Optimizing Post-Covid Return to Work Strategies – A Conjoint Case Study?

Click here to access the slides

Click here to access the slides


Transcript of recording with Abby Lerner and Megan Peitz – generated automatically by HappyScribe which means it will be about 80% accurate – if you spot confusing errors, please email ray@new-mr.com. The timestamps are included to help you jump directly to a point of interest.


[00:00:08.060] – Abby Lerner

Hi, everyone, and welcome to our webinar today, I’m Abby Lerner. I work at NewMR and I’ll be joined by my colleague and founder of numerous Meghan Pipes. We’re excited to talk to you today about optimizing post covid return to work strategies using conjoint a case study. So we all know that covid-19 of has had quite a bit of impact over the world the past year, restrictions have eased in some parts of the world, including here and businesses who’ve had employees working from home for the past few months.


[00:00:40.340] – Abby Lerner

They now have decisions to make about do we bring them back and require them to come back to work office or can they keep working from home? Some companies have taken a stance already. I’m lucky to work at a company that is fully remote. So this is not a decision that my employer had to make. Facebook is not a remote company, but they have offered their workers at any level to request to be able to work from home full time if they like.


[00:01:05.150] – Abby Lerner

And on the other end of the spectrum, you have places like Morgan Stanley who fully expects by Labor Day that most employees have already kind of made their way back into the office. And at Apple, they expect employees to be back to in the office at least three days a week come September. But as researchers, we know that employers shouldn’t be making these decisions in a vacuum. It’s not so simple as flipping a switch and picking a day and having everybody come back in this era or any truly of caring about employee needs.


[00:01:34.100] – Abby Lerner

We need to be cognizant of what’s going to make them feel safe. And it’s going to be a good, healthy environment for employers, employees to come back and feel good about doing work and being back in the office. Now, not all businesses have this option or have had this option all along in terms of remote work. But we hope that the ones that do are going to consider listening to the employee to understand what might work best for them.


[00:01:59.840] – Abby Lerner

Just as researchers we listen to, we pride ourselves on listening to the voice of the customer. And that’s some of the value we bring to the table. It’s important now that businesses listen to their employees. So one approach to go about this, when you want to hear from your customer or from your employee is to do a simple approach and do a simple survey and say, what are the things that would make you feel comfortable about returning to work in the office, select all that apply.


[00:02:28.280] – Abby Lerner

And you could see here we have some great options like hand sanitizer stations and wearing masks in the office and things like deep cleanings and whatnot. Well, you can see here it’s not an accident that every box is checked because thinking through these, all of them sound pretty good and would make me feel more comfortable about returning to work. But how is that going to help you? Well, at NewMR, yes, we would say one better approach might be using conjoint analysis, because that’s going to help us get deeper insights about what we could really do here.


[00:03:03.830] – Abby Lerner

What conjoint does for those of you who are unfamiliar is, for instance, let’s start with a cell phone. We want to break down our product and service into different components or parts. I might make a decision about linking cell phones just based on brand. However, at a certain price point, that brand becomes less important to me, or I might be willing to pay more for a special camera. All these things can be shown in different combinations, and then I’m forced to choose which combination I like best, meaning I’m making.


[00:03:36.470] – Abby Lerner

I’m making tradeoffs many different times throughout this task. And as a result, we’ll have models that can provide terms of reference and scores for all possible combinations. So in this case, what what this does is it helps us to answer questions about what product or service is most appealing to people. What price will I switch to a competitor? As I mentioned, there might be a certain price that I do that should be launch a higher end product. Well, will we be cannibalizing our own cells if we do that?


[00:04:10.230] – Abby Lerner

What we’re here to tell you today is it’s not just about products and services and buying and selling, we could use conjoint any time we’re forcing someone into a trade off. In this case, what we’re set out to do today is use this in the context of post covid return to work strategy. So we’ll be able to answer from our research that what is the optimal number of work from home is to offer. How does our return to our strategy compare to other companies?


[00:04:40.650] – Abby Lerner

What’s going to impact employees the most, whether we offer it or don’t offer it? So we’re going to set out to answer these questions today. And I’m excited to kick it over to Megan, who will walk you through the methodology and take you through the results.


