How would you prioritize new product features for Facebook?

An Execution Question for Product Management Interviews

How Are You Being Evaluated

Prioritizing product initiatives and product features is at the core of product management. This interview question tests whether you have a clear process of prioritization and whether you can boil down your prioritization to numerical comparisons.

  • Do you have a method for prioritizing features?
  • Can you formulate how to assign numerical values to different levels of priorities?
  • Can you boil down your prioritization to a numerical comparison?
  • Do you explain your assumptions clearly behind the values used in your formulas?
  • Are you organized and structured in your analysis?

Answer Structure

  • Begin by describing your prioritization process. Drawing a diagram will help the interviewer follow your explanation better.
  • Apply your process to an existing product to demonstrate how it works, or ask the interviewer to choose an example.
  • Before moving to the next step in your prioritization process, tell the interviewer which step you just concluded and which step will be next.
  • Use a benefit vs. cost analysis or ROI when prioritizing initiatives.
  • Explain your reasoning behind each term when formulating equations.
  • Formulating all equations before assigning values will keep your work more organized.
  • Create a two column table to evaluate your equations numerically. Use the left column to write assumptions and numerical facts about the terms used in your equations. In the right column, replace your equation terms with numerical values.
  • Calculate benefit/cost (ROI) for each initiative. Check if the winning initiative meets the result expected from the Objectives & Key Results (OKR). State which initiative is the winner and continue prioritizing features for that initiative.
  • Use a scorecard method for prioritizing features.
  • Summarize your analysis and provide your recommendations of what initiatives and features to build.

Answer Example

INTERVIEWEE: To prioritize new product features for Facebook, I would start with the OKRs the business wants to achieve for the period of interest. Once I am clear on that, I would assemble all product stakeholders to generate ideas for product initiatives that align with the OKRs. By initiatives, I mean high-level projects, like a new mobile app, or a new platform feature, like Facebook Live for example.

After assembling a list of initiatives, I would prioritize them based on a cost/benefit analysis. Following the initiative selection process, I would start determining which kinds of features or processes to develop for that initiative. If the initiative is a new platform product, like Facebook Live, then I would put together a list of features to design and develop for that product. I would use insights from UX research, feedback from users, customer support, sales, and any other secondary data source to write user stories. And these user stories will drive the ideation of new features, which I group into themes. Themes help me organize features and identify dependencies.

After compiling a list of features and grouping them into themes, I would prioritize their implementation using a scorecard that evaluates the features based on their contribution of positive impact vs. effort. Attributes that denote positive impact could be, for example, must have features, frequency of use, ability to leverage existing technology. And, attributes that denote effort could be feasibility, engineering complexity and others. The types of attributes I select for the scorecard depend on user needs and alignment with business objectives (or OKR). Using a Value Point System between 1 and 10, I would assign a number to each feature’s attribute. Attributes whose higher values are considered “good” get added, and attributes whose higher values are considered “detrimental” are subtracted. Next, I would calculate a weighted sum of the attribute values for each feature to arrive to a score number. This score number is what I would use to prioritize the features.

Let me summarize this prioritization process with a flow chart:

Flow chart of the prioritization process
Flow chart of the prioritization process

INTERVIEWER: Okay. Now imagine that you are asked to prioritize between Stories, Live, and Marketplace as new features for the Facebook mobile app? How would you use your process to do this?

INTERVIEWEE: I would start by asking what the Objectives & Key Results are?

INTERVIEWER: Let’s say Facebook wants to increase ad revenue by 10% this year.

INTERVIEWEE: Okay. Stories, Facebook Live, and Marketplace are not small features. They are products. So, I consider them initiatives. Do you agree?


INTERVIEWEE: Okay. First I would determine if these initiatives align with the OKR. And they do. The initiatives aim to increase time spent on the platform, and time spent is directly correlated with ad revenue. Facebook Stories aims for the user to return or spend more time on Facebook by viewing or creating stories. Facebook Live wants users to spend time watching real-time videos, such as live sports. And, Marketplace aims for users to spend time on Facebook searching for items they would otherwise find using Google, Craigslist or another tool.

Now that we have established that these three initiatives align with Facebook’s OKR, the next step is to prioritize the initiatives based on benefit over cost. For all three initiatives, we can use ad revenue as a proxy to measure benefit. To measure cost, we can use engineering wages as a proxy.

