Benchmarcs

AI-driven analytics for the funeral industry
Benchmarcs

Challenge

Collecting and displaying funeral home sales metrics that would help owners know where their business is successful and where they could improve

Solution

Build a failure-aware network of spiders to crawl public obituaries, use natural language processing to parse location data, and display it all in a simple, readable interface

Results

A product that gives funeral home owners a simple way to see their business compared to their competitors’ and where future opportunities might be

Massive, automated data collection

To build an analytics tool to show sales data across multiple funeral providers, we needed data… and the only way to get it was to collect it ourselves. To populate our database, we built a failure-aware network of spiders to crawl millions of pages a month. Once we had that, we could display it in a responsive front end where users could slice and dice the data and better understand how they stack up against the competition.

Visit the site

Services

  • UX
  • Design
  • Advertising Strategy
  • Content Strategy
  • CMS UX

Technologies

  • Marionette.js
  • Python/Django
  • Elasticsearch
  • AWS EC2
  • Nginx
App Screenshots

Data Collection

Failure-aware spiders crawling millions of pages a month

Websites get updated all the time, but we still needed Benchmarcs to stay up to date with the latest obituaries posted. To populate and maintain our database, we built a failure-aware network of spiders that crawls millions of pages a month. As websites and data sctructures change, our network adapts itself to crawl the correct data or throws warnings our way so we can be up and running again as soon as possible.

Benchmarcs Trends page

Deep Learning

Neural networks that read obituaries the same way people do

Collecting millions of obituaries was only half the battle. We needed a way to turn them into usable, structured data. And the only way to do that was to teach a neural network how to read through obituaries to parse locations, dates, and more. What this means is that our data gets populated in nearly real time. As obituaries get posted, we’re able to crawl, parse, and display them with only a small delay.

Benchmarcs NER Tagging

Interface Design

Converting millions of points of data into coherent visuals

After refining signal out of noise, we wanted Benchmarcs to deliver customers a more insightful look at their data, so we tailored their interface to provide coherent data visualizations that generate actionable metrics.