Should companies eat their own dogfood?

Transcript

Akilah Scharff:
So my visceral reaction was a little bit of nausea.

Paul Slade:
Am I a dogfooder? Yes, yes. I am an addict.

Christine Dela Rosa:
Welcome to Work Check, an original podcast from Atlassian, makers of teamwork software like Jira, Confluence and Trello. I'm your host, Christine Dela Rosa. At Atlassian, a big part of my job is to understand how teams collaborate. And within that, I've seen leaders jump on the bandwagon of popular new practices, and it's made me wonder are these really the best ways to move work forward? On this show, we take workplace practices and separate the hype from the helpful. Each episode, two Atlassians debate how the practice should be applied, if at all.

Christine Dela Rosa:
In today's episode, we tackle the unfortunately named practice of dogfooding. Let's meet this week's debaters. We have Marshall Walker Lee on the anti-dogfooding side.

Marshall Walker Lee:
I want to say for the record, I am pro-dog. I'm just anti-dogfooding.

Christine Dela Rosa:
And Shannon Winter, our dogfood defender.

Shannon Winter:
I'm defending the practice, not necessarily the name.

Christine Dela Rosa:
Sure.

Shannon Winter:
And I want to state for the record, I don't literally eat dog food.

Christine Dela Rosa:
We know.

Shannon Winter:
Although, I did eat cat food a few times as a child. But that's a story for another podcast.

Christine Dela Rosa:
Shannon, I'm going to need you to fill me in on that one off the record. Moving on. If you're not familiar with this term, don't worry. Here's a little dogfooding 101. First of all, no, dogfooding does not involve eating dog food. It's actually a common workplace practice. It just means that employees use their own company's products in their day-to-day work as a part of user testing and product development. Think of it like a chef tasting a dish before it leaves their kitchen. That's how they can learn if it needs more salt or spice before it gets in front of the customer.

Christine Dela Rosa:
The term is most common in big tech, but all kinds of industries use dogfooding. In fact, it said that the word dogfooding was coined at Microsoft back in the '80s. An exec there named Paul Maritz sent an email to their testing manager with the subject, “eating our own dog food.” The email was about getting Microsoft employees to use the company's products at work. And the name stuck.

Marshall Walker Lee:
Okay. We can all agree that it's a terrible name though, right?

Christine Dela Rosa:
Definitely.

Shannon Winter:
Okay. If we're giving dogfooding a hard time because of the name, I feel it's my duty to tell you about some attempted re-brands. An executive at Microsoft actually suggested the term “ice creaming” and another big tech company pushed for the phrase “drinking our own champagne,” which I can absolutely get on board with.

Marshall Walker Lee:
Okay. Okay. I'm definitely pro ice cream, although I do prefer a nice dry Prosecco.

Christine Dela Rosa:
Whatever.

Marshall Walker Lee:
But let me just point out, though, that none of those re-brands stuck. It's still dogfooding.

Christine Dela Rosa:
Okay. Okay. Save it for the debate, which starts now. Now, debaters, the rules: You get three rounds to make your case. Now, I want a good clean debate, no low blows and keep each round snappy. Shannon, Marshall, tell me: should all companies use dogfooding as a practice? Round one. Shannon, you are arguing that every company should dogfood. What have you prepared for us today?

Shannon Winter:
Okay. Let's start with the big one. Dogfooding is just faster than a full-blown user testing process. There's a reason it's used by Microsoft, Google, Apple, YouTube. The big tech company list could go on and on. It helps teams quickly iron out kinks using resources they already have on hand – themselves. I'd like to introduce you to my first guest, Nick Muldoon.

Nick Muldoon:
Can I say g’day day, folks? G’day, folks.

Shannon Winter:
Nick is the co-CEO of Easy Agile, a company that makes software for software development teams. And dogfooding helps them understand their product from their customer's point of view and iterate fast.

Nick Muldoon:
The cool thing about dogfooding is that you'll round off the rough edges. You're just trying to find the best and quickest feedback loop. I mean, that's ultimately what it comes down to. It's like, how do we get the product into the hands of someone and get their feedback as quickly as possible?

Shannon Winter:
So, Nick's company designs tools that help companies do scaled agile transformations. In plain English, they help teams change the way they work on a big scale. In a large company, agile transformations could take a full quarter of planning, if not more, and that would be a whole lot of dog food.

Nick Muldoon:
So to enable us to test and dogfood the product more regularly, we actually use a five week cycle. So we can go through those same activities that our customers are maybe doing once a quarter, just give us a bit of insight into how the customer's using the product and what they're doing.

