The O in OKR Should Also Trigger Outcomes

(and Not Just Objectives)

Hello everyone! Happy February. Happy Valentine’s Day! Happy New Year (lunar calendar, the year of the ox)!

We’re in the middle of the first quarter in 2021. I’m hearing from friends and colleagues who I’ve worked with in the past that they are busy doing either of two things – one set is doing strategic planning for the year. The other set is gearing up for upcoming performance review cycles. (And there are a few who are doing both – heavens! Who thought of combining both of these in the same quarter? Talk about stress and cognitive overload.) 

I see and hear a myriad of people using Objectives and Key Results (or OKRs) to plan out what they are doing for the new quarter. Using OKRs, which Andy Grove invented at Intel, helps teams and companies proceed with focus and achieve results. You can find several videos and writings out there about OKRs. The talk given by John Doerr at TED Talk is one that many people find useful as an introduction. 

Using OKRs can give you the drive and results. However, I find a lot of teams and companies have not fully harnessed their power. Why? Instead of measuring the outcomes, I see many of them measuring output for their OKRs. The OKRs they write are couched in terms of releasing something on a particular date.

What's up with the feature & date?

Therein lies the problem. A release date doesn’t make an OKR – the impact that happens after a release is the one that does. What did your release do? Did it make your customers happy or sad? Did it help sales sell more, or are sales still scrambling to meet their quotas? 

KR means Key Result – but most people still think that releasing a product by a date is a key result. It is not – it’s an event, a milestone point in time. So how do you create measures that you can track after something releases? A lot of metrics like sales numbers are lagging indicators that can take a few months. 

Impactful OKR Metrics Drive Outcomes

One example of a better metric that one of my teams used is tracking the rate of adoption. We had outcome trackers to see how many of our customers deployed the new software release to their enterprise. We also measured the time they took from the release to actual deployment. 

We also looked at incoming calls to our customer care – if there was an uptick in either issues or negative feedback from our customers. In a nutshell, we created counterpoint OKRs alongside our main OKRs. We do this so that we don’t negatively impact current metrics and make them worse.

We used all these metrics since we could gather on our own. We collaborated with teams who were immediately and directly impacted by our release (like customer service). The end result – we changed the focus away from releases during our quarterly planning and instead concentrated on actual results and impact.

I’d love to hear your thoughts on some of the ideas I presented here. 

Till next time,

-JF

 

Let Them Eat Cake

How do you proceed to eat an entire elephant?

One bite at a time.

Wait – I thought this was a post on cake? Why are you talking about eating an elephant? Let me get to that in a moment. (Unless you made an elephant cake, of course.)

I’m not trying to be facetious here with the eating an elephant analogy. I’m highlighting a struggle that many Agile teams face when they are first going about their Agile journey. How do you break down a project or product delivery into smaller pieces? How might you deliver something that has value to someone every two weeks (or even much less, like three or fewer days for Kanban)? You will need to create small bites – or slices – that someone can use if you want to deliver within your iteration. In other words – it has some value to someone. 

Marie Antoinette and Cakes

Think of it this way – say you’re serving a 5-layer cake (are you noticing copious food analogies here? 😉 ). When you give the cake to someone, you don’t give them the entire bottom layer (since it’s the only one you’ve finished baking, perhaps), and then wait for them to finish before giving the next layer (which hopefully is done by the time they consume the first layer). That’s just too cruel and disappointing, right? No, you give them a slice – a small sliver that contains all five layers and icing.

When you break things down to their smallest value, you have to think about delivering in terms of valuable (consumable) slices (of yummy cake). You might say – yeah – but I still had to bake all five layers, make frosting, put it together, and then get a slice of cake to serve. For cakes, yes – you have to do that. But when it comes to work and what you deliver, do you genuinely need to build every layer?

A Simple Slicing Strategy to Start

One of the most common ways to start slicing is looking at the Happy Path Scenario and building that. We solve for all sorts of flows, where someone starts from point A and ideally ends up on point B. 

