What we have learned about weekends and how fast to hire: The November 2013 Buffer Happiness Report

I just hit send on an email to the Buffer team with all of the numbers and trends for November. November was super exciting as we were still recovering from the October hack, and we brought on 2 really talented new heroes. Here’s the full report! :) 

Statistics:

Emails via Help Scout:

Customers helped: 4,614 (a 21% decrease from October)
Total Email Replies sent: 6,855 (a 22% decrease from October)

Note: this month stays on the trend of a general increase from month to month, but since October’s volume was so high due to the hack, November is actually a decrease if compared strictly to the previous month.

Percent within 1 hour: 27%
(October was 39% within 1 hour)

Percent within 6 hours: 56%
(October was 77% within 6 hours)

Happiness via Hively:

94% Happy (no change)
3% Satisfied (no change)
3% Unhappy (no change)
702 total ratings (no change)
19 “Wows”

Twitter via Sparkcentral:

Tweets and DMs sent: 5,105

Discoveries:

1) Weekend support can be drastically improved.

On weekdays in November (including Thanksgiving), our average “new conversations” per day is 209. To handle that, we send an average of 269 replies per day.

Contrast that with weekends, where our average “new conversations” per day is 109. We’re sending an average of 118 emails per day.

(This is a bit inaccurate due to the fact that Åsa in Australia comes online during our Sunday. If we isolate Saturdays, when no hero officially works, we sent an average of 61 emails per Saturday, despite 95 new conversations per Saturday.)

Simplified: each weekend day has half the volume of a week day. Thus, each weekend has the volume of one whole weekday.

Not having anyone staffed on the weekend is affecting our stats, and more importantly, causing a pretty unhappy experience for hundreds of customers. It also makes Mondays quite tough.

We staff between 3 and 4 happiness Heroes each weekday. Generally, if our weekend day load is half the normal weekday load, then right now 1 person working full-time on the weekends could make a big dent. That would be a significant improvement over what we’re doing now, which is that everyone simply pitches in when we have time.

So, our proposed solution is to hire someone full-time whose schedule will include the weekend. Joel came up with an awesome title of “weekend warrior.” It will be quite exciting to start hiring for this. Stay tuned to our jobs page for this listing!

2) Onboarding new heroes doesn’t take as long as

we thought.

A few months ago, we dug into the onboarding process for the three previous heroes. We found that new heroes are up and running within 1 month. This time, we dug into the weekly stats, and it turns out that our two new heroes were at or near normal volume by some point in week 3.

Meaning, heroes are basically onboarded within 2 weeks.

So, we should keep hiring rockstars like Mary and Adam, and we can count on a change immediately. The hiring process is still quite slow and deliberate, but once we get the right person, we can expect a big change within 2 weeks. This is especially evident when we compare the speed stats for each work week in November.

November speed by work week

3) We should hire 1 new hero every 2 months.

Over the last 6 months, our total replies sent have increased an average of 10% per month. A 10% increase for a team of 5 is the equivalent of half of a person.

So, if we’re going to just keep up with the increasing volume, we should hire 1 new Hero every 2 months.

We’ve spent some time thinking about this one, and we’ve made the decision to keep growing our team instead of trying to “scale” support by directing people to self-help resources. We intend to continue to create resources (like the FAQ) and solve bugs and areas of confusion by keeping the product team informed, but we want anyone who uses Buffer to feel completely happy and comfortable to reach out to our team and expect a prompt and friendly answer.

4) Not a discovery, just one number on the hiring front:

Between Happiness Hero and Community Champion hiring, the Happiness team received 897 applications in November.

  • Gabe Smith

    I know Helpscout provides this data for you in the reports section, but how are you analyzing your month-to-month and week-to-week changes? I don’t see an export option in helpscout, so I would assume you’re collecting it all in a spreadsheet manually to analyze trends?

    • carokopp

      Hey Gabe! Thanks for asking. That’s right, Helpscout provides the data for any date period, and then I add the numbers I want to see in a google drive sheet. :) My customer happiness spreadsheet is a beast! I’m sure there are better ways to do it, I just don’t know it. :) It’s quite fun to see how the numbers change over time, all in 1 place.

      • Gabe Smith

        I would love to see how you’re processing the data from Helpscout. Do you mind sharing the google drive sheet you’re using? I love the insights that Helpscout provides, but wish it gave more options when viewing the data through the website itself. Maybe this is something they’re still planning to develop.

        • carokopp

          Great question, sure thing! Thanks so much for asking. Here is a link to the last 6 months’ worth. I just pull the data for each month and then the spreadsheet calculates the change. Does this give some insight, Gabe? :) https://docs.google.com/spreadsheet/ccc?key=0AoNp29qBm0uxdHNIdnl5Vi1xZ21HVEtJekZXWHpjM0E&usp=sharing

          • Gabe Smith

            This is very helpful! Thank you so much for sharing it with me. I plugged our numbers into it and they’re all over the place! +23% one month and -40% the next month. We’re dealing with much smaller numbers than you are though, so they percentages will be more erratic. Thank you again, Carolyn!

  • Heather YamadaHosley

    Thanks for the note reminding us of the wonky stats in October due to the hack. At first glance, I couldn’t believe the happiness team could’ve decreased in responses! :) Glad the new HH’s are getting caught up quickly and that you plan to keep growing CS.

    • carokopp

      Thanks so much for mentioning that, Heather, I had the same thought until I realized what was happening there! :)

  • http://www.romaingiacalone.com/ Romain Giacalone

    Thank you for sharing this experience. Wow 897 applications in november, it’s a big worldwide competition

    • carokopp

      Hey Romain! Thanks so much for reading, yes, quite exciting to hear from so many people all over the world. :)

  • http://seandillonsmith.com/ Sean Smith

    Those are some wild statistics! Glad to hear the new recruits are killing it for you guys! Doing a great job as always! Love the attention to community service as your main marketing initiative! It’s 5x easier to keep a client than it is to acquire one! (and you get a word-of-mouth referrer for life)

    • carokopp

      Thanks so much, Sean! You’re so kind to say so and for joining us on the journey here. Definitely agree that it makes good business sense to take care of your existing customers, well put! :)

  • nthampy

    Great post | Wondering why Buffer wouldn’t consider using one single product like http://www.freshdeks.com instead of HelpScout + Hively + Sparkcentral?

    • carokopp

      Great suggestion, we are always open to exploring new tools; we’ll take a look! We’ve generally found that a few tools, each specific to one task, works well for us instead of 1 major tool, but that could definitely change. Thanks for the recommendation! :)

  • Terry Harmer

    Just curious, but how do you onboard someone living in a timezone where there is perhaps only 2-3 hours of crossover per day ?

    • carokopp

      Great question, Terry! :) We do a lot by email and try to cover as much as possible in that 2-3 hours. But, ultimately, most of the onboarding gets done by the folks nearer to the right timezone. :)

  • T-Moor

    Thank you for such an interesting post! I want to apply for the Weekend Warrior position and I was wondering if you receive around 900 job applications, do you look through all of them and reply to each applicant? How do you handle such a big amount of applications per advertised position?