OpenTable's CEO on How to Dodge Company Breakdown as You Scale
Management

OpenTable's CEO on How to Dodge Company Breakdown as You Scale

OpenTable's leader Christa Quarles explains how startups can spot and survive their breaking points.

Here’s the scenario: Management starts to lose control over a company’s work. Teams are operating without full context or feedback loops. Product becomes highly reactive. HR’s busy and top performers are looking to jump ship. Those who stay start to complain about the lack of direction.  Nearly all technology companies — from small startups to big successes — have experienced at least one of these symptoms.

Take it from OpenTable CEO Christa Quarles. Over her career, she’s become virtually clairvoyant when it comes to identifying when, how and where companies are about to break — and early enough to save the day. Without question, Quarles comes from resilient stock. She grew up with six siblings, her husband’s a former Navy Seal, and four months after joining OpenTable as its CFO, she was named interim CEO and elevated to permanent CEO two months later.

In this interview, Quarles draws from a rich career across industries to share her war stories and hard-earned wisdom about how startups can survive their breaking points. She offers three strategies to fortify companies for change, and charts four key areas to monitor for fissures during growth: headcount, revenue, users and geographic expansion. With her predictive acumen and figurative duct tape at the ready, Quarles may be your company’s Nostradamus meets MacGyver.

Three Principles To Reinforce Your Company

From the vantage point of the C-suite across four companies in six years, Quarles has become an expert at picking out the sweet and weak spots in companies as they scale. Through recognizing those patterns, she has honed her approach to leading a company during dynamic and uncertain times. Here are the three pillars of her philosophy:

Reactively lean-in.

Most startups either proactively establish process (like standing meetings, systems, protocols) to support fast growth or push until something breaks and then assess if it’s worth fixing. Instead of swinging to either end of the spectrum, Quarles champions reactively leaning-in, which requires continuous calibration.

“If you try to put in a lot of process before any pain is felt, there won’t be the urgency to do anything. And you’ll drive away your engineers, who especially hate banal processes,” says Quarles. “So you have to watch for the little cracks and nudges before the big hammer hits you over the head. Post mortem why things happen. Don’t write off edge cases as one-time exceptions. They are there to teach you and should make your antennae go up.”

To strike the balance, seek better context on the current situation, not preemptive ways to change it. The former is a contribution, while the latter is inevitably about control. “At Playdom, we treated our LTV [life-time value] calculations like they were gospel. Driving new users through advertising was a big part of our business and we were spending a lot of money doing it,” says Quarles. “At times, I had my doubts. As CFO, I had the benefit of hindsight on the ad spend as I could see the near-term ROI calculations and look back on the financials for the past quarter. If it didn’t add up, I could push on the teams through simple, side calculations that uncovered issues with how we calculated the viral coefficient for new users. That way we were quick to make refinements, without blindly pushing wholesale changes that’d shock the system.”

Work toward — and be affirmed by — problems.

The success and survival statistics of startups are referenced often, but, according to Quarles, more attention should be paid to how often companies temporarily break, not completely fail. When a company — or anything — breaks, there’s an interplay of counteractive forces, and related movement, tension and traction. There’s an opportunity to re-channel or redirect energy. Failures become one-directional and are characterized by stagnancy more than dynamism.

To grow, you must break. Like a snake shedding its skin.

“Get comfortable with breakages —  it’s part of the process. They’re footholds, not pitfalls,” says Quarles. “Build systems around them, so that you’re prepared to identify their origin, learn from and repair them. Then, the breakages won’t seem so scary or dramatic. They’re signals that you’re ready to take on something that, up until this moment, you weren’t able to handle.”

Take when to introduce basic HR (T&E policies, employee handbook and systems) at a startup. “Early on at Playdom, we were keeping all of our employee information in spreadsheets. We didn't have an HRIS [Human Resources Information System]. We didn't have policies,” says Quarles. “But as we grew to 175 to 200 people, this process started to become unwieldy. We didn't have salary histories. We started seeing T&E costs grow faster than it should given the employee growth. Incorrect information flowed into payroll. All of these things are symptoms of your system are breaking down. It’s means growth, but also the need for a new approach.”

