Make sure you've built a functional admin panel before you launch.
Sorry, but this is bad advice. For 99% of startups, their biggest risk is not going to be spending too much time mucking around in the production database because they don't have an admin panel, but that no one is going to show up or care when they launch their product. Building cool stuff like admin panels, metrics dashboards, and loggers is just more procrastination to keep from doing the most difficult thing: launching.
If you're going to build those things at some point, make it after you've seen that users care enough about this product that you're not going to scrap it and move on to the next one. Otherwise you're just wasting effort.
EDIT: To be fair, I think this is otherwise a great post. It is important to build those things early but I think building them pre-launch and pre-validation is a big mistake.
I think it's symptomatic that the author does not mention building the part where people PAY YOU MONEY.
Payment integration is a big pain and is something you should get into the product early, because grafting it on later will hurt. The same goes for ad management, if that is your business model.
Given that my startup I not only had people paying me money and then turning it around and giving it out to others, I definitely know that pain. I just totally thought of it as core to the product; people generally have asked me about gateways and whatnot when they have their "ohmygodIhavethisidea" conversations with me.
This was more about the stuff they didn't ask about that they should have, but your point is exceptionally well taken - people who are picking a revenue model should support that revenue model before launch.
Are you acting as middleman ? In the sense, people are paying you money for products/services you don't own ? (An example would be online travel companies)
If that is the case, which gateway did you choose. From the gateways I spoke to this type of operation is difficult to get approved unless the final vendor itself has a merchant account. Was that the case in your startup?
Yes, Dawdle.com acts as a middleman between buyers and sellers. We got a merchant account through Paymentech and used Payflow Pro as our gateway to take money in and used PayPal's Mass Pay API to send money out (after an aborted attempt to use ACH for disbursement). Dawdle is the merchant of record and so holds the proceeds in escrow for some period of time to mitigate risk.
I'll just offer some counterpoint, as we didn't do this. We created accounts manually with a script for a long time. By the time we needed a panel (leads were starting to get out of hand), I knocked out a fugly one in Rails in one 20 hour coding marathon. Then I banged up an email parser in Perl to add all of our current accounts / correspondences. That was just a couple months back.
The biggest problem that I could see with building a panel up front is that you might be building it for the wrong product. In my eyes, testing your product to see if you get any traction would probably win.
Ditto. When we needed to do a little admin, we just modified the database. (A little risky, I suppose, but we were doing regular backups.) We only wrote administrative tools when things got too complicated for the least-technically-sophisticated member of the team - me - to easily tweak things.
As it turned out, some of the most common customer support issues we had weren't anticipated by us in advance, so building out an admin panel would've still only met part of our needs.
Make sure you build a real metrics dashboard and implement a real logger before you launch.
Just launch it, then figure out how to instrument and admin it. Yeah, it will be painful, but at least you'll have some idea what you're trying to accomplish.
+1 for Django admin! The amount of effort it alone saves would make Django worth while (for me) even if the rest was no better than PHP (luckily that's not the case though).
There's an overwhelming amount of cool-looking options for analytics. Is there a standout option for logging and visualizing everything that one would need, or is it better to have two or three different analytics suites?
The pitfall I've run into when building CMS interfaces for the sales types is that they always want to do something you didn't enable it to do. The image always displays on the right, so naturally they want it on the left, and you end up having to go muck with the code anyway.
Sure, there are plenty of vetted CMS systems out there (e.g., Drupal, Wordpress, etc.), but they're massive overkill when you only need an events list or something.
Sorry, but this is bad advice. For 99% of startups, their biggest risk is not going to be spending too much time mucking around in the production database because they don't have an admin panel, but that no one is going to show up or care when they launch their product. Building cool stuff like admin panels, metrics dashboards, and loggers is just more procrastination to keep from doing the most difficult thing: launching.
If you're going to build those things at some point, make it after you've seen that users care enough about this product that you're not going to scrap it and move on to the next one. Otherwise you're just wasting effort.
EDIT: To be fair, I think this is otherwise a great post. It is important to build those things early but I think building them pre-launch and pre-validation is a big mistake.