP and general manager of eBay North America. Follow him on his blog and @jeff_jordan.