Establish a clear line between the finder and fixer.

In Quarles’ experience, it is rarely the person who spots the breakage who fixes it. “Often it is the folks who are in jobs with a horizontal purview — such as finance, HR or the CEO — who can and should spot fissures,” she says. “A CEO may have to dive in (depending on the size of the company) and outline the problem for teams, but ultimately the groups where the breakage is happening should be the ones to fix it. Chances are teams won’t get it right the first time, so watch and iterate. Post-mortem the tiny breaks and try a little process to offset the problem.”

Quarles illustrates this point with her account of Playdom’s users and their LTV. She was leading Finance, but was able to spot a material issue with how the startup spent its marketing dollars. Companies can even take steps to narrow the gap between “the finder” and the “the fixer” of a breakage. “At Nextdoor, our BI [Business Intelligence] team would find anomalies in the data, which could mean that the damage was already done inside the product,” says  Quarles. “As a result, we set up dashboards for the relevant product teams so they could be on the front lines to see things break. By democratizing the data and sharing the analysis, the these teams could catch their errors as they rolled out new features, instead of waiting for them to show up in the top line metrics.”

Know Your Company’s Fault Lines

For startups equipped with Quarles’ three principles around company breakages, here are the four areas to keep a close eye on for cracks. Because no two companies experience growth the same way, Quarles has discovered that asking the right questions periodically is the tactic, rather than prescribing generalized solutions. So, interlaced with examples and tips, she highlights key questions that leaders — and their teams — should ask themselves to stress test their organizations.

HEADCOUNT

While at Playdom, an online social network game developer, Quarles saw the company through a $760 million acquisition by Disney, but not before she helped the startup buy and integrate seven companies — at an approximate clip of one acquisition per month.

One Monday morning, Quarles took stock. It was the first day for 30 new hires. And over the preceding six months, the startup had onboarded 500 new people. This rapid growth splintered their HR processes, finance systems and engineering org chart. Quarles has learned to ask the following three questions to identify any chinks in the armor due to ballooning employee growth.

What percentage of first-time managers are in action?

As companies in hyper-growth add headcount, it’s common to elevate individual contributors to leadership positions. While internal promotions come with cultural fit and invaluable institutional knowledge, they may not yet be equipped to see around corners as managers.

“I can’t tell you how often I’ve seen new hires pour into a company and an overwhelmed leadership internally promotes by pedigree, elevating the Stanford grad with a PhD in CS over onboarding a seasoned manager,” says Quarles. “Now imagine that happening weekly for a year. It’s a perfect storm: a promoted, star individual contributor who’s a rookie in the role and the leader who’s forgotten the difficulties of becoming a decent manager at the start.”

A concrete way to gauge potential breaking points as a team doubles, triples and quintuples is to monitor the number of active first-time managers (who are less than a year into the role). If that number is roughly 30% of total management, take pause and ask the following questions to determine if you’ve trained your contingent of green leaders sufficiently:

  • Is it easy to clearly outline how we delegate and escalate key decision making?
  • Are managers having 1:1s with their people?
  • Are they giving feedback back to their reports on their progress?
  • Are new managers’ reports happy in their work?
  • Do new managers have people they can turn to as a lifeline?

If you don’t like the answers, assign people and activate processes to create checks and balances, before a big break happens. “For example, we once made a rule that disallowed adding more than 3 reports to a new manager without adding support. It could be formal management training or pairing the new manager with a seasoned counterpart. They’re used to being top contributors; they need a safe forum or confidant to ask the ‘dumb’ questions.”

How quickly do managers get back to you with status updates?

Another early warning sign is the inability of managers to quickly summarize the status of their work. “If there are blank stares, it doesn’t necessarily reflect the capability of the manager, but perhaps the complexity and velocity of the challenges they’re facing,” she says. “The solution is not to make them do more busywork reporting, but to investigate why they’re having a hard time pinpointing their location on the map.”

Throughout her career, Quarles has found that when her reports don’t have answers to simple questions, it wouldn’t be long until the wheels fell off the wagon. In one instance, she noticed that features of the company’s website stopped functioning. The short term solution was fixing the code, but the underlying trouble was in the Q&A process. She knew that because her engineering managers couldn’t answer where the release was in the approval process.