To quantify benefits, I will formulate some equations to estimate ad revenue and engineering labor for each initiative. Then I will compare the initiatives based on benefit/cost ratio, which is ROI, to prioritize them. Does this sound reasonable?

INTERVIEWER: Sure. Go ahead.

INTERVIEWEE: Okay, let’s start by talking about the different types of ads Facebook offers and how we can use them to estimate ad revenue. Facebook has various types of ads, but the main ads are: impression ads, click ads, app install ads, page link ads, and video ads. Prices for these ads fluctuate depending on various factors, like bid value, and quality of the ad. So, I am going to use an average price for ad types to simplify my estimates. Do you agree?


INTERVIEWEE: Ok. Based on my knowledge of price fluctuations for Facebook ads, the average price of an ad impression is about $0.001 or $1 CPM; the average price of a click or CPC ad is about $0.2, and the average price of a video ad is about $0.01. The price of other ads, like app install and page link, are close to the click ad’s price. To simplify my analysis, I will group these ads under the same click-ad bucket. The three types of ads I need to consider are: impressions, click ads and video ads.

Now, I would like to introduce equations that will help me calculate ad revenue for each of the three initiatives.

Let’s start with Facebook Stories.

How does Stories generate additional ad revenue for Facebook? When a person gets notified that a friend posted a new story, this may prompt the user to open or return to the Facebook application. So, Stories may contribute incremental ad revenue by displaying impressions or clicks ads when the user opens the app. Here are two equations to estimate impressions and click ad revenue for Stories.

(1) Stories impression ad revenue

Impression Ad Revenue
= ( DAU x ( %Stories users )
x ( %creators )
x ( #stories / week )
x ( %stories that cause user to open Facebook app )
x ( #impressions/story )
x ( $CPM / 1000 )
x ( #weeks / year )

(2) Stories click ad revenue

Click Ad Revenue
= ( DAU )
x ( %Stories users )
x ( %creators )
x ( #stories / week )
x ( %stories that cause user to open Facebook app )
x ( #click ads / story )
x ( CTR )
x ( $CPC )
x ( #weeks / year )

Here’s an explanation for these equations:

  • There is a number for Facebook’s mobile daily active users (DAU).
  • A percentage of these users use Stories (%Stories users).
  • Of these Stories users, a percentage creates stories (%creators).
  • These creators produce some number of stories per week (#stories/week).
  • And, only a percentage of these stories cause other users to reopen the Facebook app (%stories that cause a user to open the Facebook app).
  • A number of impressions are displayed when the users re-opens the Facebook app to view a story (#impressions/story).
  • Each impression generates CPM/1000 of dollars.
  • And, since we are calculating revenue for a year, we multiply by the number of weeks in a year.
  • To calculate revenue from click-ads due to Stories, we can use the same equation except we replace the number of impressions with the number of click ads. And, replace CPM/1000 with CTR x CPC.

Let’s move on to the equation for Facebook Live.

(3) Live ad revenue

Facebook Live generates ad revenue from video ads shown while users watch videos. Here is an equation to estimate ad revenue for Facebook Live:

Video Ad Revenue
= ( DAU )
x ( %video users )
x ( #minutes / day )
x ( #video ads / min )
x ( $cost of video ad / video ad )
x ( #days / year )

Let me explain the equation in more detail.

  • A percentage of daily active users (DAU) of Facebook mobile app watch live videos(DAU) x (% video users).
  • These live video watchers watch some minutes a day (#minutes/day)
  • During these watch minutes viewers are exposed to a number of video ads (#video ads/min).
  • And, each video ad has a cost to advertisers ($cost of video ad/video ad).
  • We multiply by the number of days in a year to get the total ad revenue per year.

Ok. Let’s move on to the equations for Marketplace.

While using Marketplace, a user does not see ads. But, ads are shown to users when they log into facebook and access Marketplace. And, perhaps after using Marketplace, they might stay to read their newsfeed. So, ad revenue attributed to Marketplace can come from these instances.

In these situation, I will use these two equations to estimate ad revenue from impressions and click ads.

