Estimate Highway Billboard Impressions Using Traffic Data

Quick Overview

Evaluates estimation of highway billboard reach, impressions, and frequency from traffic data. Strong answers define metrics, convert passings into viewable impressions, estimate unique reach, and communicate uncertainty.

Estimate Highway Billboard Impressions Using Traffic Data

Company: Pinterest

Role: Data Scientist

Category: Statistics & Math

Difficulty: medium

Interview Round: Onsite

##### Scenario Out-of-home advertising campaign wants to estimate reach. ##### Question How would you estimate how many people actually saw a highway billboard and derive the total number of impressions it generated? ##### Hints Segment the population, collect traffic counts or mobile-location samples, model exposure probability, then multiply by average views per person.

Quick Answer: Evaluates estimation of highway billboard reach, impressions, and frequency from traffic data. Strong answers define metrics, convert passings into viewable impressions, estimate unique reach, and communicate uncertainty.

|Home/Statistics & Math/Pinterest
Pinterest logo
Pinterest
Jul 12, 2025, 6:59 PM
mediumData ScientistOnsiteStatistics & Math
31
0

Estimating Highway Billboard Reach and Impressions

An out-of-home advertising team wants to estimate reach and impressions for a single highway billboard over a specified time window, such as one week.

Design a defensible estimation approach suitable for a data-science interview.

Constraints & Assumptions

  • Define the billboard face, direction of travel, time window, and whether the board is static or digital.
  • Distinguish passings, viewable impressions, reach, and frequency.
  • Make assumptions explicit for traffic volume, vehicle occupancy, mobile-device representativeness, and visibility.
  • Discuss uncertainty rather than presenting the estimate as exact.

Clarifying Questions to Ask Guidance

  • What is the campaign time window and share of voice for the billboard?
  • Which lanes and travel directions can see the billboard?
  • Is the objective total impressions, unique reach, or frequency distribution?
  • What data is available: DOT counts, mobile location samples, camera counts, ad-server logs, or survey calibration?

Part 1 - Define Metrics and Scope

Define impressions, reach, frequency, and the measurement scope.

What This Part Should Cover Guidance

  • Impressions as expected ad views, not merely vehicles passing the location.
  • Reach as unique people with at least one exposure during the window.
  • Frequency as impressions divided by reached people or as a distribution.
  • Scope choices for direction, daylight, weather, digital rotation, and audience segment.

Part 2 - Estimate Passings and Exposure Probability

Describe how to convert traffic into expected impressions.

What This Part Should Cover Guidance

  • Segment traffic by time, direction, lane, vehicle type, speed, and weekday or weekend.
  • Combine traffic counts with vehicle occupancy and a visibility or attention probability.
  • Account for digital share of voice, view angle, time in view, obstructions, lighting, and weather.
  • Produce a formula such as sum of passings times occupancy times viewability times share of voice.

Part 3 - Estimate Unique Reach

Describe how to estimate the number of unique people exposed at least once.

What This Part Should Cover Guidance

  • Use mobile-panel IDs, origin-destination data, commute recurrence, or modeled duplication rates.
  • Correct for panel bias and device-to-person scaling.
  • Avoid double-counting repeat commuters across days.
  • Report uncertainty intervals and sensitivity to duplication assumptions.

Part 4 - Validate and Communicate the Estimate

Explain how you would validate the model and present results.

What This Part Should Cover Guidance

  • Compare against camera counts, independent traffic sources, mobile exposure vendors, or brand-lift surveys.
  • Run sensitivity analysis for occupancy, viewability, and repeat-rate assumptions.
  • Communicate point estimates with confidence ranges and caveats.
  • Separate operational measurement limitations from decision recommendations.

Follow-up Questions Guidance

  • How would the method change for a digital billboard with rotating creatives?
  • How would you estimate incremental reach across multiple billboard locations?
  • What bias would you expect from mobile-location data, and how would you correct it?
Loading comments...

Browse More Questions

More Statistics & Math•More Pinterest•More Data Scientist•Pinterest Data Scientist•Pinterest Statistics & Math•Data Scientist Statistics & Math