Episode description

The guys answer a listener question about how to ensure customers are happy with your product and how to be proactive in finding out if they aren't. This can be especially difficult if your customers tend to be a little bit introverted...*cough* developers *cough*. Plus Ben reveals the number one reason customers leave Honeybadger!

Full Transcript:

Starr:              00:00          Yeah. I totally have been playing dumb this whole time as a strategy, 100%. It's not like I don't actually know anything, or I'm just making it up. It's 100% a strategy.

Announcer:          00:11          They're just three amigos making their way in the crazy old world of software as a service. Welcome to FounderQuest.

Ben:                00:22          In other news, I deployed the Logplex to Lambda, and it worked like a champ. Unfortunately the economics just don't work.

Josh:               00:31          Again, bit by the economics again.

Ben:                00:34          Yeah. I penciled out the math and it worked. You know you can allocate the amount of ram to the function, right?

Josh:               00:42          Yeah.

Ben:                00:42          I did the math based on 128 megs of ram, which is the lowest option, because it doesn't use that much. It uses about, I don't know, 30. The problem is, the performance and the concurrency was such that we were running, like, hundreds of concurrent Lambda functions in order to service the level of traffic, well, one-third of our production traffic. AWS has a soft limit of 1,000 concurrent Lambda invocations. You can of course get those soft limits raised if you can justify it, but of course, that comes with money, right? More invocations means more money.

Ben:                01:19          I was like, "Okay, 600, 700 concurrent indications, that's really not great. Let's see if we can get that down a bit by increasing the ram," which increases the CPU allocation as well. That works, but at that point, the economics didn't work. Yeah, it was just the combination of concurrency and the amount of time taken to actually process each request.

Josh:               01:46          I've taken a shot at this, and now, Ben's taken a shot at this. You guys are Seinfeld fans?

Ben:                01:51          Mm-hmm (affirmative).

Starr:              01:52          Sure.

Josh:               01:52          Katelyn and I have been watching Seinfeld, re-watching it because she's never seen it. Last night, we were on the episode where Newman learns about Kramer's failed Michigan recycling scheme, where they basically like take the bottles in New York, which are worth 5 cents, and they have to figure out how to work the economics out to truck them to Michigan to get 10 cents. I feel like Ben and I are kind of like Kramer and Newman. Yeah, Logplex on Lambda is kind of like our Michigan recycling scheme.

Josh:               02:31          Well, in the Seinfeld episode, the way they solved the recycling problem was that, Newman works for the U.S. Postal Service, as you know, and on one day of the year, on Mother's Day, the U.S. Postal Service has overflow, they had a fifth overflow mail truck that goes to Michigan. On one day of the year, they could co-op that mail truck and fill it with recycling, and get a free truck which changed the math in their favor.

Ben:                03:05          We need Amazon to provide us a mail truck.

Starr:              03:08          The solution to our economic problems is fraud. That's what you're saying, Josh, isn't that?

Josh:               03:14          Yeah. Amazon does have a truck. What is that truck that you can like move your data center with?

Ben:                03:24          Snowball.

Josh:               03:24          Yeah, Snowball.

Ben:                03:25          Snowball, yeah.

Josh:               03:26          It's like the world's biggest USB drive, right? It's like this semi truck, and it's got a little bit of USB jack on the back of it, and you stick it in your computer.

Ben:                03:34          Yeah. I think our customers and might have some issue with the amount of latency that would introduce.

Josh:               03:38          Yeah.

Ben:                03:39          Oh well.

Josh:               03:40          We just need to find a Newman at Amazon that has access to a Snowball truck, and there you go.

Ben:                03:47          Yeah. You know those Snowballs actually have like an EC2 environment on them? It's pretty wild.

Josh:               03:52          Could take this show on the road.

Starr:              03:56          Oh, man. Today, we have another listener question, and it's from, let's see, I can't find this name on here. Tony. It's from our old friend Tony. Tony asked us another question back a couple of episodes ago, and we answered it, and so we're back for round two. The previous question was all about marketing to developers and stuff like that, and now we're on to customer success around developers. I'll just read it. How about that, instead of me just making stuff up?

Josh:               04:34          Sounds good.

Starr:              04:36          It says, "For an early stage startup in the developer tool space, it's important to talk to customers, get product feedback, build social proof with testimonials and stuff, provide top notch support and love." I like the love part, you know? I like where Tony's coming from. "However, engineers are busy people and do not want to talk to someone," I totally understand that sentiment, "Unless something is broken or if they want to cancel." Yeah. Personally I still don't really want to talk to people.

Josh:               05:06          Yeah.

Ben:                05:06          I was thinking the same thing. I stil...