Hightouch Raises $12M To Empower Business Teams With Operational Analytics
Reverse ETL is the hot, popular trend within the modern tech stack. As software firms of all sizes generate more data than ever, their data warehouses are becoming more and more important in facilitating the rise of operational analytics in various internal organizations. Hightouch has ushered in the era of operational analytics by turning “any SQL query into a live data sync to over 60 SaaS tools.” Hightouch cofounders Kashish Gupta, Tejas Manohar and Josh Curl have raised $12.1 million in their Series A round led by Amplify Partners with participation from Bain Capital Ventures, YC, and Afore Capital earlier this year. The startup is based in San Francisco, counts Autotrader, Nando’s, Lucid Chart, Plaid, Retool, Blend, and Ramp as customers and claims to have grown 37% month-over-month (MoM) since their seed round.
Frederick Daso: It’s well known that the main key innovation in data warehouse technology, such as separation of storage and compute, have allowed the data warehouse to assume its role as the centralized “source of truth” fully, but there are other factors as well that have affected. What developments in the data warehouse architecture allow the latter to be included in a company’s operational workflows?
Kashish Gupta, Tejas Manohar and Josh Curl: The main reasons for the adoption of the warehouse, beyond the separation of storage and compute, are technology and cost improvements, a better tooling ecosystem and the proliferation of SQL and SaaS tools.
As the warehouse became cheaper due to technology improvements (like separation of storage and compute, alongside usage-based pricing models), even early-stage (ex: Series A) companies began adopting them. Data warehouses also moved to the cloud, so now very small companies can easily spin one up without full-time staff to maintain the infrastructure for the warehouse. As a result, most companies can now afford a data warehouse and benefit from the cloud’s speed and infinite scalability.
ETL tools like Fivetran, Stitch, and Matillion make it easy to set up and send data to a warehouse with the click of a button. For transformations, dbt has made it easier than ever to organize models across a company. Lastly, Airflow lets teams orchestrate data workflows in a standardized way. SQL users can accomplish much more than ever without having to learn how to code.
In addition to architectural innovations, organizational and people-related changes have also given way to the data warehouse becoming the obvious source for organizational workflows, not just a single destination used for analytics.
The proliferation of SaaS tools and various databases across a company created the need for one central source of truth that replicates all of the other data silos. Analytics is no longer just done on one database but rather by joining data across many databases and data systems. The data warehouse serves as that repository that stores backups of all the data systems across a company.
More and more users within a company know SQL now (such as RevOps). Instead of consuming data within a BI tool or spreadsheet, accessing data directly via SQL is becoming more common for more organizations.
Daso: Since your last fundraise, have your customer’s needs for reverse ETL evolved or expanded? If so, how have you shaped your product development to ensure Hightouch can maintain product-market fit?
Gupta, Manohar and Curl: Hightouch started as a significantly more technical product since the product’s initial users were primarily data engineers. Over time, as we began working with bigger and bigger companies, analysts and business users wanted to use Hightouch to power their own workflows. As a result, the scope of the product expanded. We continued to build out our core competency with technical users while expanding the feature set to make the product more accessible to everyone within an organization.
Hightouch now supports multiple ways to access data past utilizing SQL queries and filtering customer segments visually on top of existing models. Customers such as Autotrader use our Visual Audience Builder, which allows team members who may not know SQL to filter customer segments visually on top of existing models (for example, targeting users who visited your site in the past 30 days). We also allow data team members to select and use their existing data models from popular tools like dbt.
Hightouch also supports workflow-like actions such as creating tasks and tickets in Asana, Jira, Zendesk, or Salesforce. Hightouch also supports creating notification feeds in Slack when customers do certain app actions, such as putting down a credit card. We’ve seen companies like Blend use Hightouch for automation and workflows, putting their data into action by automating Slack messages for their services team to help meet customer deadlines instead of relying on static reports.
While a core focus of Hightouch is to serve less technical team members, the majority of our users are data engineers, so we dedicate a lot of our roadmap to unique developer features. Hightouch can now be triggered via API, dbt, or Airflow. Our live debugger helps users understand the API requests that Hightouch makes on their behalf. Hightouch also supports version control through git and configure the syncs from their command-line interface (CLI).
Daso: Hightouch’s view is that everyone in a company using the product will become a “data person.” What are the underlying assumptions and implications of the perspective?
