Partnering with Goliath: A Tale of Two Announcements
By John O’Farrell, contributor
SUNNYVALE, Calif., June 2, 2003
HP and Opsware Inc. Join Forces to Deliver Enhanced Automation for HP’s Utility Data Center
SUNNYVALE, Calif., Feb. 13, 2006
Opsware Announces Worldwide Distribution Agreement with Cisco
These two headlines sound pretty similar—“Small company partners with giant company to reach a bigger market”—but they led to two very different outcomes. Our 2003 deal with HP HPQ didn’t generate a single dollar in revenue, whereas our 2006 agreement with Cisco CSCO drove tens of millions of dollars in sales and helped to make Opsware the uncatchable leader in data center software. Why did one succeed spectacularly while the other never took off?
As a startup with the best product, your challenge is often getting it in front of enough customers and getting them to buy. In theory, striking a deal to have an HP or an EMC EMC or a Vodafone VOD sell your product to their customers is the way to cover the market and exponentially increase sales velocity. In practice, however, most “David-Goliath” distribution deals turn out like our 2003 HP deal: great PR, but not much else. Here’s the way it typically plays out:
- The deal is announced with great fanfare and high internal and external expectations.
- Goliath needs product changes, training and lots of help to even attempt to sell your product.
- Goliath deluges your already overloaded people with feature and support requests.
- You can’t justify assigning dedicated people to support Goliath because you can’t bank on any new revenue.
- Goliath sells nothing, or even worse, ends up competing against your sales team for customers you would have won directly at a higher margin.
- The “partnership” quietly withers away, leaving a damaged relationship with Goliath and a bad taste inside the company.
The temptation for some startups is try to “make it up in volume”—sign as many distribution partners as possible on the basis that none of them is likely to deliver much on their own. While that degree of over-coverage might feel temporarily reassuring, it only multiplies the challenges exponentially in practice. The only thing worse than having no partners is trying to manage multiple ineffective partners competing with each other in the market and drowning the company with their demands.
In my experience, one well-constructed, high-impact partnership is better than a hundred run-of-the-mill arrangements. Distribution partnerships take an enormous effort to make work. It’s just not realistic to have many of them, which is why picking the right one and constructing it intelligently is fundamental. At Opsware, we only had one significant distribution relationship over eight years, but it had a massively positive impact on the company.
Time to find a partner
It was early 2005, and things were starting to go pretty well for three-year-old Opsware. We had just closed our $33 million acquisition of Rendition Networks, giving us a lightweight but powerful network automation software product that we christened Opsware NAS (Network Automation System). NAS would allow us to penetrate and rapidly deliver value to new accounts, then upsell them to our considerably heavier and more expensive Server Automation System. (Our average NAS deal size was $115k, took a month to sell and a couple of weeks to deploy, whereas SAS deals averaged $765k and could take a year to sell and six to twelve months to deploy.)
It made sense to market and sell NAS as aggressively as possible—but we faced a challenge: Our direct sales force was small, only covered US and UK, and had few relationships with network buyers. We needed a big partner with the buyer relationships and global coverage we lacked.
The right partner
The right partner was obvious: Cisco. So was the product: NAS.
Thanks to our investment in strategic BD (see Part I and Part II of this series), we’d been calling on Cisco since the founding of Opsware, from the C-suite to mid-level data center managers. Now we finally had something to talk about. For all its pre-eminence in router and switch hardware, we knew from customers that Cisco’s management software capability was weak for its own devices and non-existent for, say, Juniper devices in the customer’s network. We had the market-leading, cross-platform product. Within days of closing the acquisition, we visited the Cisco SVP responsible for management software and offered him the solution his customers were demanding.
Before you can construct a high-impact deal, you have to be crystal-clear on what you want to achieve. We spent a substantial amount of time up front to achieve internal clarity and consensus across the company on what we wanted from a partnership. These were our main objectives:
- A real commitment by Cisco to market and sell a lot of NAS product, at an attractive margin to us.
- Opsware’s right to target every Cisco NAS customer for an Opsware SAS sale.
- Opsware branding on the Cisco NAS product.
- Minimal product changes and a single code base.
- Enough assurance of financial potential to allow us to invest in the support Cisco would need to succeed.
In a seemingly endless series of meetings in Cisco’s blue-gray conference rooms over the spring and summer, we gained a good understanding of Cisco’s main objectives. They wanted:
- The ability to sell NAS to any Cisco account, with active support, not competition, from Opsware.
- Cisco-branded product.
- Freedom to sell at any price they chose, including throwing in the NAS software free as part of a big equipment sale.
- “Insurance” against a surprise acquisition of Opsware by a competitor.
- Access to Opsware NAS source code, and even the ability to make derivative works. (As you might imagine, this one caused us a lot of heartburn.)
All through the fall and early winter, we shuttled back and forth between Tasman Drive and Opsware’s office on Mathilda Avenue, alternately negotiating a deal structure with the Cisco software team and updating and strategizing with our sales and product organizations to make sure we were staying true to our objectives. Some issues, like source code, were extremely complex. Nonetheless, like a figure emerging from the fog, a compelling deal gradually took shape.
