diff --git a/_posts/2019-02-07-calculating-customer-lifetime-revenue.md b/_posts/2019-02-07-calculating-customer-lifetime-revenue.md index d4c679e..53d0f16 100644 --- a/_posts/2019-02-07-calculating-customer-lifetime-revenue.md +++ b/_posts/2019-02-07-calculating-customer-lifetime-revenue.md @@ -10,7 +10,7 @@ team: Applied Research Why LTR? (Lifetime Revenue) -When I joined the[ Data Science team](https://www.scribd.com/about/data_science) at[ Scribd](https://www.scribd.com/), my first project was to update a dashboard that our Marketing team uses to make ad-buying decisions. For all of our acquisition sources, this dashboard displays key indicators that we use to assess the future value of a new group of subscribers like: bill-through rate(% converting from a trial to a paid subscription), 1st day cancellation, etc. While these indicators are valuable and give our team guidance on where we should be investing our marketing dollars, they aren’t explicit signals of the future value of subscribers. You can imagine a high bill through rate not resulting in a high value group of subscribers if most of them cancel after using Scribd for only 1 month. +When I joined the[ Data Science team](https://www.scribd.com/careers) at[ Scribd](https://www.scribd.com/), my first project was to update a dashboard that our Marketing team uses to make ad-buying decisions. For all of our acquisition sources, this dashboard displays key indicators that we use to assess the future value of a new group of subscribers like: bill-through rate(% converting from a trial to a paid subscription), 1st day cancellation, etc. While these indicators are valuable and give our team guidance on where we should be investing our marketing dollars, they aren’t explicit signals of the future value of subscribers. You can imagine a high bill through rate not resulting in a high value group of subscribers if most of them cancel after using Scribd for only 1 month. To truly solve this problem and empower our Marketing team to make data-driven decision about their spend, we needed a way to calculate the future value of the subscribers. Enter Lifetime Revenue (LTR)! Below is a fake version of our marketing dashboard that includes our LTR column, “Expected Lifetime Revenue Per New Signup”. diff --git a/_posts/2019-03-04-experiments-with-seq2seq.md b/_posts/2019-03-04-experiments-with-seq2seq.md index ff10bb4..88e47df 100644 --- a/_posts/2019-03-04-experiments-with-seq2seq.md +++ b/_posts/2019-03-04-experiments-with-seq2seq.md @@ -104,4 +104,4 @@ In conclusion, we modeled the dependency of seq2seq on data as a function of inp We found that the relationship to be roughly linear. (It’s possible that at some point the relationship becomes no longer linear, but we didn’t encounter that within the parameters of this experiment.) -This result can give you a very rough idea of how much data you will need for your seq2seq project. At Scribd, we use or have explored using seq2seq for a variety of projects, including query parsing, query tagging, and spelling correction. If working on one of these projects is something you think you might be interested in, go ahead and give us a holler at [https://www.scribd.com/about/data_science](https://www.scribd.com/about/data_science) +This result can give you a very rough idea of how much data you will need for your seq2seq project. At Scribd, we use or have explored using seq2seq for a variety of projects, including query parsing, query tagging, and spelling correction. If working on one of these projects is something you think you might be interested in, go ahead and give us a holler at [https://www.scribd.com/careers](https://www.scribd.com/careers) diff --git a/_posts/2019-08-28-real-time-data-platform.md b/_posts/2019-08-28-real-time-data-platform.md index c90bcf8..b12d6f9 100644 --- a/_posts/2019-08-28-real-time-data-platform.md +++ b/_posts/2019-08-28-real-time-data-platform.md @@ -16,7 +16,7 @@ team: One of the harder parts about building new platform infrastructure at a company which has been around a while is figuring out exactly _where_ to -begin. At [Scribd](https://www.scribd.com/about/engineering) the company has +begin. At [Scribd](https://www.scribd.com/careers) the company has built a good product and curated a large corpus of written content, but where next? As I alluded to in [my previous post](/blog/2019/platform-engineering-at-scribd.html) about the Platform @@ -40,7 +40,7 @@ admit this, but much of our direction was heavily influenced by two conversations, both of which took less than an hour. The first was with [Kevin Perko](https://www.linkedin.com/in/kperko) (KP), the head -of our [Data Science team](https://www.scribd.com/about/data_science). His team +of our [Data Science team](https://www.scribd.com/careers). His team interacts the most with our current data platform (HDFS, Spark, Hive, etc); in essence Data Science would be considered one of our customers. I asked some variant of "what's wrong with the data infrastructure?" and KP unloaded what diff --git a/_posts/2020-08-10-datadog-backup.md b/_posts/2020-08-10-datadog-backup.md index 736627d..5efede6 100644 --- a/_posts/2020-08-10-datadog-backup.md +++ b/_posts/2020-08-10-datadog-backup.md @@ -89,4 +89,4 @@ Core Infrastructure team is up to by [reading more of our posts](https://tech.scribd.com/blog/category/core-infrastructure#posts). If shipping metrics and managing cloudy things is up your alley, you just might be the kind of person who'd love to work here, so [click -here](https://www.scribd.com/about/engineering) to review our open positions! +here](https://www.scribd.com/careers) to review our open positions!