Gupta, Manohar and Curl: We believe that all users of Hightouch will become data people and that most people in a company already are data people. Most roles require the use and understanding spreadsheets, BI tools, or SQL, but these roles don’t have “data” in the title. It’s not that people don’t understand data. It’s that tooling wasn’t built for data-savvy people to use data in their workflows. Data is still primarily accessible to developers, and that’s the problem.
We think that the status quo of data management forces ownership of data into the hands of a small team, largely for two reasons. 1) The data team understands the underlying structure of the raw data. 2) The data team has the knowledge and skillsets to transform and model that data into something usable downstream.
In other words, the “translation” from raw data to usable data has historically happened through esoteric, often repetitive work done by data experts—tools like dbt and Hightouch help to centralize and automate that translation layer. Thus, business users who own downstream tools (like CRM, email, ads, etc.) can work with refined, modeled units of data. They can do so even if they don’t 1) understand the raw data or 2) know how to model the raw data with SQL.
Daso: Operational analytics, so far, has been oriented around event-driven changes in the data of a user in the data warehouse being updated in the end destination, the SaaS tool. However, there may be cases where operational analytics doesn’t simply stop at updating data but requires teams to react to that event causing the update. How is Hightouch integrating itself deeper into SaaS-supported teams to facilitate their workflows at the most granular scale?
Gupta, Manohar and Curl: Traditionally, users may think of Reverse ETL as a simple copy and paste operation that takes data from one table called users and moves it to a user’s object. Our design partners were a lot more creative and have helped us build end-to-end workflows powered by the data warehouse in Hightouch. We are the only Reverse ETL provider that connects with workflow tools like Asana, Slack, and Jira. On top of that, our integrations go beyond data syncing and power workflows within tools like Salesforce, where we let users convert leads to contacts upon the closing of a sale, for example.
Another example is triggering ad campaigns or emails based on events and user statuses within the data warehouse. B2C and B2B customers alike use Hightouch to upload audiences to ad tools like Facebook, Google, Bing, and even Tiktok. The best part is that those audiences dynamically update as the data changes in the warehouse. For example, when someone buys a product, they can automatically get added to a campaign that excludes them from future ads, so now you won’t keep getting ads for those sandals you bought last month! With our Visual Audience Builder, marketers can filter users and upload them directly to their ad tool of choice without any help from engineering. The campaign becomes automated for good.
A good example of this use case is with our customer Blend, who uses Fivetran, dbt, and Hightouch.
Daso: Hightouch was originally built on your ability to write a SQL query. Now, you’re removing that barrier with what you call visual querying. What is visual querying, and how does it benefit even experienced engineers who are well-versed in SQL?
Gupta, Manohar and Curl: Our Visual Audience Builder is an interface that allows users to filter user segments with a point-and-click experience, regardless of how well they know SQL. In the past, those audiences and models were created with SQL. With the Visual Audience Builder, users take existing models prepared by the data team and add their own filters on top to create audiences that can be synced downstream. This means that business teams can be more autonomous and agile in sending data to their downstream tools without waiting on engineers, and on the flip side, data teams can spend more time focusing on strategic work.
Even experienced engineers who are well versed in SQL find themselves reaching for our Visual Audience Builder because it’s simply a lot easier to use than writing SQL, and they can get more done faster. Funnel queries are a really good example of this because they’re uniquely complicated to write in SQL but quite intuitive to understand visually.
Daso: How will Hightouch enable “data activation as a whole” in future products it will release for users?
Gupta, Manohar and Curl: For us, activating data doesn’t just mean getting data out of the data warehouse. Not only do we support production databases as a source, but we also support Google Sheets, Airtable, and Looker as a source. That’s why saying that Hightouch gets data from a data warehouse into a SaaS tool is an extremely limiting way to describe our product. The idea is that anywhere there is organized and structured data in your company, you should be able to activate that data with Hightouch.
On top of that, we also plan to enable real-time streaming use cases by connecting with more real-time data sources and destinations. We’re currently working on source connectors with faster data warehouses like Materialize and Firebolt, and we hope that Snowflake will develop similar features over time. Additionally, there are many more destinations where fresh data needs to live to power core workflows and personalization. For example, Internal ETL helps moving data from the data warehouse into production databases, and our customers then use that production database to power in-app personalization. Finally, we will continue to invest in our API to enable more flexibility with data transfer and embedding hightouch into existing workflows.
This means that Hightouch will be responsible for all operations that utilize data in the warehouse for anything more than BI. Getting data into faster databases is just the first step. Hightouch plans to release features that make SQL queries more and more powerful. Imagine being able to turn a SQL query into something that powers a production app or being able to click a few buttons and choose the latency at which you want your metrics to be available and how often you want them to be calculated. Hightouch can enable all of these processes with the same underlying technology that we have today, but with different UX to enable different business users.