[00:04:55.250] – Megan Peitz

Thanks, Amy. Let’s talk about the sample in this research, we conducted a quantitative online survey earlier this month among about 600 respondents in the US. We make sure that they were at least 18 years of age and currently working full time or part time. And the niche about this sample, as is that we wanted people who had worked at an office prior to covid-19 and then had worked from home during covid-19. We felt like this was the best way that we could understand how people actually wanted to return to work, given that they’d been at home for so long.


[00:05:31.550] – Megan Peitz

There’s two other call outs that I wanted to note here specific to. If you don’t do a content analysis, there should be other things that you consider the first one being. Our survey was rather short. Only eight minutes to complete and people open ended. Lee responded that they really liked that. We only asked questions that felt relevant to the survey. The second thing is that forty three percent of our respondents took the survey on a mobile device. Now you could see the screenshot here where our content actually renders to look good on a mobile device.


[00:06:00.590] – Megan Peitz

But please, please make your surveys in twenty twenty one mobile compatible because people are taking their surveys on their mobile devices. Now, just a brief flashback to what our exercise is going to look like, we’re going to show respondants three return to work strategies at a time, and we’re going to ask people which one would make them most comfortable returning to work. Post covid-19 and see our list of attributes on the left hand side. And then within each of those concepts, we have our levels, which we’ll talk about in just a second.


[00:06:37.150] – Megan Peitz

And the respondent is going to go through this task multiple times in our survey. They are going to go through at eight different times and then pick one environment on every screen. We’re also going to leverage what’s called a dual response here, because if you think about it, there are going to be some people that aren’t ready to return to work no matter the environment that you put them in. So after they tell us which one feels most comfortable to them, they then answer a follow up question that says, based on the office layout you chose above, would you actually return to work under these conditions, yes or no?


[00:07:10.600] – Megan Peitz

I really like the spill response approach and I actually use it in my KPG type conveyance or product optimization contracts or service optimization contracts as well, because this way I at least learn what they like about the return to work environment. But I don’t force them to say that they will go back to work. So let’s take a quick look at our attributes and levels, we’ve got our attributes in the left hand column and then our level spread throughout country recommendations are five to seven attributes, two to seven levels within each.


[00:07:43.790] – Megan Peitz

So we’re sticking to those rules. And you can see each of the concepts on the previous page choose one level from every attribute to create one of those return to work strategies. So we’ve got a number of days in the office, a week ranging from one all the way to five capacity of the desks that are going to be occupied having one hundred percent, meaning everyone’s back down to as little as twenty five percent. And then some other measures that will help employees feel safe when they return back to the office.


[00:08:15.010] – Megan Peitz

Now, after we conduct the research and collect all the data, we’re going to build a model and the output of that model, our utilities, and we get utilities for every level of every attribute. And if we build a hierarchical Bayesian model, we actually get those utilities for every respondent as well. Now, utilities are going to be the crux of your analysis, but at Numerics, we don’t actually show them in the report because what we do is we transform them into shares of preference.


[00:08:42.910] – Megan Peitz

And those shares of preference can be used to simulate different scenarios in a market simulator run optimization searches or what we really like is sensitivity analysis, which is what we’re going to show you here. We have a whole nother presentation on why and why not to use utilities. But the main outcome is, is that they’re just values, they’re interval scale data. They’re typically rescale to each attribute. So you can’t really compare the score for one level of an attribute to the score for another level of another attribute.


[00:09:13.780] – Megan Peitz

And because their interval scales, you can’t do ratio operands. So a utility of a 10 is not twice as good as a utility bill five. And people automatically want to do those with those numbers. So once you transform those utilities into shares, we can do that. And so that’s why we’re going to show you is starting off with a market simulation center, then sensitive analysis and then optimization searches. Now, let’s assume I’m the employer and I want my employees to return to work and be back in the office full time.