Shannon Winter:
So, dogfooding basically allows Nick to use his own team as a microcosm for their customers' experience. With fewer people and less time, they can still come away with insights about the way their products might work at scale and fix any problems before they reach that broader audience.

Christine Dela Rosa:
Totally. That's dogfooding working as it should. Marshall, anything you disagree with there?

Marshall Walker Lee:
Yeah. Listen, if Nick claims that dogfooding works for Easy Agile, I'm not going to dispute him. But we're not here today to talk about what works for Easy Agile. We're here to talk about whether or not dogfooding is a good idea for all companies. And here's my concern. What if you or your designers or your developers can't represent your users? What if you have different experiences or you come from a different background? Then this dogfooding process is inherently risky. It threatens to push your team towards delivering a product that works first and foremost for your team.

Marshall Walker Lee:
All right. So my first guest is Akilah Scharff, a consultant at Project Inkblot. They help companies design more equitable products and services, and she has seen a ton of companies struggle and even fail because they relied too heavily on dogfooding.

Akilah Scharff:
At least in my experience, development happens in an echo chamber. So if we're looking at statistics, most development and design groups are pretty homogenous. And so if you're designing for everyone, but your teams don't look like everyone or have the same lived experience as everyone, testing on yourself creates this cycle of validation that isn't truly effective or accurate if your ultimate goal is truly to design for everyone.

Marshall Walker Lee:
It's bad for business if your product excludes a chunk of the population that might want to use your product.

Akilah Scharff:
We've had clients that have completely missed out on great business opportunities, but also opportunities to serve humanity by focusing on their target market so myopically that they've missed out on folks who would be ideal customers for whatever it is that they're developing.

Marshall Walker Lee:
This brings to mind all kinds of examples of design that didn't account for gender or race.

Christine Dela Rosa:
True.

Marshall Walker Lee:
For example, do y'all remember the viral video of the racist soap dispenser?

Shannon Winter:
What?

Christine Dela Rosa:
No.

Marshall Walker Lee:
Okay. If you missed it, a video is shared around Twitter – this was a couple of years back – and it showed an automated soap dispenser that would only dispense soap into white hands.

Christine Dela Rosa:
What?

Marshall Walker Lee:
When a Black user waved their hand under the dispenser, nothing, nada. It just didn't recognize that user's existence. Now that's some truly terrible design.

Christine Dela Rosa:
Terrible. Yeah.

Marshall Walker Lee:
And that should remind us that our products are not born, right? They are made by people and those people can be short-sighted and naive. And in the world of tech, historically, those teams have been pretty homogenous and those people have looked the same. Now that points to a larger problem about diversity and equity in tech hiring. We don't have time to get into that. But what I'm saying is the practice of dogfooding, it assumes that the team creating the products and the customers that they're creating them for are alike. And that's not always true.

Marshall Walker Lee:
Here's Akilah again.

Akilah Scharff:
It's time for us to be thinking outside of our echo chambers. And I always say, when you're designing or developing something, it's got to go beyond like the last 20 people you texted or emailed or DM'd on Instagram. That's not everybody. That's not representative of the public.

Marshall Walker Lee:
What do you think of that, Christine?

Christine Dela Rosa:
No. It's a valid point. While dogfooding might save you time, it can't replace an entire review process. That's cutting corners.

Marshall Walker Lee:
Yep.

Christine Dela Rosa:
So... this round goes to Marshall!

Marshall Walker Lee:
Yes.

Shannon Winter:
Oh, man.

Christine Dela Rosa:
All right. Let's switch the order up. Round two. Start us off, Marshall.

Marshall Walker Lee:
Okay. So last round, I talked about the dangers of dogfooding for the customer. But what about for the employees? You want your team to be proud of their products, right? But what if they eat their own dog food and it sucks? All right. Are y'all ready to meet my surprise guest?

Shannon Winter:
Eagerly awaiting.

Christine Dela Rosa:
Bring it on.

Paul Slade:
My name's Paul Slade. I'm the head of engineering at Atlassian.

Shannon Winter:
Oh, man. This one hits close to home.

Christine Dela Rosa:
Bold tactic. Dogfooding, as you know, Marshall, is embedded into the culture here. Good luck turning one of our own against it.

Marshall Walker Lee:
No, no, no. You're totally right. Atlassian uses dogfooding all the time. In fact, here's how Paul describes it.

