My first project in consulting was working on strategy for a large restaurant (I obviously can’t share McWhich). They were trying to figure out how to modernize their menu and dine-in experience. And so like good management consultants we gathered up a bunch of industry data, we did a consumer survey, we talked to experts. And our recommendation was basically something like “make wraps, the millennials, they love ‘em”.
Our final deck was 110 pages, with lots of tidy charts and graphs. We were paid eleventy trillion dollars.
You know the one thing we didn’t do? Go into a single restaurant. Which is important (and idiotic), because if we had what we would have quickly realized is that many, many of them had become decrepit, Mad Max thunderdome shit shows. And their menu proliferation had resulted in food to match. Fixing those two things became the pillars of their ultimate turnaround, no thanks to us.
The moral of the story is this, and it’s somewhat counterintuitive for a lot of folks: data will give you averages, anecdotes will give you insights.
At DoorDash we called this search for the root cause of any problem ‘getting to the lowest level of detail’.
Encourage your customers to yell in your face.
A group of us spent the first many months running CX at DoorDash very similarly, analyzing data from the comfort of our Aeron™ (brag) chairs. And in our infinite wisdom we’d pretty much figured it out. Just get orders delivered on-time, and don’t cancel too many. What humble geniuses we were.
It wasn’t until our first visit to our front-line customer support teammates in Manila that we saw the picture a bit differently. Sitting in a windowless room, hopped up on Jollibee fried chicken and ketchup spaghetti, we were, for over an hour, torn to shreds by our friends.
“Do you know how #$%^ing irritating it is when a customer doesn’t notice that they’re ordering to their home address when they’re at work? And then, being unable to cancel, having to watch their driver slowly inch ever-further away from them with their hot, delicious Chipotle quesadilla1?” Heads around the room nodded; their quiet, piercing eyes all said one thing: “shame...”
One after the other, we heard stories of how agonizing this experience was. It wasn’t the most frequent but it was apparently the most likely to have someone send their phone screaming across the room, smashing into their NASCAR commemorative plates.
The learning; listening to your customers (no kidding) will give you insights into not just the volume of their bad experiences, but also the severity, and help you prioritize.
Find the disasters.
Most late DoorDash orders are about 5 minutes late. Every now and then, however, you get one that’s 9 days late.
Every Friday afternoon for 5 years I met with a large group of senior people - DoorDash’s COO included - and we would review, to a meticulous level of detail a single delivery that had gone not a little south, but that had gone slipstreaming into the cold quiet void of never-before-seen, Upside Down heinous awfulness. And that was because you could learn so much more about where your people, processes, and product were broken by looking at just one horrible edge case that went completely off the rails.
Our 9 days late guy? Restaurant was closed, we didn’t know. Order was accepted via tablet by a cleaner who randomly hit a flashing button. Dasher was sent to another location of the same restaurant and he picked up someone else’s order. Left his parking brake off. Car rolled into a ravine. Exploded, destroyed his phone, launched the order 2,000 feet in the air. Landed on a seagull. Seagull flew for 9 days, was shot over SFO (it lived). Order landed on the runway. Baggage handler ate it.
Or something like that. We would go through every detail, every button click, every customer communication, every dispatch trigger, trying to understand all the breakages that had allowed such a seemingly impossible nightmare to happen, and then assign out actions to fix the root problems for the next week’s meeting. The result was that by fixing the root causes of the worst of the worst, we slowly built a robust system that kept every delivery better on track.
It might sound crazy but it ain’t no lie: why, why, why.
Take the example above. The first issue was that the restaurant was closed in reality but still open on the app. Well why (1) was that?
Oh it was because the restaurant owner hadn’t marked it closed before she left for a long weekend. Got it. Makes sense.
No, Mr. Turtle it does not make sense, we are nowhere near the center of the Tootsie Pop.