On February 13, 2006, a year after closing the Rendition acquisition, we announced a worldwide distribution agreement with Cisco. Cisco’s enterprise sales force would sell our NAS product, rechristened Cisco Network Compliance Manager, to their worldwide customer base. Opsware’s sales force would follow their Cisco counterparts, supporting their efforts to sell the network product and in the process building the relationships and laying the groundwork for a subsequent million-dollar-plus server software sale.
Behind the headlines – a deal with teeth
While the press and public endorsement by Cisco were valuable, what made this partnership work unusually well in practice was a smart deal structure designed to achieve both sides’ objectives:
Minimum revenue commitment: It took us almost nine months, but we persuaded Cisco to make a binding three-year quarterly revenue guarantee that totaled almost as much as we had paid to acquire Rendition. We helped our Cisco counterparts to understand that this would be a win-win: Guaranteeing several million dollars a quarter to Opsware would give Cisco huge motivation to drive sales. In turn, it would give us the assurance we needed to back off our own NAS sales efforts and do everything—including assigning some dedicated sales and product headcount—to make our partner successful.
Substantial revenue share: We secured a revenue share and pricing floors that made a Cisco NAS sale almost as financially attractive to us as if we had sold it directly—because Cisco could actually charge more for the software than we could.
Co-branding: The product was marketed as Cisco Network Compliance Manager, “built on Opsware automation technology”.
Channel-neutral compensation: To motivate Opsware’s sales people to support their Cisco sales counterparts instead of competing with them, we compensated them for every Cisco sale in their territory. While this cost us real money, it made sense because every Cisco sale opened the door for us to sell a much bigger, higher-margin product.
Acquisition insurance: We committed to give Cisco advance notification of any agreement to sell the company, and the opportunity to enter the bidding if they wanted to. (Note this is very different from a Right of First Refusal.)
Source code license: We gave Cisco a three-year license to view our source code and build on it, but came up with an innovative licensing approach that protected us fully and even generated millions in additional revenue. In any case, we bet they would never actually do anything with the code, and we were right.
Within nine months, Cisco was selling large network software deals into accounts like DHL, Costco, Sprint, USPS and Telstra, and a short time later, they were selling enough to exceed the quarterly minimum revenue guarantee to Opsware. The Cisco deal would go on to generate over 25% of our bookings. Cisco’s massive distribution power vaporized our NAS competitors. Just as important, every NAS sale opened the door for our SAS product in accounts we had never penetrated before, providing a major thrust to our high-margin server business.
Of course, there were many challenges along the way. Cisco people needed lots of training and support to sell the product. There were occasional conflicts over accounts that had to be arbitrated. We ran a separate weekly pipeline call with Cisco to track and drive their sales and direct our follow-on server sales calls. Product change and bug fix requests from Cisco had to be triaged and folded into our roadmap.
However, we were able to handle these demands with equanimity, because we had a couple of dedicated heads in sales and product management to handle them, and a large guaranteed revenue stream that helped the whole company to realize this was a deal worth supporting.
Like Opsware’s 2003 HP deal, the vast majority of David-Goliath type distribution deals don’t deliver beyond PR.
On the other hand, like Opsware’s Cisco deal, the right distribution deal with the right partner can be truly transformative. In formulating your partner strategy, consider four golden rules:
- 1. Quality trumps quantity: One well-constructed partnership will have far more impact, and be far more supportable, than lots of toothless arrangements.
- 2. Mission clarity: Invest the time up front to get crystal-clear on your objectives and those of your partner.
- 3. Demand hard commitments: Big companies are easily distracted. Binding commitments focus the mind and create incentive to deliver long after the deal is signed. (Note this means finding a senior executive sponsor with the authority to commit.)
- 4. Make hard commitments in return: You will need to invest substantially in making your partner successful. Commitments like sales and product support and channel-neutral compensation are worth making in return for the right commitment from your partner.
Only a few startups invest in the ongoing BD effort to systematically cover the landscape of potential partners and create that game-changing partnership when the opportunity presents itself. For Opsware, that investment paid off three years after our start as a software company. Like the Loudcloud-EDS deal and the M&A campaign that followed, the Cisco partnership provides another example of the power of strategic business development.
John O’Farrell is a general partner at Andreessen Horowitz. John led the firm’s investments in Facebook, Twitter and Groupon, and works with portfolio companies on partnering, strategic transactions and global expansion. Prior to Andreessen Horowitz, John held Executive Vice President positions at Silver Spring Networks, Opsware and Excite@Home. Earlier in his career, John worked in the United States and Europe with US WEST, Booz Allen, Telecom Ireland, the EU Commission, Digital Equipment and Siemens. John has an MBA from the Stanford Graduate School of Business and a Bachelor of Electronic Engineering from University College Dublin. He speaks English, German, French and Portuguese. John blogs at http://john.a16z.com/.