- The Analytics Note
- Posts
- Larry rule
Welcome to the fourth edition, it’s about now in the life cycle of a new newsletter that you’ve covered the first few topics that were top of mind - and then begin wondering what else you wanted to cover! But everything is moving so fast right now, I think we’re covered, have you tried the Gemma model? I had a quick play but think its use cases are a bit narrow.
Someone recently asked me at an event - hey what’s the last SQL query you ran? What an ice breaker. But the point was, with AI you can get it to run it for you, or to help craft the right query. To be fair, I couldn’t recall ha.
Zack Kantar shared a tweet which I thought was excellent. He was sreenshotting this from Larry Ellison:
"We used to deliver incomplete software, and you would finish it. Now we're delivering nearly complete software-software that will meet eighty to ninety percent of your needs without any changes. So we're asking customers not to try to put in the last ten, fifteen, or twenty percent. If you do, you'll fall into an old trap. If you heavily modify our software, it will be difficult and expensive for you to take a new version of that software a year from now. If you heavily modify our software and you call us with a question, we'll have a hard time answering the question because you wrote a lot of the system, not us. You are better off, I submit to you, with an eighty percent solution installed and working in six months than fantasizing about a hundred percent solution that you might finish in two years after you write lots of custom code."
-Excerpt from Softwar: An Intimate Portrait of Larry Ellison and Oracle
In the last coupe of editions, I’ve mentioned some of these analytics projects with the road to nowhere. And on balance I think the above quote from Larry is spot on. Would you believe it’s from over 20 years ago? It could be uttered today and be just as on point.
It’s because he’s right. It is better to have that 70/80/90 percent on a common platform that hundreds or thousands of other customers are subsidizing the cost, the R&D and ongoing improvements. And finding a way to work with the last amount.
Because if you do that, then as a team you can focus on other things that help lift the needle.
I may be biased, given I offer an analytics platform, and we do do a similar approach. But I only caught this tweet recently, it’s not that we go by the Larry Ellison rule, it’s that it’s a good product mantra. You can’t build every feature that every customer needs. But you can build as much as you can and offer some generalized solutions that help solve the rest or fill the gaps. Hence custom fields in nearly every piece of software.
If you’re looking at or specifying out a new project, do keep it in mind and use that quote as support internally. If you try to tackle everything, you end up bearing all the cost. At least if a platform integration goes afoul you have access to their talent/knowledge to get it back on track! Call it the Larry rule, he’ll love you for it.
Notable stories this week
Amplitude had their earnings call and shared their Q&A. Worth digging in to.
Using the Wayback Machine and Google Analytics to uncover disinformation networks. Smart.
Tableau Pulse uses generative AI to create data analysis on its own.
Haleon says Attention is a good measure of media quality, but it’s too soon to make it a buying standard.
How I Left Facebook to Forge My Path in Data Consulting.
Deals/M&A
Data visualization of the week
Tips receiving during 10 months as a server.
Smartest commentary
“If you manage to avoid the trap of building custom software, the next boss to beat is avoiding the trap of customizing off-the-shelf software. The question to ask over and over again: how can you eliminate ""requirements"" in order to use software ever-more idiomatically?” -Zack Kantar
Datapoints of note
$60m/year is what Google is paying Reddit.
That’s it,
-Ben
..
See our most recent on why we measure attention.
The Analytics Note Resources [continously updated] ← please do send any you recommend, books, podcasts, websites, even other newsletters.