Introduction
“The most important distinction between constructing a highly-technical enterprise infrastructure product and a extra conventional SaaS product is that you simply’re not simply designing for human customers, you’re additionally considering machine customers,” says Nate Stewart, Chief Product Officer of Cockroach Labs.
The corporate is the creator of CockroachDB, a cloud-native distributed SQL database that’s designed to construct, scale and handle data-intensive functions. (Because the founders put it, they named their startup after the resilient insect as a result of it’s practically unimaginable to take down — which is kind of the tall order for Stewart to reside as much as as a product chief.)
No matter who you’re designing for, with the top of the 12 months quick approaching, annual planning is probably going prime of thoughts — analyzing whether or not you hit your marks within the earlier plan and what would possibly information your work for the next 12 months. However the best-laid plans can simply stall when your technique makes its approach off the web page and the rubber meets the highway.
“In a market as giant as ours, there are totally different doable methods that may be profitable. After we determined to focus our product technique on mission-critical, transactional workloads that concentrate on holding information near customers, different possible methods opened up for our opponents,” says Stewart. “You begin to see prospects whose characteristic requests you stated no to enter the arms of your opponents — and that may be demoralizing when you don’t have conviction in your path and its long-term success.”
There’s rather a lot written about startup technique and the way you should utilize OKRs to go from technique to execution. However there’s not a variety of literature on the market about why some methods fail in the true world.
As product people begin sharpening up their annual plans, Stewart unpacks his playbook for gaining conviction and making a resilient technique that’s nonetheless standing this time subsequent 12 months. On this unique interview, he shares the failure modes for product leaders to maintain an eye fixed out for (like specializing in buyer high quality over amount), and the way he’s sidestepped these errors in his function at Cockroach (together with a brand new method to design companions).
Let’s dive in.
When a startup is simply starting to tread its path, it may well appear to be there are seemingly limitless routes on the map to product/market match. Cockroach confronted an analogous level of competition within the early days when deciding what to construct. “Databases work in each business and in many alternative use instances. As the primary Head of Product, I needed to put in place the processes for making choices round what to construct. What are the capabilities that we must always say sure to, and what are the options that we must always keep away from?” he says.
Even at first, you want a way of claiming no and filtering out the entire requests and potential use instances on your product.
So Stewart started wanting on the other ways of segmenting the database market, eyeing the sting instances the place Cockroach would possibly slot in most naturally. This led the crew to a fork within the highway, an early alternative that will go on to be one of many defining choices for the corporate.
“We might construct a database that was an analytics system like Snowflake, that’s glorious at operating giant aggregations and doing computations throughout many rows of knowledge. Or we might construct a database that was very performant when it got here to writing and studying transactional information,” says Stewart. “Probably the most pure early choice level was: What sorts of queries can we need to make Cockroach most performant for?” he says.
It’s a query that many product leaders have confronted within the early days: The place must you focus the ICP? “In the end, the massive choice that we made was to concentrate on transactional use instances — as a result of the factor that makes CockroachDB distinctive is the product’s resilience. In case your analytics database goes down for 2 seconds, that’s okay. You simply hit refresh and run the analytics once more. But when the database that’s powering your fee processing goes down for 2 seconds, you possibly can lose tens of millions of {dollars}. Our secret sauce is holding your information on-line, up and operating, and near customers,” he says.
This drew the road within the sand that Stewart (and all different product leaders) are searching for as they start to sketch out the product roadmap. “Getting alignment with the founders and the income crew that we have been going to be a transactional database helped us say no to a variety of issues that will push us nearer to opponents like Snowflake, and enabled us to say sure to issues that will construct upon our product’s power over an extended time frame,” he says.
However, as Stewart places it, this was a controversial alternative amongst the crew and the retelling of this story smoothes over among the sharper edges. “There have been some real-world pressures that made it tough to carry the road on this choice to turn into a transactional database. On the time we have been pre-revenue and had some giant prospects saying, ‘We’ll offer you your first $100,000 or $200,000 — however provided that you construct out these analytical capabilities. It’s extremely arduous to say no to a startup’s first potential income sources,” he says.
To shore up conviction within the option to concentrate on transactional databases, Stewart shaped shut design partnerships with potential prospects. However right here, he notes a key distinction with regards to constructing enterprise infrastructure versus other forms of SaaS tooling. “Enterprise infrastructure consumers are usually extra conservative — by definition, the infrastructure is what helps some core enterprise course of. In case your infrastructure is unstable or goes down, your core enterprise can go down,” he says.