“If managers cannot tell you in an instant where they stand on a goal, it’s best to re-visit and retrofit your team’s objectives. Perhaps they’ve outpaced them or haven’t set them in the first place,” says Quarles. “Regardless of the reason, if it takes them a week to get back to you on a goal, then they probably aren’t monitoring what that goal is. They're not necessarily in charge of what their group is actually working on in that point in time. That’s the time to step in.”

How monocultural is human resources?

When Disney acquired Playdom, Quarles knew to zero-in on one specific department: human resources. Attention is more often given to how engineering or product management teams integrate. Yet, during hyper-growth and milestone mergers, Quarles likes to gauge the integrity of the team that’s logistically defining how a company’s people are brought together.

“People like to talk about the fabric of a company as if it’s one material. But in moments of rapid growth — whether organic or via acquisition — it’s undoubtedly quilt work. That requires hands that assemble in unison,” says Quarles. “Imagine a 90 year-old media company under the same rules as a high-paced, frenetic gaming-as-a-service startup. The latter, Playdom, had almost no process, while the former, Disney, expertly coordinates 100,000 employees during peak amusement park season — in just one corner of its operations. More often than not, HR is a reliable and representative microcosm for companies’ cultures at any point in time.”

So, when the media behemoth and gaming startup came under the same roof, Quarles stepped in as translator, until they became comfortable and conversant with the other. “During change and fast growth, HR should be one of the first places to be unified by the same cultural mores and procedural philosophies. In this case, it meant adopting some structure from Disney and keeping some flexibility for Playdom, until they both saw their individual identities represented and then melded together. That unified front was critical to the health of a growing team.”

Frequently conduct litmus tests to spot the breaking points in your company, especially when it’s scaling quickly. Growth covers a lot of sins.

Acquisitions aside, rapidly growing companies can lean on a few techniques to keep its HR strong and singular. Quarles has worked at startups that scaled to over a hundred people before instituting “real HR” — that’s just too late. Here are a few places to monitor during fast growth:

  • Compensation: “If 30-40% of your employee base is over two years vested, it's probably time to look at stock grants.”
  • Annual reviews: “Do you have them? You can keep it informal and simple, but not giving star performers feedback on their performance isn't a good idea. People want to know they're having impact. If you don’t have a formal performance review cycle, lean on or bring in some seasoned managers. It’s harder for first-time managers to navigate.”
  • Unfiltered feedback: “An exec at a fast-scaling startup will stop getting the straight scoop from her team. Much will be sugar-coated. More people means more politics. Try to keep your truth-tellers. If you can’t, create mechanisms for anonymous feedback.”

REVENUE

Having held the post of CFO at several companies, Quarles typically has the strongest command of a company’s financial health and potential in the boardroom. Compared to increased hiring or skyrocketing user growth, breakages along the financial vector often occur due to the overwhelming absence of revenue. Here are the two key questions she keeps in mind regarding revenue to weatherproof for breakages:

What’s different about how the company’s investing in its growth versus in the past?

When Quarles started at OpenTable, the company had been acquired for about a year, public for six years and in business for 17 years. Long past were the days when it brought restaurants online by installing cable to show the value of its web-based reservation service. Today, OpenTable doesn’t need to invest in connectivity; the catalyst for growth has changed.

“The path of how you once grew will be fundamentally different from the way you now need to scale. That’s especially true for startups who should be testing and confirming their path to growth about every two years. Don’t get your horizon stuck at the next quarter, but think deeply about where your think your current growth strategy will land you,” says Quarles. “Right now, for OpenTable our growth comes from market adjacencies like casual dining and entering new geographic markets — both organically and through acquisition as we did in the UK and, most recently, in Australia. Our benchmark has largely been the US, and within that, specific markets.”

