Discuss Home · Bootstrapped Podcast · Scribbleton Personal Wiki · HelpSpot Customer Service Software · Thermostat NPS

Supportengine.co - tier 1 support


#1

Hey guys - this is an idea that a buddy of mine and I have been playing around with recently:
http://supportengine.co/

The idea is to basically crowdsource tier 1 support. Let me know what you think. I’d be interested particularly in your feedback as a potential customer, given that most of you guys probably manage some support.


#2

Looks interesting.

Though I am somewhat sceptical that even just tier 1 support can be handled satisfactorily by what looks like “one click closet match from faq and reply goes out with no manual editing” system (e.g. a simple one - what happens when they ask two different questions?). I’ve got lots of “Canned Responses” setup but I very often have to edit the email manually even after some of it is built from the canned responses.

I very much like the idea of the “Dry Run” though - an idea to supplement that could be a ‘hold queue’ where you could turn on simple approval process where you could quickly browse through replies and hit “Yes Acceptable” and it goes out or “Nope” and it doesn’t. Perhaps could enable this for particular (e.g. new) “Support Geniuses”.

I would try this out once built, if I were on ZenDesk but I use Helpspot.


#3

Thanks man! Ya I’m somewhat skeptical as well :). What I know for sure is that some percentage of tickets greater than 0% and less than 50% could be answered in this way. The key would be to improve that % over time while keeping quality high.

Two answers in one would be more of a two click situation. But zero typing none the less.

What made me think about this initially was in seeing how formulaic most support emails were - even the ones that I knew we’re coming from well compensated, highly trained support staff like mailchimp.

Should be relatively easy to add support for other support platforms since that side of the integration would be pretty simple.

Would you be interested in being on the early access list? I.e. You give us access to your account for testing while developing it (read-only) in exchange for a free account for 3 months or something like that once live?

Sent from phone


#4

Oh and the holding queue idea is awesome. I had been trying to think of how to close the loop on the dry run mode. Love it.


#5

Question: who would you want these support responses to be from?

It would be ideal if the end customer felt like they were coming from an individual support team member on your end - which they are. But they would probably be able to realize that they weren’t coming from a specific individual if for no other reason than the timing of them.

They could also come from the name of the actual Support Genius on your team - of which let’s say you have a total of 3 to 5.

But it might seem weird to them that whenever they reply back to John with a more detailed follow up question they don’t hear back from John but rather from Bob (tier 2 support).

So maybe it’s best to have them come from “Company Support”. Possibly even with a link in the footer explaining how the service works?


#6

I’d have to think very carefully about giving even read only access to all support tickets to be honest. Quite a few have sensitive info in them (either mine or customers).

Could do a sanitised list of only tickets that would be happy you seeing (at prospect stage)

  • Logistically how would it be done, maybe give a list of ‘ticket ids’ rather than open acess
  • Its a lot of work for a prospect so we would have to be pretty sure that its worth a go to bother going through the process of identifying tickets to send you.

For me - its unlikely that I could justify the work involved on my side at this very early stage - sorry!


#7

Personally :-

  • I would use individual names - it always helps for customer to feel like there is a human on the other end.
  • With most ticketing systems you may get different person replying so don’t see this as a problem and I don’t think anyone cares as long as they get their question answered well and its not some irrelevant canned reply.
  • “Possibly even with a link in the footer explaining how the service works?” Wouldn’t bother myself - don’t think customers care about what’s going on behind the scenes.

#8

Gotcha - thanks for the feedback re: the From.

For me - its unlikely that I could justify the work involved on my side at this very early stage - sorry!

Not at all - completely understand. Appreciate the feedback none the less.

I was actually worried about sensitive information myself as well. I think that on top of getting the answer quality high, another key thing here would be building trust between business and the support agents. Pretty much in a similar way to how airbnb builds trust between hosts and guests.