Paul Slade:
So we use our software to build our software, which is nice and inception-ish.

Marshall Walker Lee:
Now you might think of that as a good thing, but dogfooding has caused some really big problems, even here at Atlassian.

Paul Slade:
And so we have teams like legal and finance and talent all using our tool, all using Jira, for their day-to-day job to keep the business moving. And when we impact their ability to get their work done, well, that's when things get a little exciting on the dogfooding front and we're told to cease and desist. We've had some examples where a business team has been busy creating all their issues and their project plans in Jira, like good Atlassian citizens only to have those things deleted because the development team didn't necessarily understand there were any examples of them around in the universe, and didn't really think about any necessary data migration. So, that can sometimes be the peril of too much bleeding-edge dogfooding is your data can disappear.

Marshall Walker Lee:
I think we can all relate to the nightmare of lost data.

Christine Dela Rosa:
Oh yeah.

Marshall Walker Lee:
Just imagine, Christine, that you come into work tomorrow and you discover that you've lost hours, days of work.

Christine Dela Rosa:
No thank you.

Marshall Walker Lee:
And now you have to start over again from scratch because of your own product. How would that make you feel about your dog food?

Paul Slade:
Often with dogfooding, you can create a loss of trust in the product. So internally, teams, they want to be proud of the products that their company is building and selling. And so they want to see a high quality product. But with dogfooding, you're exposing your employees to something that might be a little more work-in-progress, and that then can create bugs and unexpected behavior and you can lose a bit of trust there.

Marshall Walker Lee:
As designers and developers and anybody else working on delivering these products, our job is to bring value to the customers. And we should use the best tools for the job, whether they're made by our company or made by somebody else. Wasted time means less value for our customers, and that seems like an unacceptable trade off to me.

Christine Dela Rosa:
Strong showing, Marshall. Shannon, can you counter that?

Shannon Winter:
Challenge accepted. So in that case, we didn't have a great feedback loop setup for the dogfooders to let the product development teams know how they were using the product. It's really not enough to flag random problems. There should be feedback on exactly how the product is being used when you encounter those problems. And that's the beauty of dogfooding company-wide – you get a sense of how a product might be used by various teams.

Christine Dela Rosa:
Sure. Sure. The feedback loop is essential in getting accurate audience input, totally.

Shannon Winter:
But, I'd say my bigger issue with Marshall's argument is that dogfooding done well can actually build team morale. It allows employees to feel like they're part of a team creating something really great together. Just look at Apple. In 1980, the CEO of Apple at the time, Michael Scott.

Marshall Walker Lee:
I'm sorry?

Shannon Winter:
No, not that Michael Scott... Announced that the company was no longer going to use typewriters. They were going to put their money where their mouths were and use their own word processors. Here's the memo he wrote, according to Inc Magazine.

Speaker 8:
Effective immediately, no more typewriters are to be purchased, leased, etc., etc. Apple is an innovative company. We must believe and lead in all areas. If word processing is so neat, then let's all use it. We believe the typewriter is obsolete. Let's prove it inside before we try and convince our customers.

Shannon Winter:
So our friend Michael Scott wasn't actually using the word dogfooding, but he might as well be. “Let's prove it inside before we try and convince our customers.” This idea is at the very core of dogfooding. You have to believe in your product to be able to convince others to believe in it, too. In this way, dogfooding allows teams to know their product inside and out. Sure. If people on the product development team use their own products, they'll be able to iterate faster and better. But if the marketing team also dogfoods the product, they can market it better. The sales team can sell it better. The customer service team can understand and empathize with their customers better.I think you get the point. It's a win-win all around because dogfooding makes for more informed and empathetic employees. And fun fact – so does catfooding.

Christine Dela Rosa:
What?

Marshall Walker Lee:
There's no catfooding. Are we allowed to just invent our points?

Shannon Winter:
No. No. You got to give the cats some love as well. Catfooding is sort of reverse dogfooding, where a group of employees will use a competitor's product for a period of time to really get to know what else is out there. Sounds pretty purrrfect to me. Am I right?

Marshall Walker Lee:
Come on.

Shannon Winter:
Okay. I'll stop. But really, you can't have a solid competitive strategy without this kind of competitor knowledge. With either practice, you're getting to know the user experience more deeply, and that's going to make for a team with more faith in and understanding of the product they're building.

Christine Dela Rosa:
So I think this round comes down to the question, does dogfooding or catfooding, for that matter ...

Shannon Winter:
Yeah.

