Some friends wonder why, as someone who considers himself a strategist, I care so much about execution. Is it because I’m a perfectionist? Well yes, but that’s only one reason (and perfectionism is a weakness  in business, not a strength). Is it because I like metrics, measurement, and incremental improvement? Yes, I do. Is it because I like improving efficiency and minimizing cash burn? Yes, even though that doesn’t matter nearly as much as it once did.
But are any of those answers the real, big reason why I care so much about execution? No. This quadrant is:
In business, while we all may have strong opinions about what’s going to work and what isn’t, if we have even a trace of humility we must admit that we don’t know. Therefore, everything we do is an experiment. When an idea succeeds, great — let’s scale it up. But it’s also quite important to know, when an idea fails, why.
That’s why execution matters. Because when an idea doesn’t clearly succeed you need to be able to determine whether you’re in Box 1 or Box 4 of the above quadrant .
Saying a new corporate initiative failed because of bad execution is like saying a lab experiment failed because the petri dishes were dirty. It shouldn’t even be a possible cause of failure. You should have controlled for that. You should have hired professionals to run your experiment.
If you’re going to spend $5M of your investors’ money to try an idea, if it doesn’t work you should be able to explain why — and be damn sure that execution is not a probable reason. The whole Silicon Valley model is about isolating and removing risk from the equation. That’s why boards want startups to pay high salaries and lure experienced talent with lucrative stock options. It’s not because VCs like high compensation packages and dilution. It’s because they want to optimize the chance of something working and, when it doesn’t, they want to be able to say, “we spent $5M and proved that was a bad idea,” as opposed to, “we spent $5M and we’re not sure whether it didn’t work because it was a bad idea or it was just poorly executed.” 
If Sartre said, “hell is other people,” I’d say, “hell is not knowing why you failed.”
Let’s take European expansion as an example. You have a nice US-based $30M SaaS business and you want to expand into Europe. But you’re not fully committed, the board is split on the decision, you’re worried about the CAC impact of initially unproductive sales investments, and you’re not sure what to do.
You think, if you had the proper budget, you’d:
- Open in both the UK and Germany to run the experiment in parallel
- Hire only sellers with experience in your space, with the same profile as the ones you hire in the US.
- Hire 3 in each country to make sure the outcome isn’t dependent on any one bad hire.
- Hire 2 SDRs and 2 solutions consultants in each country to support the sellers (the same ratio you use in the US).
- Hire a few other staff like in functions like customer success, support, and services to support the team.
- Hire a VP of Europe to lead this — someone experienced but who still enjoyed being close to the action; they’d be expensive but they could scale the operation as it grew.
- All in all, you’d hire maybe 20 people and spend $4M to get going.
But given the board situation, you decide to do it on the cheap:
- You hire one seller in UK and one in Germany, who are both junior in their careers and have never worked in the space, but they’re cheap and seem aggressive. Good sales DNA you convince yourself.
- You tell them to work primarily through partners because you don’t think you can afford to go direct.
- You have them share a UK-based solutions consultant who doesn’t speak German.
- And that’s it. You spend less than $1M, but at least you’re getting started.
Now, zoom forward to the end of the year. It didn’t work. The two sellers both quit and the solutions consultant is trying to support the few customers they sold. It’s a mess. And what have you learned? That Europeans don’t want software in your category? That a partner model doesn’t work in Europe? That it’s hard to find good talent in Europe?
Nope. You haven’t learned anything — except that bad execution leads to failure. And you knew that already, didn’t you?
Maybe you think you learned not to hire junior sales reps, that reps need a proper level of supporting staff, and that customers like to get demos in their native language? But you knew that already, too. You haven’t learned a thing, but you’ve wasted $1M, damaged your brand, and most importantly, lost a year.
By the way, if you really wanted to know if you could hire junior sellers and make them successful — if that was really the question you were trying to answer — then shouldn’t you have held all the other variables constant and just tested that? In the US, at perhaps smaller companies but in your same target market. With your standard support ratios. With experienced leadership.
Imagine if you ran the other play, the one with the proper budget. If it worked, you’d be worried about scaling up, not still worried about how to open Europe. If it didn’t work, then you’d have some really hard questions to answer. Because we can be pretty sure it’s not the people or the support ratios or the sales model — or execution in general. We’re probably in Box 1 — there’s likely something about the idea that’s wrong. Maybe, to pick a trivial example, Europeans don’t want to buy your compliance software because it’s weak on supporting European regulations .
The Idea/Execution Quadrant
It’s no secret that I like quadrants. I made this one because I wanted to separate business outcomes from idea quality and execution quality. Box 2 and Box 3 were easy to label. I frankly struggled with the labels of Box 1 and Box 4. So let’s talk about them in more depth as I’ve lived in both .