Why (2) hadn’t she marked it closed? She’d done it so many times before so we know she knew how. She must have forgot. Why (3) did she forget? Well we called her after the fact and turns out she had Thirty Seconds to Mars tickets, was way too excited and forgot on her way out the door. It happens.
But wait, our 9 days late order was the 6th order sent to the restaurant that day, the prior 5 going unconfirmed and then abandoned. Why (4) didn’t we use that signal to automatically shut the restaurant, or at least ping the owner’s phone to ask what was up?
Good question. Seems like it might be a good potential solution…
The more “whys” you’re able to answer - ideally until you can ask why no more - the closer you’ll get to the final and ultimately root cause of any problem.
Put it in your eyeballs.
Back in the day there was a problem a group of DoorDash operators were trying to solve (I am going to butcher this, so light up those comments). They knew from their own experience doing deliveries that wearing a bright red DoorDash shirt helped get identified inside a restaurant, and pickup their customer’s order faster (thus improving the customer experience for everyone involved). What they didn’t know was how often Dashers were wearing the shirts they’d been given, and as such what to make of this effort.
They tried a number of things to figure out the answer. They texted Dashers, they tried to get Dashers to send pre-Dash photos, they thought about asking restaurants to confirm how many shirts they saw.
Ultimately all it took was one person going and sitting in the parking lot of the restaurant for a week straight, watching internal tools and noting every time a Dasher was or wasn’t wearing a shirt. That experience ultimately became the case study that all prospective operations hires were given.
Another story. In 2019 DoorDash had a very popular regional Atlanta chain that unfortunately had very long wait times for Dashers picking up their orders. No one could figure out the problem so eventually a group of folks got on a plane and flew to Atlanta and spent an entire day sitting in the restaurant, watching what was happening. And what they observed was that the tablet had been strategically located behind the bar so that whenever a new order came in the bartender could easily completely $%^&ing ignore that order. And rightly so; he was mixing drinks, flirting with customers, trying to make tips, saving up to buy a new ficus for his apartment. Meanwhile this stupid tablet was helping him not at all and as such he’d use the Dasher walking in the front door of the restaurant as the prompt to submit the food to the kitchen. A huge wait ensued. The solution here was complicated but it started at looking the ultimately very obvious problem square in the face.
So in summary, a few ideally helpful tips for successful problem solving:
Start by letting your customers tell you where the real problems are
Use seemingly impossible, long-tail ‘disaster’ experiences to light the path toward insights
Keep asking why until you can’t; that’s the root cause
If all else fails get up and go physically look at the problem
Some psychos must be ordering those things.
OMG YES! SO many times my team would show me dashboards with unsatisfactory results (e.g. downloads not hitting target) and I would ask WHY NOT? What do the users say about our app? How can we make it better? And they'd stare blankly at me. Nobody thought to ask a user. Any user. Anything. They'd just push out features and pray.
Then when our marketing team did research and figured out what features customers liked best, they put out marketing messages about it on social media. Except that people don't need a fancy AI--powered multi-stop crowd-avoiding elevator-finding mall navigation tool when they're on their butt scrolling the 'gram on the sofa at home. They need it when they are LOST AT THE MALL. So I asked 'what's our in-mall marketing plan? Blank stares.
Finally I said PLEASE SOMEBODY GO FIND AN INTERN TO STAND BESIDE THE MALL MAP FOR AN AFTERNOON AND TELL LOST-LOOKING PEOPLE THEY CAN DOWNLOAD THIS MAP INTO THEIR PHONE SO THEY CAN REFER TO IT WHEN THEY ACTUALLY NEED IT, WHICH IS WHEN THEY ARE NOT NEAR A PHYSICAL MAP. THEN MEASURE HOW MANY DOWNLOADS AND ACTIVE USERS WE GET THAT DAY. YEAH?
No. Blank stares. Nada.
The weeds are where the ugly truth lies, but they're ugly so... nobody likes to look there ;)
👍