All informed, people weren’t precisely lining up the door to be early testers. “It wasn’t as straightforward as I anticipated to get individuals to give suggestions on the beta software program — as a result of there was such an enormous engineering funding required for a brand new characteristic that wasn’t absolutely baked. It wasn’t actually price it to our prospects to check originally — they only needed the product to work,” says Stewart.
And whereas most people need the most recent tech— the most recent iPhone launch, or the brand-new automotive mannequin off the lot — with enterprise infrastructure, consumers are inclined to take the alternative method. “When our prospects have been buying the software program, they didn’t need the most recent and biggest. They needed it to be six months or perhaps a 12 months behind,” he says.
So an already lengthy product suggestions loop was stretched even additional. “Numerous the suggestions that we acquired from our prospects was primarily based on choices that we made a 12 months in the past — which implies that you’re driving by means of the rearview mirror,” says Stewart.
So the Cockroach crew leaned closely on dogfooding the product within the early days, even when it wasn’t fairly so elegant. “There was a degree the place I insisted that the entire product analytics have been carried out on the CockroachDB database. CockroachDB shouldn’t be an analytical dashboard, nevertheless it was a approach for us to see what the SQL interface was like and the way straightforward it was to get information out and in of CockroachDB,” says Stewart. “Initially a few of these dogfooding alternatives have been just a little contrived, and admittedly fairly painful, nevertheless it was essential for us to check the perimeters of the product and paid enormous dividends afterward.”
However the transition from dogfooding along with your inner customers to placing your product out in the true world could be a sizable leap. That’s the place design companions will help cushion the autumn. However reasonably than method design partnerships in a conventional approach, Stewart tried a special method. His recommendation to different product leaders dealing with comparable headwinds is to bucket your design companions into totally different classes as a way to de-risk particular features of the enterprise.
Strategic Design Companions: “With these people, we actually talked about the place we thought the database market was going. What have been among the macro issues that individuals have been experiencing and the pains that their group was going by means of?”
Roadmap Design Companions: “These have been individuals offering suggestions on the precise roadmap and the capabilities that we need to add to the database. For instance, it was clear over time that if persons are putting in a mission-critical database, they need it to plug into the remainder of the group and ideally ship updates to different techniques in real-time. So we talked about including a real-time information streaming functionality to the roadmap — there was a variety of pleasure round that concept and these design companions helped us de-risk the aptitude and the engineering funding it might take to construct.”
Characteristic design companions: “That is probably the most tough design partnership, which is getting suggestions on specific options and getting engineers to attempt to use them. These are the design companions who really feel probably the most ache that your product can remedy. For instance, we had an early buyer that was doing a big quantity of fee processing, and all of that processing was taking place in a single information heart. If that information heart went down for any motive, whether or not there was an operator error or one thing out of their management like an earthquake or an influence surge, their enterprise could be fully offline. They needed to get outdoors of a single area and construct a extra resilient system. They have been prepared to speculate a few engineers to be sure that what Cockroach constructed would assist them on that journey to a cloud-native structure.”
MBA-style stories about market measurement are nice. However whenever you begin to see the conviction out of your design companions, you get an image not simply academically of how huge a market is — you begin to really feel it viscerally. It’s just a little artwork and science.
Significantly within the early days, when the product roadmap is a bit squishier and prospects are fewer and much between, it may be tempting to decide to new options, within the title of signing a contract. However from the start, Cockroach has made it extremely tough to commit options to prospects. “When new product managers and new sellers be part of our crew, they’re typically caught off guard by our unwillingness to decide to constructing one thing for a selected buyer. It’s like getting a invoice by means of Congress,” says Stewart.
That is fairly uncommon within the early startup section — when plans are amorphous, commits are far more frequent. “You possibly can say, ‘Hey, we don’t have product/market match but. So if a buyer pays us income, we are able to decide to constructing options A, B and C.’ If you will get your first 500K buyer by means of buyer commits, that may materially change your run fee. Nevertheless it additionally considerably modifications your agility. The chance price is large,” he says.
Buyer commits are tech debt for product groups. You restrict your flexibility — and what’s extra beneficial to a startup than the power to alter course rapidly?
He estimates that in his 5 years at Cockroach, they’ve made 5 or 6 buyer commitments whole. “That flexibility has been enormous as we’ve navigated the altering database panorama,” says Stewart.