Box 1: Bad Idea, Good Execution. For me, this was MarkLogic, an XML database system . During my six years there we grew the company from $0 to $80M in revenues, so I have trouble labeling that box — as I initially did — “failure” . While I also considered labeling it “partial failure,” in the end I chose “partial success.” We brought great execution to a hostile environment  and had enough success to confuse people — by the numbers we resembled many destined-for-greatness companies, but those who looked beyond the numbers knew we were not . As one very smart VC summarized it, “you’re still pushing” — a great definition, in fact, of partial success. You’re driving growth and making numbers, but you have no tailwind from the market. When you’re in Box 2, you’re HODL-ing to stay on top of the market — in Box 1, by comparison, you are fighting for every yard. That’s why I labeled it “partial success.”
Because if you bring good execution to a bad idea , smart people can often figure it out, and you can drive some success. But it won’t be easy and it won’t scale .
Box 4: Good Idea, Bad Execution. For me, this was Ingres, my first job out of college. If I told you I joined a startup after school, stayed there 7 years, and it grew from $30M into a $240M division of a $400M company, you might be tempted to say “success.” But you’d be wrong. One clue is obvious: division. We grew, but not so much that we were able to stay an independent company. The other is non-obvious. During those same 7 years our archrival (a company called Relational Software) grew from $30M to $1B. During that period Relational Software changed its name to match that of its product: Oracle.
The idea was clearly good. In terms of wealth generation, the RDBMS was the second best idea of the 20th century . Did Ingres succeed? While my quip is that I experienced great success and great failure simultaneously , there is no question. No. Ingres failed. It got third place in the race of the century. Second place (Sybase) was a set of steak knives. Third place was for our acquiror, ASK, to get acquired by CA for less than 1x revenues. Was our execution bad? I think so, not only because of hindsight bias based on the result, but because I worked there. While I go into considerable depth in this post, I think the simple answer is that Ingres never really understood either the stakes of the game it was playing or the power of increasing returns in software market leadership.
You could argue I should label Box 4 “failure,” but we did manage to grow the company, go public, and achieve other key milestones — so, optimists could argue that we were succeeding. Semantically, would I argue the partial success is not success and thus failure? Maybe. Practically, do I want to label Box 4 “failure” and potential enable misguided optimists to argue that some success equals success and that they’re ergo in Box 2? No.
That’s why Boxes 1 and 4 are so difficult. You’re not sure if you’re succeeding or failing and you’re not sure why. Which, in the end, is why execution matters. So you succeed when your idea is good and so you can rule out execution as a factor when it’s not.
# # #
 As it took me way too long to realize, you need to focus on getting what matters right. It’s one of several points discussed in this deck.
 See the second part of the post for more discussion of the quadrant itself.
 And now they need to decide if we want to spend another $5M to try it again, but this time with professionals.
 And a professional team would likely figure that out and tell you. The amateurs might offer 100 different excuses and/or forms of hope.
 Happily, I’ve also done plenty of time in Box 2!
 I am not speaking of MarkLogic in its contemporary form (about which I know fairly little), but as it was in the 2004-2010 period when I ran it. Disclaimer: I own some residual shares in MarkLogic.
 Though a VC might easily do so. The investor view on these things is pretty simple: did I make a lot of money? The operator view is different. We grew a business, in an environment as hostile as a sea-floor vent, to $80M and solved plenty of hard problems for plenty of happy customers in the process.
 Something like 15 of the 16 original XML database companies basically went out of business. I’d call that hostile.
 Specifically that NoSQL was emerging and that those systems (e.g., Hadoop, MongoDB) and their associated open source models, were going to become the mainstream way to solve the problems that our technology solved.
 This begs the question of how to define the idea. In MarkLogic’s case, what was the idea? Building a search engine with database parts and using a distributed scale-out architecture were good ideas — very good, in fact. But the overall business idea was that XQuery would do for MarkLogic what SQL did for Oracle. That, because XQuery never took off (and was arguably a case of infantcide by the megavendors) was a bad idea, and the one that put us in Box 1.
 Critical thinkers may be wondering if I’ve rationalized the company into Box 1 because (as a non-founder CEO) I was responsible for the execution and not the idea. Let’s test that. The team who took over the company was definitionally “better” in the minds of our top-tier VCs who brought them in and who are presumably skilled at evaluating such things. Ergo, if you were in Box 4 and upgraded the team (and ergo execution), you should move to Box 2 and see improved results. That didn’t happen. Similarly, if you look at the idea compared to other companies pursuing the same idea (see endnote ) you can’t find any evidence that it was a good idea. Had 1-2 competitors done materially better than we did in size, growth, or exit value that would suggest the idea was good and our execution bad, but that didn’t happen either. Hence my belief that we were in Box 1.
 First place was PC operating systems with Microsoft.
 Because we did face the challenges of high growth.