Don’t be afraid to look cross-industry for inspiration on how companies have sparked growth and avoided top-line atrophying. For example, one way Uber chose its target cities for expansion was by monitoring where its customers opened its app in regions where it didn’t operate. “We’ve employed the same tactic by analyzing requests from OpenTable users and have started to evaluate their top searches on the platform in regions where we don’t operate. If they’ve booked before, we know that they’re familiar with the type of restaurants on the site. When we noticed that they were searching for casual, chain restaurants versus one-location eateries, we knew to take a closer look and change tacks to fulfill that need.”

Learn to call audibles. An above-average read of your environment is better than sticking to a perfectly drawn-up play — even if that plan once brought success.

Are there limitations to the network effect?

Around 12 years into OpenTable’s operation, network revenue surpassed subscription revenue for the first time. This was a milestone for a business that, for many years, was almost entirely subscription-based because there was little to no network of restaurants. Since 2010, OpenTable’s revenues have been increasingly — and very heavily — composed of network revenue. “The trade-off between subscription which is fixed and network revenue which is variable has made for a very dynamic evolution,” says Quarles.

As the network became more powerful, OpenTable needed to take a more nuanced approach with its pricing strategy. “We weren’t matching price and value or giving restaurants a way to spend more to get more. We needed to take the simplicity of $1 per diner, which served the company well in its past and look at a different solution for different states of equilibrium in our marketplace,” says Quarles. “Some restaurants have too many diners and too much demand. As such, we’re testing paid reservations in New York with the very popular restaurant Cosme. On the other hand, some restaurants have too few diners. For those, we are focused on how OpenTable can the marketing mechanism and demand-generating engine for these restaurants.”

OpenTable CEO Christa Quarles

GEOGRAPHIC EXPANSION

Given its broad network of restaurants, OpenTable is a geographically-defined business. More restaurants mean more revenue, but also more diversity in customer behavior. For example, what works for patrons in Nashville doesn’t work for diners in New York. If OpenTable expands with its product in a regionally agnostic way, fissures will form and start to weaken the business.

“Our research showed that our Nashville diners were eating out twice as often as they reserved on OpenTable. So, if they dined ten times a year at restaurants after making OpenTable reservations, they were actually eating out around 20 times annually,” says Quarles. “Through talking with customers and looking at restaurant inventory, we realized that we had an issue with our supply. The restaurants that we offered were the ones booked for special occasions, like an anniversary or birthday.  We started to add more casual dining restaurants that are primarily walk-in, but accept reservations.”

The bottom line: repurpose your best use case as an initial comparison, not a template. For OpenTable, that means juxtaposing its American cities with its first municipal market, San Francisco. Or, if taking a broader regional analysis, by comparing the US market with the European market. “The exercise is to compare, not replicate,” says Quarles. “The starting point, development and solution for that market will not be the same. But it will accelerate the path to unveil that market’s levers, economics and cultural peculiarities."

Here are the follow-up questions to ask around geographic expansion to prevent any breakages across time zones:

  • How exportable is the product? Have you built your product to be ready made for new languages? “OpenTable has learned this lesson the hard way. It didn't originally build its product to be global; multi-lingual capabilities weren’t a part of the initial product roadmap. It’s difficult to adapt our architecture for other markets, but we are doing it,” says Quarles. “If you're doing construction and the walls are all opened up, it's much easier to put the outlet in the wall. You don't have to plug in the lamp just yet, but have the outlet there when the house is being built. Same goes for international. A little forethought into your product architecture can save pain down the road.”
  • Can you configure your product easily for cultural differences? “In America, we list breakfast, lunch and dinner, but in Australia, people reserve for breakfast, lunch, teatime and dinner,” says Quarles. “All of a sudden, there’s an entirely new category of dining experience. On the back-end, it was an easy fix, but was important to provide restaurants with this capability as it matched their hospitality style.” Also, OpenTable has discovered a spike in reservations from restaurants offering deals in both the UK, so it has started to pilot specific offer-based products in select markets that it plans to take globally.
  • Does your business require payments, what is different about how you will make revenue as a result? “Germany's the classic example. They don't use credit cards and so, show different behavior in how they would like to pay,” Quarles says. “So, it was important for us to consider that preference, before setting up an identification and reservations system centered around credit cards.”
The buildings that stay up during earthquakes are designed to move with the vibrations, not outfitted to resist them. Learn to tango with the tremors.