Christine Dela Rosa:
... Boost or hurt the workforce's faith in their products? While I hate losing files, Marshall, I feel more confident and honestly have more loyalty to the products I use every day. And if the dogfooding process is done carefully, it can build faith. So based on a sample size of one: me, I award this round to Shannon on the pro side!

Marshall Walker Lee:
I got robbed.

Christine Dela Rosa:
And the means we are tied up.

Marshall Walker Lee:
Catfooding.

Christine Dela Rosa:
And just in time for our third and final round. Shannon, go ahead.

Shannon Winter:
Awesome. And thanks for the win. So my last argument is simple. Dogfooding just works. I mentioned Apple earlier with that typewriter memo.

Speaker 8:
Effectively immediately...

Shannon Winter:
But the Apple team didn't stop there. They actually doubled down and created an intense dogfooding schedule. Every day, the engineers and designers would put a new prototype in front of a colleague who'd never seen it before. They'd collect their feedback, work all night to adjust the prototype and then do it all over again the next day.

Christine Dela Rosa:
Ooh.

Marshall Walker Lee:
Wow.

Shannon Winter:
It’s a lot. Believe it or not, a lot of the results they got are still in use today, like the system-wide menu bar at the top of the Macbook screens that I think you're all looking at right now.

Christine Dela Rosa:
Ooh.

Shannon Winter:
We can give a big thanks to dogfooding for that navigation friend.

Christine Dela Rosa:
Eloquent, Shannon, and a relatable reference. Over to you, Marshall. What is your final point?

Marshall Walker Lee:
All right. So I'll concede that dogfooding works, but that doesn't necessarily mean it's the right tool for every company. If you need to whack a nail into a wall, a frying pan works, a shoe works, but a hammer would be a lot better. And dogfooding, it might have worked well for most tech teams in the 1980s. And Shannon, if you want to meet up next week and debate the merits of slap bracelets and parachute pants, I'm all in.

Shannon Winter:
See you there.

Marshall Walker Lee:
But a lot has changed since the 1980s in the early days of Apple and Microsoft. Don Norman didn't even coin the phrase “user experience” until the mid '90s. And since then, our ideas about user-centered design have become a lot more sophisticated. Now there are tons of simple strategies to help us put real users at the center of our work, like AB testing or multivariate testing. The internet has made all of this relatively cheap and easy to do. I mean, sure, it's not as cheap as asking the designer at the desk next to yours if they like your new feature ...

Christine Dela Rosa:
Sure.

Marshall Walker Lee:
... But bringing real users in early and often, it's a worthwhile investment. And one more thing. Before we wrap up, I want to circle back to where we started, the name dogfooding. Now you mentioned, Shannon, that people have tried to re-brand this practice, but those names, they didn't stick. We're still calling it dogfooding. And my hypothesis is that we keep returning to the term dogfooding because the practice that it describes is just like that name. It's a little bit icky. It isn't entirely nutritious. And when we do it, we know deep down inside that something is wrong.

Shannon Winter:
Rude.

Christine Dela Rosa:
Way to bring it full circle, Marshall. And that is a wrap on round three. So does dogfooding work for select companies or is it a practice that should be adopted everywhere? You've both made excellent points and you've led me to believe that dogfooding does work. But if it can't be done effectively, it shouldn't be adopted by all ...

Marshall Walker Lee:
Yes.

Christine Dela Rosa:
... Which means ....

Marshall Walker Lee:
Yes.

Christine Dela Rosa:
... I am declaring Marshall the winner ...

Marshall Walker Lee:
Yes.

Christine Dela Rosa:
... On the con side.

Shannon Winter:
No.

Marshall Walker Lee:
Down with the dogfooding. And catfooding too, although I think you made that up, Shannon.

Christine Dela Rosa:
Dogfooding isn't a perfect practice for every company, but it does have its merits. So how should you apply it? One: remember dogfooding alone does not equal product testing. It's just one tool in the tool belt, not a replacement for thorough, real user testing. Two: think carefully about which teams you're sharing your product with and at what stage. Make sure you have a clear feedback loop in place. As Paul Slade said, dogfooding too early can risk losing your team's faith in their product. And finally, three: dogfooding only works if your testers truly represent your customers. That means representation is key. And if your whole team shares the same lens, make sure they're not your only test audience. And that's it for our episode on dogfooding. Thank you so much, Shannon and Marshall, for your excellent arguments.

Shannon Winter:
Thank you.

Marshall Walker Lee:
Till next time.