(4) Marketplace impression ad revenue

Impression Ad Revenue
= ( DAU )
x ( #Facebook sessions / day )
x ( %sessions started due to Marketplace )
x ( #impressions / session )
x ( $CPM / 1000 )
x ( #days / year )

(5) Marketplace click ad revenue

Click Ad Revenue
= ( DAU )
x ( #Facebook sessions / day )
x ( %sessions started due to Marketplace )
x ( #click ads / session )
x ( CTR )
x ( $CPC )
x ( #days / year )

The details for ad revenue from impressions are:

  • Daily active users (DAU) start a number of session per day (DAU) x (#Facebook sessions/day).
  • Of these sessions, only a percentage are started by users searching in Marketplace (%sessions started due to Marketplace).
  • During each session the user is exposed to a number of impression ads (#impressions/session).
  • The cost-per-thousand impressions charged to advertisers is CPM, so the cost for each impression is (CPM/1000).
  • To get an estimate for the entire year, we multiply by the number of days in a year.

We can use the same equation to estimate click ad revenue for Marketplace, except we replace (#impressions/session) x ($CPM/1000) with (#click ads/session) x (CTR) x ($CPC).

Now, I will make some assumptions to replace the terms with numerical values.

(The interviewee creates a two column table. The column on the left is to state assumptions and known facts. The right-hand column is to show how to apply these assumptions and facts to the calculations.)

Stories ad revenue

Stories ad revenue
Stories ad revenue

Facebook Live ad revenue

Facebook Live ad revenue

Marketplace ad revenue

Marketplace ad revenue

So, it looks like Stories would bring in $127M, Facebook Live $18B, and Marketplace $648M.

Now, let’s look at the cost of developing these platform features to complete our cost benefit analysis.

I will use the cost of engineering labor as a proxy. Of the three platform features, I think Marketplace is the easiest to develop and maintain. Marketplace requires uploading and storing static photos, presenting product information, and enabling users to message sellers about the products they are interested in. There is no video streaming involved, nor functions to transact purchases. So, I would guess that it takes a team of five engineers to build and maintain. Assuming that each engineer’s salary is about $200K a year, that means Marketplace costs $1M a year.

Between Facebook Live and Stories, I think Stories is less complex to develop and maintain. It involves uploading videos, sharing, annotation, and notifications but no live streaming like Facebook Live. Stories are viewed asynchronously. So, I am going to guess that it takes a team of 10 engineers to build and maintain which is 10 x $200K for a total cost of $2M a year.

The real-time nature of Facebook Live makes the engineering requirements more challenging. Facebook Live needs broad and fast network bandwidth, optimized frame buffering, and be synchronous with various functions like audio and feed conversations. So, I am going to estimate that it takes a team of 30 engineers to maintain Facebook Live which is 30 X $200K for a total cost of $6M a year.

Now, let’s compare these three initiatives with their benefits and costs with an ROI metric or benefits/cost ratio.

Comparison of the three initiatives with their benefits and costs with an ROI metric or benefits/cost ratio.

Facebook Live is the platform feature with the highest ROI and would be the one I would select to build. But, we also need to check if this option is likely to meet the mandated OKR. The OKR was to increase ad revenue by 10% YoY. Given that Facebook mobile revenue is expected to be about $30B in 2017, then $3B (10% of $30B) would be the desired incremental revenue. We estimate $18B for Facebook Live ad revenue. This is $15B more than the $3B YoY incremental ad revenue desired. Therefore this results favors Facebook Live.

The next step is to start working on the product features for the Facebook Live offering. As I mentioned before, this entails thinking about user needs, creating user stories, and prioritizing them. Would you like me to demonstrate how I would do this for the Facebook Live offering?

INTERVIEWER: No, we are running out of time. However, could you summarize your analysis to wrap up?

INTERVIEWEE: In summary, I explained my process for prioritizing features. It starts with OKRs, and is followed by listing initiatives that are aligned with the OKRs, prioritizing these initiatives based on benefit vs. cost or ROI, listing features for the winning initiative, grouping features into themes, and finally prioritizing features using a scorecard system. I applied this process to the prioritization of three Facebook platform features: Stories, Facebook Live and Marketplace. In this example, these platform features are product initiatives, so I prioritized them as such. In this example, Facebook Live was the winning option with an ROI that was multiple times better than Stories and Marketplace. Therefore, my recommendation would be to build Facebook Live first.