CASE STUDY

Harry's Accelerates Global Marketing Analytics with Intelligent Data Pipelines

Accelerated the release cycles of new pipelines to hours instead of weeks.

Built-in alerting and Pagerduty integration means response and resolution of issues within minutes.

75%

Reduced pipeline release cycles with CI / CD via CLI, git integration and version control.

25-50%

Ascend automation frees up 24-50% of time for each user.

Industry:

Consumer packaged goods, primarily Personal Care products.

Technology:

Google Sheets

Looker

Spark, pySpark

AWS, Redshift

SAP

PagerDuty

Team:

Data analysts, analytics engineers, and data engineers from multiple teams are using Ascend.

Spanning SQL and Python skills.

Prior Solutions:

Custom data stack on AWS native services and databases.

Ready to explore Ascend?

Build Data Pipelines 10x faster at 67% less cost.

The Harry's Story

Harry’s is an early direct-to-consumer product led business that makes grooming products for men, and has become the clear #2 player in the US men’s non-disposable razor market. Harry’s believes that you shouldn’t have to compromise when it comes to the products you use, and engineers them to provide a great experience. Harry’s has expanded into many retail channels, which demands broader, more diverse omni-channel marketing attribution tools.

Technical Results:

Accelerated development to create new analytics within days instead of weeks.

Reduced sources of error in the data with built-in data quality features.

Rely on Ascend automated data partitioning to manage processing costs of large datasets.

Economic Impact:

Reduced running costs with reusable dataflows and eliminated redundant single-use pipelines.

Growth of data pipelines is faster than growth in the consumption of resources.

Data pipelines are key to driving ROI for a media mix model that is worth millions of dollars.

Wow Moments About Ascend:

Unparalleled engineering-level partnership and support by the Ascend team.

New team members build and release production pipelines within two weeks.

Ascend scales across three separate business units, and is now acting as a data broker across all those businesses

The Business Imperative

As Harry’s expanded its business from direct-to-consumer to embrace retail channels, their analytics team needed to quickly onboard new types of data sources to start to measure omni-channel attribution of marketing spend. The sources include sales data, market data, and marketing spend and impressions data, which requires significant cleansing and normalization before loading into the data warehouse running in Redshift. Ascend has proven to be a powerful rapid development platform for pipelines, enabling quick iterations and business feedback to arrive at the desired data product.

Ascend is creating great value quickly with marketing attribution tools. This has to be highly tuned to specific marketing and retail channels. Harry’s operates in 30+ channels, so understanding marketing ROI alone uses over 20 inputs to understand the return of every marketing investment, and its interaction with the direct-to-consumer and retail businesses.

"We use Ascend end-to-end, from data ingestion through to our BI tool. For example, we power marketing attribution tools via Ascend data pipelines, by using Ascend to combine and transform marketing and sales data."

Harry's Intelligent Data Pipelines

Ascend transformed the speed with which the Harry’s team is able to implement, test, and launch the required data pipelines. Initially, a BI team of two was able to meet business needs for new reports by relying on Ascend automation to create intelligent data pipelines, and reduce turnaround times ranging from a few months to just a few hours or days. The pipelines ingest data from all sorts of messy sources, including Google Sheets, CSV, databases, and applications, utilizing both the built-in connectors as well as plug-in custom Python libraries.

Harry’s has consolidated many of these sources into connectors that eliminate a lot of duplication and sources of error, improving reliability and achieving frictionless scale. The pipelines primarily populate the Redshift data warehouse, which powers Looker reports and ML models, and also feed datasets into applications like Harry’s demand forecasting tool.

"So we got out of the business of maintaining Python scripts, and got into the business of SQL transformations, which has been a huge upgrade for us in terms of maintenance and scalability."

Business Results

Marketing attribution tooling alone lies at the heart of millions of dollars of marketing spend optimization. Getting this right is a significant competitive advantage, and doing it at speed and with agility means Ascend returns its investment many times over.

A key benefit of Ascend is the elimination of system maintenance, infrastructure concerns, and most other IT aspects of running a business. In the past, these areas got complex quickly and consumed a lot of time and people to keep up to date and running. The result is much faster iterations on data and reports, which is where the value lies.

Harry’s data team relies heavily on the advanced functionality of Ascend, such as pausing and resuming without reprocessing to inspect and update data pipelines. They also use native incremental data processing to handle large datasets, some of which would take up to six hours to process. With Ascend keeping track of changes and processing only what is new, this throughput drops to a fraction of the time and costs, and speeds up delivery of the data products.

A key differentiation for Ascend is the partnership and support that Harry’s has to address any issues, and the depth with which the Ascend team remains engaged. From individual engineers to the field CTO to the executives, Harry’s is confident in the use of the solution and in their ability to get the attention they need to continue their journey together.

"You’re saving probably over 25% in data processing because it [Ascend] knows that the data hasn’t changed. And it’s nice that you can select the exact transformation node that is going to be associated with every error alert, and instantly pinpoint where the trouble lies. That’s nice."

Harry's Intelligent Data Pipelines

The Harry’s team continues to grow their relationships with retail partners and large global distribution channels, building on new data sources and creating new insights that drive their brands. They are also taking on responsibilities in marketing other brands under the Harry’s umbrella, including Flamingo, Lume, and Cat Person. These brands were acquisitions and incubations that operate their own data stacks, with opportunities to extend the winning Harry’s marketing attribution tools to other customer profiles, products, and buying behaviors.

On the technology side, the team will be increasing the use of Ascend built-in data quality features to further automate the integrity of their data products and customize the Ascend error and notification settings. The ease of integration with Ascend will allow them to extend notifications into PagerDuty to scale their incident response capabilities. The team will also take on more “hit-level” data from websites and e-commerce, moving into big data, near-real-time processing. For this, they have been exploring the Ascend partitioning features to wring the most value out of their spend. These capabilities are not available anywhere else.

"We are just operating in an environment where there’s just so much variety of data... In that universe, Ascend is a really good tool. That’s why it works so well for us..."