For example, if you’re building a website, one small flow you can work on is creating an account. The Happy Path here is landing on a sign-up page, providing an email, providing a password, and voila – you now created an account. 

Does your sign-up page have to be high-fidelity at this point? Not necessarily. When Google first started, all they had was a simple space bar and a Search button.

What if the user entered an email that already existed in the system? Well, you may not have built that yet. As I said, this slicing example is the Happy Path. This example is what I mean by a slice. A slice is a small piece of work that is achievable in a tiny amount of time. And it has value – we enabled a user to create an account. And we can get feedback to help us enhance and modify what slice we potentially will do next: Do we continue with the current cake? Or do we pivot to another part of the cake? Or do we make another cake entirely?)

You might be saying, “Yeah, well, if this was your cake, you still can’t serve one piece to the customer. They are buying the entire cake.” Yes, you are correct. The beauty here is that you’ll find yourself building a bunch of small slices. And when you aggregate these slices together, you will end up with a “good enough cake” that you can serve. It’s valuable enough for people to use. You may not have built everything you wanted yet. But you may have something usable by a small but sufficient amount of people to be valuable enough to release.

Other Slicing Strategies

In the previous section, I already alluded to two other slicing strategies. Were you able to spot them?

The first one that may be obvious is slicing for different error conditions: What if the email account exists? How do we prevent a bot from creating a bogus account? And so on.

The list of error conditions can be infinite. Do you need to build for all of them? Not necessarily. You can now look at prioritization strategies. You might create a variant of an Eisenhower Matrix. Your X-axis can be the probability of the error (low probability to high probability). The Y-axis considers your risk (low risk for the customer to high risk for the customer). You can target the error conditions in the high probability, high-risk quadrant as the ones to build out first. (Or, another alternative instead of prioritization is to build a story map.)

The second strategy may not be as evident with the Google Search example: start at very low fidelity, and build up from there. Using this strategy is most effective when you’re collaborating with UX Researchers and Designers. Working together, you might target a low fidelity scenario for the first slice of cake you’re producing to get feedback on what you’re creating. You’ll be continuously learning from feedback while still building for the whole.

Beyond Software Delivery

You might be saying to yourself, “JF, this is fine and dandy for software product teams, but I’m not in software.” 

The same slicing strategies apply. For example, let’s take a look at HR and onboarding a new hire. From the moment a new hire clicks on Accept on their offer letter to the first week they start, an onboarding flow exists. And you might want to build a flow that is engaging and frictionless for the new hire.

As a startup company, your new hire flow might be somewhat low fidelity. You might start with a simple document that they sign, a questionnaire for what machine and login/email name they’d like so that these are ready on their first day. You might start sketching out a manual process to start and then slowly automate it over time.

As the startup gets bigger, your onboarding flow might become more complicated since your company may be hiring many people. You might want to automate the procurement of laptops, for example, so that it gets automatically configured and delivered to the new employee in these times of pandemic.

And here’s the third slicing strategy that you can use. When you have a long and complicated flow (whether existing or new), you can break down the long, complicated flow into smaller flows that you can target. 

It’s like driving from San Francisco to New York. You can do it one straight shot, but that’s going to be too tiring and leave you with no time for enjoyment. You can break the trip down and have some pit stops for exploration and fun. You can design your trip to stop in Arizona to explore the Grand Canyon. Then head on over, making a left at Albuquerque so that you can grab some good ol’ barbeque in Kansas. 

You can do something similar with HR onboarding. You can concentrate on a subsection of the flow and fix problems. You may want to modify it. You can also add new, additional flows to make the whole onboarding process better by adding one small flow at a time.

These are just a few quick ways to break down things into deliverable pieces of value during each iteration. There are many other slicing strategies that people have written about, which you can find through Google. I typically start with these strategies when I’m coaching teams that are new to Agile. Hopefully, these few suggestions can help you on your path to getting comfortable with breaking down items that fit in an iteration.