Right here’s how the Cockroach crew decides which buyer commits make the minimize:
“The primary is that it must be one thing that we have been planning to construct anyhow, or is someway higher for the corporate than what we have been in any other case planning on doing. The shopper commits are roadmap accelerators, not detractors.”
“We now have to grasp the answer from a technical standpoint — do we have now the engineering crew and the important thing companions essential to just accept the dedication?”
“There must be an enormous reward on the different finish. We now have to know that if we construct this characteristic, the prospect will turn into not only a buyer, however somebody who will deploy and develop their utilization with CockroachDB immensely.”
Stewart additionally flags that if consumer requests are arising again and again, you’ve acquired to begin digging deeper with the go-to-market movement.
The request for buyer commits is an attention-grabbing indicator that you could be not have product/market match or you may have a misaligned GTM.
“If the gross sales crew is feeling like they must stretch, it is a probability to have a look at all the GTM operate. Do they not know place the product? Or are they not seeing the sorts of prospects within the pipeline that they anticipated?” he says. “You additionally must dig into the gross sales incentives, which are usually extra short-term targeted. What does this deal imply for somebody’s quarter or somebody’s 12 months? They might be making a dedication that they don’t must cope with the longer-term impacts. You must be conscious of that, particularly whenever you’re within the enterprise house with a bigger gross sales crew.”
One of many largest tripwires that Stewart sees, notably for resource-and-manpower-strapped startups, is that folk dream huge with regards to the technique — with out considering the true capabilities of the org. “You might have a few choices right here: you possibly can have an sincere accounting of your crew’s capabilities, given your technique, and work out how one can shut the hole between the 2. Or you possibly can resolve that that is the crew you may have, and work out a technique that this present crew can execute,” he says.
There have been a number of inflection factors in Cockroach’s journey the place Stewart has confronted this actual predicament. “This was a problem we skilled instantly as we began shifting into extra of a SaaS product by delivering CockroachDB as a service, the place our prospects don’t run the database themselves — we run the database for them. There was a macro development pushing extra individuals to dump their operations to distributors, and it had turn into more and more clear that we would have liked to supply CockroachDB as a service,” says Stewart.
So his subsequent query is one thing he suggests that every one product people ask themselves repeatedly when crafting any type of technique:
What must be true for this technique to achieve success?
“The very first thing that we realized is that our prospects have been having hassle hiring these database operators — however now we must discover a solution to rent them. And these website reliability engineers aren’t going to only run one database, they’re going to run hundreds of them. How do we discover these SREs? How can we interview them? How can we combine them into our tradition and make them profitable? If we resolve to pursue this technique, it’s an enormous functionality that we have now to construct out,” he says.
Stewart sketches out one other instance of how your technique must leap off the web page. “We’re within the enterprise of dealing with mission-critical fee ledger information. If we’re asking individuals to belief us with that information, we have to construct a considerable safety skillset. So we have now to rent safety specialists on the product and engineering aspect to assist us pull that off. Should you’re not very particular about outlining all of the totally different contingencies along with your chosen technique, your technique will fail,” he says.
There’s an oft-repeated startup adage that competitors doesn’t matter — simply preserve your head down, and don’t get too distracted by what people round you might be doing. Stewart has been caught up on this mantra however has since modified his tune. “I initially thought that competitors didn’t matter. Simply concentrate on the shoppers, work by means of first rules, and also you’ll find yourself in a superb place. However with regards to touchdown an enterprise sale or going by means of a radical analysis course of, it’s inevitable that opponents are concerned. It’s helpful to grasp the place your opponents are sturdy and the place you’re sturdy, and speak about these differentiators in a approach that favors you,” he says.
So with regards to crafting firm technique, preserve your opponents in your peripheral imaginative and prescient. “We now have a category of opponents which have made essentially totally different strategic choices than us. For example, Amazon Aurora is a superb database. Fairly than make all the database cloud-native, they’ve made simply the bottom degree cloud-native. For them, that’s a strategic alternative as a result of it means it’s a a lot easier migration to take an present database and run it on Amazon,” says Stewart.
“The factor that’s in our favor is that these databases aren’t all cloud-native, so that you don’t get the identical resilience traits. You possibly can’t transfer your information anyplace on the earth, and you’ll’t scale to the identical degree as you possibly can with CockroachDB — that’s a basic trade-off,” he says. This key differentiator then arms the gross sales crew with the messaging they should place the product to win prospects.
When you consider the strategic investments which you can make, spend money on the core differentiators that set you other than opponents.