USERS

When the number of active users goes through the roof, it’s cause for celebration only if the growth can be corralled. Quarles has been inured to breakage by user growth due to her background in the gaming industry. “Most fissures form from technological scale. We would see significant spikes around product launches and could see step-function breakages at users thresholds that hadn’t been previously been hit,” she says. “We would ensure we had surge capacity on AWS [Amazon Web Services]. But that’s a short term fix. And it’s also dangerous to buy the 747 when all you need is a nimble biplane. You can ‘over invest’ without knowing exactly what you need causing waste as well. It’s a tricky balance.”

The best policy is to take small steps to build the system to scale. Quarles starts by asking these questions:

What are the edge cases with users and how can we get to them faster?

Quarles admits that it’s difficult to extract general tactics around users, given how distinct and disparate their behavior can be between — or even within — industries. “My friends at [mobile gaming and entertainment network] Scopely recently showed me a game that got 4 million downloads in a week. It reminded me how dynamic user growth can be in the gaming space,” says Quarles. “ At Playdom, we’d monitor Daily Active Users (DAUs). We had a couple of games that'd hit one million DAUs. Then five million DAUs. And boom. Every time, we’d hit 5 million more DAUs something would break.”

There’s internal (e.g. sandbox-testing) and external (beta testing) exercises to run, but there’s not always a way to replicate five, ten or 20 millions active users on your app or game at once. “The real time nature of gaming makes it more complicated when more people are added to the system. Even e-commerce flows are much more straightforward,” says Quarles. “Once, one of our engineers coded the leaderboard so that it was updated in real time. But when 10 million active users were on, it crashed the entire game.”

While no amount of testing helped Playdom pinpoint the breakage at 5 or 10 million users, it helped them write and fine-tune a game plan for further breakages at scale after the fact. “Yes, success meant millions of active users, but it also was finding the tipping-point that brought it all down. If you start with the premise that it will inevitably break at some point, you learn to quickly speed toward the breaking points, so you can be on hand and prepared to better fix them.”

Skilled drivers know that the most controlled way to take a curve at high speed is to accelerate through it. And so it goes with testing for breakages during rapid user growth.

Are there dedicated eyes on analytics around launches and releases?

Given that no amount of testing will prevent breakages from user growth, Quarles champions continuous monitoring of user activity. Depending on a company’s resources, the responsibility may be concentrated in a role or shared across the broader team.

“At Playdom, I felt that we had a good system in place. We used home-grown analytics, which everybody had access to at their desk,” says Quarles. “It was mostly in real-time and so any team member could send out an alert if output mechanisms started to look funny. It helped us to identify and fix any issues — typically within an hour. That made our reactions feel more proactive. It beat apologizing to users saying, ‘Oops, it broke.’”

For key launches, it’s wise to build an analytics bunker and staff it with a team to monitor activity before, during and after a big release. “For notable launches, we’d have engineers on hand, each assigned to monitor a metric. It was a bit of a war-room mentality, but very rarely did things break — and not for long, if they did — when we organized ourselves in that way,” says Quarles. “We take the same approach at OpenTable. Inevitably, every Mother's Day the system gets taxed and something breaks. This year, we had all teams on watch and nothing got too overheated. The day passed without incident and set a new record for the holiday.”

The Takeaway

When Quarles asks a company when it last broke, she’s not digging for vulnerability, but searching for accounts of strength. She gets concerned when she sees a company that has been around for decades that hasn’t experienced a breakage due to hiring sprees, dried up growth strategies, market expansion or extreme user gain or churn. There’s no foolproof plan to keep fissures from eventually shattering your company, but you can almost guarantee extinction if aren’t preparing for — and getting affirmation from — a breakage in your company.

“It can literally be your company’s big break. And importantly it’s a chance to entrust and embolden your people,” says Quarles.  “They shouldn’t see a breakage as a failure, but as an eventuality during which everyone steps up as custodians of their corner of the company. As a leader, it’s your job to ask these questions before a breakage, so that your people are not only prepared for it, but also looking forward to the new growth it’ll bring.”