How to Figure Out What to Automate

I think everyone who owns a web business wants to do as little work as possible to make as much money as possible.  That’s pretty obvious right.  Given that we’ve built our cart from scratch, we have the ability to automate just about anything that can be automated.  So why don’t we?  I get that question from time to time, particularly from people who have just caught the internet business bug and think that they can just automate everything, do nothing, and sit back while the money rolls in.

The answer is pretty simple: because in many instances there are trade-offs.  There’s always the initial time spent up front. But sometimes even with the automation you end up creating more work for yourself in maintenance. Or, even worse, you lessen your customer’s experience.

I have one really simple rule when trying to decide whether something gets automated or doesn’t.  If it saves us time, requires little to no maintenance, is scalable, doesn’t negatively impact the customer experience (in many cases, we’re trying to improve the customer experience), AND the end gain is worth the upfront time and resources to make the feature happen, then we do it.  Otherwise we don’t.

A good example of a successful automation is the failed checkouts on Detailed Image.  That’s a win for everyone across the board.  Since we implemented that, the number of failed checkouts has dropped to only a few per week, with almost no multiple failures from the same user.  Previously customers would fail over and over and over again because they didn’t know what was happening.  We gave the customer the information needed to solve their own problem, and they have 100% of the time.  I used to answer emails from customers regularly, but since then I haven’t received one.  And they’re less frustrated because they can solve their own problem immediately instead of awaiting an email from us.  Win/win.

On the other side of the page, we haven’t fully automated our inventory system, specifically ordering from vendors and managing “out of stock” items.  The comments on that post were fantastic.  Probably better than the post itself.  A lot of people posed questions about why we couldn’t go any further.  I did my best to explain our situation and why a fully automated system could potentially cause more headaches for both us and our customers, in addition to the upfront time to create, test, and deploy the features.

A more extreme idea would be to try to automate customer service.  The customer emails us, we scan our large database of replies, and automatically email them back our closest match.  Sounds crazy, but it’s possible.  Most companies that do this do a hybrid version where they send back suggestions automatically but still have a real person follow up.  We might do that at some point. However, that would require a lot of upfront time/money/work, and it would probably piss people off.  One of our best competitive advantages is that we reply to our emails with real replies from real people, and we do it relatively quickly.  Instead of fighting everything with automation, the best long term business solution to handling our ever-increasing volume of customer service emails might simply be to just hire a customer service rep.  All of the programming in the world can’t create the quality answers that a good employee can.

With every single business problem we always go through this same iterative thought process.  It makes for interesting conversation because we can attempt to automate just about anything.  Just because you can do something though, doesn’t mean you should.  Or, as Kumar said, “just cause you’re hung like a moose doesn’t mean you gotta do porn!”

5 comments on How to Figure Out What to Automate

  1. Rob says:

    Great post – very insightful. Choosing what to automate is a decision that might seem easy for someone new to a particular area, but really should only be taken after a lot of thought and evaluation. Some things are no-brainers when the cost/benefit ratios give clear indications, but other things take a sound understanding and a lot of experience to be able to see, consider and work out how (or if) they should be automated. Hiring staff is always a possibility. You need to train them properly and give them the power to help the customer though, rather than just using them as a barrier like so many companies do with their phone scripts and foreign call centres that sound like they’re halfway under the ocean (have you tried turning it off and on again?…). I love the example on Zappos about the customer service rep refunding someone for shoes that were bought as a gift for someone that died, and having flowers sent out to the widow without needing to refer up to a supervisor.

    I know it’s a *little* off topic, but when you mentioned failed checkouts I first thought of abandoned carts… Do you have a system for recording or following up on abandoned carts? There’s some ad networks out there that can work with abandoned carts and follow the customer for a couple of days, I think it’s called remarketing.

    • Adam McFarland says:

      Good question Rob. We don’t have a system for following up with abandoned carts, but it’s something that’s on our to-do list. Something as simple as an email a few days later with a coupon code for a product they were thinking about buying. I think it could be successful so long as we do it right.

      • Rob says:

        Apparently the conversion rates are really high, you’ve just gotta avoid being seen as “big brother” or pestering the (potential) customers.

        • Adam McFarland says:

          Exactly. We’ll definitely spend a lot of time figuring out who to email, when to email them, and precisely what to say.

  2. Tamajong says:

    Excellent post! Quite a bit of food for thought, and many awesome points were made. Thanks!

Leave a Reply

Your email address will not be published. Required fields are marked *

Commenting Rules

I'm honored that you found this post interesting enough to leave a comment. Before posting, I have a few ground rules:

  • Please keep your comments as relevant to the post as possible.
  • No personal attacks or any other nastiness.
  • Your first comment is subject to my approval.

Thanks!