[00:09:45.910] – Megan Peitz

So I’m going to say, hey, our return to risk strategy is being in the office five days a week. One hundred percent of the desk, you’re going to be occupied. But we’re going to put some appropriations in to make sure you feel comfortable like partitions. We’re going to close the meeting room. We’re going to close the kitchen. And we’re we’re going to require masks for employees at all times. I can understand from the market simulator what proportion of employees would actually return to work under those conditions based on how they answer the contract exercise.


[00:10:14.290] – Megan Peitz

However, in this scenario, only forty one percent of my employees would actually want to return to the office, given my plan. So now if I’m the employer, I’m sitting there thinking, yikes, what do I need to do in order to get people to return to the office? And so I can run a sensitivity analysis with my contract data. And what that is, is we basically take that what we call base case scenario or that employer plan on the last screen and hold all of the levels constant except one.


[00:10:49.000] – Megan Peitz

And we change that one level at a time. So you’ll kind of see a lot of points hovering around that. Forty one percent, which is exactly what we showed on the last slide in terms of the percentage of employees that would actually go back to work, given the employer’s plan of record. But if we just change the number of days of office from five days a week to four days a week, we can see that people’s interest overall increases from about forty one percent to about forty six percent.


[00:11:16.600] – Megan Peitz

And then we go up even more. If we drop the number of days in the office, down from four days a week to three days a week and so on and so forth. So therefore we can see that biggest improvement if we only wanted to change. One thing about our plan would be reducing the number of days in the office from five to two. We can also see that if we look at the mass requirement that has quite a bit of variability in it, we might want to go there next.


[00:11:42.460] – Megan Peitz

If we are if we want to change a little bit more to the plan and then we might want to look at capacity and change that from one hundred percent down to something else. But it looks like the remaining attributes, not a whole lot of variability there. However, it looks like we really don’t want to remove the partitions because that’s going to have a negative effect on how many people want to return back to work. So in general, it seems like people want partitions between the desks as well.


[00:12:13.080] – Megan Peitz

Now, I want to double click really quick into the number of days in the office, actually, we know that we want people in the office five days a week if we’re the employer. But the employees have the highest likelihood of returning to the office when there’s two days a week. But we can see there’s minimal difference between two days a week and three days a week based on the employee. So if I’m the employer, I’m going to say, OK, I’ll give you three days of the in the office a week, not all the way to two.


[00:12:41.070] – Megan Peitz

And I’ll probably get almost the same amount of interest here. I’m optimizing on both the employer preferences as well as the employee feedback. All right, now we can use optimization algorithms to search through every single possible combination of the results to figure out what is the optimal return to work strategy that’s going to make the most number of employees willing to return to the office. So the algorithm does its magic. And we find out that 81 percent of employees in this survey would return to work if we offered them two days in the office, limited to 50 percent capacity master required for those who aren’t.


[00:13:23.110] – Megan Peitz

That’s we’ve got temporary partitions. Meeting rooms are open, but single occupancy, no communal dishware, but the kitchen is open and we’re going to throw in a professional deep cleaning every week. So if I’m the employer, here’s my answer of the optimal strategy to get my most of my employees back in the office and and happy. What’s even cooler about conjoint analysis, and yes, I say cooler because I’m married and I love this stuff, but we can actually conduct cluster analysis on the conjoint responses.


[00:13:55.370] – Megan Peitz

And when we do that with this data, we actually find three segments of clusters in the data. The first segment is whom I’m going to call the The Work from home folks. And they make up about a third of our total sample. And these guys are, you know, basically work from home the entire time that they were in Koven. And they believe that for the most part, their productivity has increased or stayed the same. And so therefore they want to continue working from home.


[00:14:25.070] – Megan Peitz

Then there’s this hybrid group of people who are comfortable going back to work and actually prefer to work from home or to work at the office more days than not. And they may or may not feel like their productivity has stayed the same and maybe even decreased while working remotely. Therefore, they want to go back into the office, but probably not five days a week. And then we’ve got segment three that makes up about 16 percent of our sample. And these folks, no matter what you do, they’re just not ready to go back to the office.