Daso: What were some of the core use cases that necessitated your partnership with dbt, allowing for the latter’s models to be accessed within Hightouch itself?
Gupta, Manohar and Curl: There is a very deep product partnership and integration here, where dbt models are available within the Hightouch UI and replace Hightouch’s modeling layer. Users connect the git repo that stores their dbt project, and Hightouch automatically pulls existing dbt models into Hightouch to be used by anyone in the workspace. This replaces the need to create models out of SQL queries within Hightouch. The idea is that this will make the data modeling work done by data teams useful to business teams in their operations by letting them use that data in their CRM and other SaaS tools. Hightouch brings dbt models into tools like Salesforce, Hubspot, Zendesk, Google Ads, Facebook Ads, etc. Almost all use cases in Hightouch require clean data of some sort, and dbt has become the best in class tool to model clean data.
Daso: With Fivetran being developed to push information into the data warehouse and Hightouch to pull out into a SaaS tool, the data warehouse can begin to realize its full potential as the source of truth. How will Hightouch’s partnership with Fivetran bolster your startup’s ambitions?
Gupta, Manohar and Curl: Hightouch does the exact opposite of Fivetran. Fivetran gets data into the data warehouse via ETL, and Hightouch gets data from the data warehouse (Reverse ETL). Because of this common layer, the data warehouse, many companies will use Fivetran + warehouse + Hightouch to power their entire customer data stack. A great example of this is using Fivetran to get Zendesk data into the warehouse and then using Hightouch to get that Zendesk data into Salesforce. This gives sales a great view of how much support a customer is requesting. By making data in the warehouse more useful, Hightouch makes Fivetran more powerful and vice-versa. Many workflows within Hightouch would not be possible without the data provided by Fivetran.
Many teams have existing definitions in their data warehouse. Think lifetime value, Product Qualified Leads (PQLs), propensity score, customer health, ARR/MRR, funnel stages, and more. You can easily create reports and visualizations using this data in BI tools or SQL. Still, these insights are much more powerful if they drive the everyday operations of your business teams across sales, marketing, finance, and more in their native tools.
With Fivetran loading data from tools like sales, marketing, support, and success into the data warehouse, Hightouch can be used to sync those records and objects to business tools. The idea is similar to getting Zendesk data into Salesforce or Salesforce data into Intercom. Still, the difference is that the data has been massaged by analysts into models that are actionable by business users.
These calculated metrics and definitions are typically locked up in the warehouse and cannot be synced between tools via point-to-point integrations. Using the data warehouse not just as an analytics layer but as a modeling layer that sits at the center of your data stack allows any team to keep their tools synced. We believe this is the key to building incredible experiences for customers.
Daso: How have the responsibilities of running the startup evolved or shifted among the three of you? Specifically, what roles or “hats” are you looking to hand over to new employees to wear as Hightouch scales and you three cofounders focus more on the company’s future than its day-to-day operations?
Gupta, Manohar and Curl: Until January of this year, the team was just us three and our first hire/founding engineer Ernest Cheng. Now the team is twenty-one people, and we expect it to grow to forty by the end of the year! Saying that we’ll go from four to fourty feels surreal, and we’re still wrapping our heads around it, but you’re right; it has caused a shift in our responsibilities. Josh manages Engineering while Tejas and I (Gupta) split Go-to-Market. Since engineering is the largest team, Ernest plays many hats, such as engineer, PM, and architect. I think his title is still Generalist in our HR system, though!
As the team has grown over time, we’ve come to realize that our primary function is to do two things: One is to build the best team possible, and two is to let them run and enable them to do their best work. It’s a simple formula centered around one thing: we build the team, and the team together builds the company. It used to be that when we wanted to do something, we’d have to do it ourselves. Now more times than not, we enable others around us to lead and take ownership of their work and the goals driving that work. I always say that our job is to make the people around us happy because happy people do good work and have a ton of fun while they’re at it. Increasingly, we can make longer-term decisions and strategize how to create a category over a longer period. All three of us love product, and we enjoy thinking of cool new things to build every day. Our customers always brag that their feature requests were built at lightning speed, and we love making that happen!
We’re currently looking to hire across the board for Engineering, Sales, Partnerships, Marketing, and we’re even looking for a recruiter to help us hire faster. hightouch.io/careers has all our open positions, and we’d love to chat!