When I first thought about this idea it seemed like a quick and easy win but the more I dig into it I realize it would be just like any other good business - requiring a lot of work to actually get it right :smile:


#9

To me the key is less about getting between 0-50% of the replies right, and way way more about not getting ANY wrong. Because this service is all about trust and the minute I get an angry email from a customer about how our support sucks because they asked about X and we sent them an obviously canned and wrong answer about Y I’m very likely to stop using the service.

So perhaps a more realistic goal is to try and take care of 5-20% of extremely common support requests with as close to 0% as possible of a bad reply. Even 10% of a true high volume help desk would probably let them either lower salary or redirect it elsewhere.

Such a low percentage will likely take you out of the bootstrapper/startup clientele I think you’re more targeting though.


#10

Interesting. I definitely need some way to outsource support of my own thing. The current best solution I have is to find someone to work an hour or two a day, and pay them for that. I do want someone who can give real answers and not canned answers.


#11

way way more about not getting ANY wrong

Absolutely agree. I’ve really had to impress this on my own (internal) tier 1 - don’t assume things, don’t be afraid to ask for clarifications and don’t be afraid to escalate.

This would have huge implications for how you ran your own service

  • If you’re paying your ‘geniuses’ per ticket answered then there is obvious conflict.
  • Depending on where you recruit your geniuses from you can get problems with culture as some are much more likely to be so keen to help they won’t/don’t know when to ask for help - see the “Indian Yes” for one common example.

#12

@ian ya that’s what I meant when I said “while keeping quality high”. It would be critical to have a very low error rate.

Ya 10% would probably be fairly realistic. It varies a lot by account. I have 3 accounts that I’ve been spot checking and some of them probably have lower than a 5% simple answer rate, whereas others seem to have upwards of 30%.

Not necessarily looking to target bootstrappers. One of the things I like about this idea is how widely applicable it would be. It’s kind of surprising to me that larger companies aren’t already doing something like this b/c of how much they could save / improve.

I think there would be smart ways to niche this down though. For example for ecommerce, you could have one click return responses that also hooked into an ecommerce api to generate a return, etc. Obviously things like that further increase the importance of accuracy.

@davidw, thanks! So ya I don’t think this kind of thing would be a fit for you if you need 2 hours of total support time. The idea isn’t to replace all support answers with canned answers, but only the answers that should be canned.

If someone emails in saying “how do I reset my password?”, you should be sending them a canned response. Spending a few minutes on writing something by hand isn’t doing anyone a service.

@Rhino you keep quoting “genius” haha. You can just call them support agents if you don’t like our whiz bang marketing copy :slight_smile:

Ya again it would be non trivial to get incentives aligned such that the error rate would be low. We were actually imagining local US people doing this kind of thing as well. Because it’s so low friction and you could kind of gamify the process, it could be something that people did in their down time.


#13

Actually the more I think about it I think a much simpler MVP would be to implement this for companies to use self-serve. This would work around the trust issue of using external support agents. I still think that’s an important and very useful problem to solve but it would be nice to be able to focus on the app itself, FAQ cataloguing, and response generation first.

Problem is that I don’t think any support apps allow you to integrate with their UI itself? i.e. being able to inject one-click response buttons into their UI. And having support agents go into a separate app in order to handle one-click responses on top of what they’re already doing in the main app would just be kind of ridiculous.


#14

Well I think the problem is that nearly all help desk apps have something very similar or exact to one click responses already. Often two clicks (or no clicks if automated with a trigger), but not sure the one click difference is enough to buy and train everyone on something different. Even worse really is that you now have ‘knowledge’ in your help desk app, but other knowledge in a second app and how do you maintain all that.

Not trying to rain on your parade or anything just some problems I see with it.


#15

No it’s cool - I get it :slight_smile: You’re not sold on the idea. I think what would help to visualize this would be a side by side video comparing regular support/canned responses to this one button jobby.


#16

I’m not sure if it’s a good idea. Most of the people who contact you for the first time, including pre-sales question will not get a detailed and personalized response, instead, they’ll get a canned message. It may be good for big companies, where you spend 20+ minutes talking with auto-responder and pressing buttons, but I doubt that will do good for mISVs.