[00:15:01.040] – Megan Peitz

They’ve been working from home during covid. Their productivity has at least increased more the segment two. But you know what? They’re just not ready. Whether this is fair, whether this is some other factors. Who knows? But they’re they’re just not ready. So what we can do is we can take these three clusters and then go back to that optimization search and actually figure out which environment is going to work best for each of these segments. And that is what we do on the slide here, so we look at segment one, the work from home folks, we look at their optimization results and we see that ninety seven percent of them will return to work if they were only required to be in the office one day a week with 50 percent capacity.


[00:15:47.490] – Megan Peitz

And on that one day, they don’t want to wear masks that we can look at segment to our hybrid folks who really want to get back in the office and see that. Ninety four percent of them would return if they were in the office four days a week with 50 percent of desk occupied and masks are going to be required for non vaccinated employees. And then we have segment three, the folks who are pretty apprehensive about returning to work. And even if we give them the optimal combination of attributes and levels, still less than 50 percent are going to feel comfortable going back to work.


[00:16:19.800] – Megan Peitz

But if they do, they want to be in the office two days a week. And again, 50 percent of the desks occupied with masks being required for those who are not vaccinated. Now, based on those optimization results, we can say, hey, we’re not going to design just one return to work strategy, but rather two, and we’re going to have people who can come into the office only one day a week, or you can choose to come into the office four days a week.


[00:16:48.540] – Megan Peitz

But we’re going to take some of that other information that seems to be pretty common. And plus, we can’t really design multiple different office strategies. So we’re going to say for everybody, 50 percent of the desks are occupied. Masks are required for non vaccinated folks. There’s partitions between desks. Meeting rooms are available for single occupancy. The kitchen is open, but no communal dishware and will give you a deep professional cleaning every other week. When we do that, we look at the results on the right and we see forty three percent of people would opt in to one day a week.


[00:17:20.910] – Megan Peitz

Forty one percent would opt into four days a week. And there’s still about 16 percent who wouldn’t feel comfortable with either of these options. We can also look at those results by segment and see they line up pretty perfectly with our optimization searches. Right. Segment one opts into the one day option segment to opt into the four days week option. And Segment three has a hard time going back to work in either case. But some will go with the one day and some will go with the four days should they choose to return to work at numerous.


[00:17:54.000] – Megan Peitz

We love running cluster analysis on Congi data because we think it’s so incredibly valuable, right? You’re getting multiple strategies to appeal to the largest group of consumers, not just one answer. That’s a I guess this one’s the best of luck. So we highly recommend always looking at clustering results based on your content utilities. Now, I’m going to turn it back to Abby to talk about the somewhat.


[00:18:21.150] – Abby Lerner

So what does this all mean? But numerics, we think conjoint analysis does a really great job at helping us get at discriminating preferences among respondents. We could do this at the aggregate level, the subgroup level, at the individual level. We believe that researchers should definitely strongly consider conservate when trying to understand anything where tradeoffs need to be made between different alternatives or when we have the likelihood for nondiscrimination, non discriminating data. Perfect example was what we showed you earlier, where everything is good, everything would be helpful, et cetera, et cetera.


[00:18:57.210] – Abby Lerner

You know, ultimately, we say and we’re saying here today, when respondents are forced to make difficult tradeoffs, we will learn what they truly value by using this approach. Now, one thing you might say is there could be pushback or a little buzz around the industry that is hard for respondents, that it makes the survey wrong. You’re going to have to much drop off. Well, we’re here to show you that within this survey, we put an open ended question after our exercise to gather some feedback from respondents.


[00:19:26.290] – Abby Lerner

You can see some of the quotes here that they enjoy doing that. Survey questions require careful thinking. That is easy to answer. In short, that thanks, nice survey. Enjoyed the layout with the choices. This was a fun and relevant survey. So whether there’s a conjoin or not, it’s important to have a short, relevant, well-designed survey. And we believe with Conjoint, if you go through all those checks and do good and you have the right audience feeding through it, that it’s super valuable and worth the time to do it right.


[00:19:55.270] – Abby Lerner

And that’s why innumerous, we often look to that approach and we hope what we saw today, two to you will make you consider it four different scenarios in the future. So with that being said, thank you. And now we’re excited to talk to you and get some Q&A.