As Stewart places it, he leads his product groups like a product working system: “Product Managers are taking exterior inputs from the market — what prospects are saying and what opponents are doing. They’re additionally taking inner inputs, like suggestions from the gross sales crew and what pre-sales and post-sales people are speaking about. They’re distilling all of that info by means of the lens of a technique after which determining the ways in which we would change or improve our product to maneuver us nearer to our mission,” he says.
The questions that come up at this stage embody:
How deeply can we perceive the issue?
How can we higher body the issue?
Danger-adjusted, is that this a superb funding for the corporate, given the engineering price?
If we add one thing to the roadmap, one thing must fall off. Are we okay with the trade-off?
As these options begin to come to fruition, how can we de-risk them?
However these choices can’t occur in a vacuum — that’s the place your cross-functional companions are available. Stewart recommends constructing out product councils to assist combine these insights from throughout the corporate. A product council is a month-to-month enterprise overview, the place a PM for a selected product space will sit down with a cross-functional group of stakeholders to speak about how issues are going. “It’s a course-correction and information-sharing mechanism that we have now month-to-month within the context of this broader annual plan,” he says
Right here’s Stewart’s step-by-step information to conducting a product council:
1. Fairly than making a deck or offering a rating card, write a six-page narrative round what’s working and what’s not working.
2. Deliver 7-10 individuals from throughout the org into the assembly.
3. Put aside one hour for the assembly; start the primary 20 minutes by letting individuals individually learn by means of the memo and add their feedback.
4. The product supervisor then pulls out the primary themes, questions and dialogue factors, that are debated for the rest of the assembly.
Stewart warns towards making this assembly an open invitation. “The extra individuals you may have within the room, the less persons are prepared to lift their fingers and say one thing. Strike the steadiness the place you may have illustration from pre-sales, post-sales, help, engineering, product administration, design and advertising. However don’t open the floodgates.”
Like most startups, Cockroach has gone by means of a number of totally different permutations throughout Stewart’s tenure — and he’s been driving these waves, too. “Once I joined Cockroach it was an open-source firm. Then we began commercializing — that’s a special enterprise. Then we began shifting into managed companies — that’s a special enterprise. Now, we’re in product-led progress mode,” he says.
Right here’s his recommendation for different org leaders seeking to keep afloat with equally shifting tides: “If somebody joins a brand new firm they usually report back to the CEO as a purposeful skilled, it’s unlikely that the CEO goes to have the ability to coach them on do properly in that operate — that’s why they’re employed to be the purposeful skilled. You want to have the ability to take a tough have a look at the place you’re succeeding and the place you’re not on a reasonably common foundation,” says Stewart.
Listed below are a number of techniques he’s leaned on to do exactly that:
Create peer networks. “Join with different people who find themselves operating product groups at akin phases. What are the challenges that they’re going by means of? What’s their recommendation for navigating comparable conditions?”
Discover your function. “In case you have a visionary CEO and that individual is searching for an operator product chief, don’t attempt to be the visionary product chief and struggle with the CEO on the imaginative and prescient. What’s your ‘higher collectively’ story with the CEO? How will you assist them obtain their imaginative and prescient, present a sounding board, or maintain up a mirror when one thing is likely to be working? It’s vital to grasp what function you have to play to help all the enterprise.”
Faucet traders. “One of many issues that I did very early on was I had a dialog with a few of our traders round failure modes they’ve seen for various Heads of Product in any respect types of corporations. What are the issues that I must be taking a look at now to get forward of the talents that I must construct to achieve success as Cockroach continues to develop?”
And his ultimate piece of recommendation for startup people up and down the org chart is to get very particular about asking for suggestions. “I request very candid suggestions from my friends about the place I’m falling quick, given their vantage level. Folks have other ways they like to offer suggestions. Some desire a telephone name, others need to take a second and write down their ideas, so I’ll share the assembly agenda forward of time with the areas by which I’m searching for suggestions. Understanding somebody’s communication norms is useful for getting actionable suggestions,” says Stewart.
However whatever the format, it’s what you do subsequent that issues most: “Should you ask for suggestions and also you get defensive, it’s not possible that you simply’re going to get the identical degree of suggestions sooner or later,” he says.
What I’ve discovered is that it’s not essentially the way in which you ask for suggestions that issues. It’s the way you present up whenever you obtain that suggestions that will increase your possibilities of getting high-quality insights.
This text is a lightly-edited abstract of Nate Stewart’s look on our podcast, “In Depth.” If you have not listened to our present but, make sure to test it out right here.
