How Props uses BigQuery and image recognition to get deeper creative insights

Props is a New York-based content marketing platform that allows companies to collaborate with authentic content creators to drive conversions. The platform matches brands with creators, such as photographers, filmmakers, and writers, that align with their target audience.

These matches are made based on the creator's personality, subject matter expertise, and visual aesthetic. Props work exclusively with creators who are experts in their fields.

Props' clients don't advertise on social platforms in a traditional way. The creators advertise on social media on the brands' behalf while the brands cover the ad spend.

Props have an internal media team to manage client ads. They use a range of channels, including Facebook, Google Display Network, TikTok, and Pinterest.

Creative plays a key role in how audiences consume content online. Props' aim is to give clients a deeper understanding of what type of content works for their brand. To achieve this, the content platform uses three different types of AI to enrich creative insights.

We spoke to Oliver Blodgett, Props' COO, to understand their data journey.

Outgrowing Google Sheets as a data storage solution

Before centralizing its marketing data into BigQuery, Props used Supermetrics' Google Sheets and Looker Studio (formerly Google Data Studio) connectors to extract marketing data from ad platforms. 

Two factors led Props to seek change:

• Build reliable reports
• Improve its product's AI capabilities

Building reliable reports

Props was outgrowing its data architecture. It was working with increasing quantities of data, and Google Sheets was no longer a sufficient data storage solution. While Google Sheets is a convenient starting point, it isn't designed to support significant amounts of data.

Before migrating to a data warehouse, the Props' analyst spent 4-5 hours a day making sure its spreadsheets and dashboards were working. This was unsustainable for both the analyst and the data. 

Improving its product's AI capabilities

Props had already built a manual prototype of its content marketing platform but found it difficult to transform and manipulate the data at scale. Without automation, Props was unable to make full use of its AI capabilities. 

With a manual process, Props could include AI-enriched insight into monthly client reports, but it needed a better data architecture to make the product scalable. 

Props started exploring options. The team initially planned on building the required data stack in-house, but the task was more complicated than expected. After talking to freelancers and competing data pipeline providers, Props also realized how expensive the project was.

Props eventually discussed their goals with Supermetrics, and the idea of a full maintenance solution came up. They could have an ongoing collaborative relationship with Supermetrics instead of purchasing a one-off project. Since the Props team had already researched alternatives, they quickly realized that the Supermetrics pricing was competitive.

Designing a better data architecture

Professional Services is a team of engineers at Supermetrics that offers training, consulting, and dashboard-building services. Props and the Professional Services team have collaborated closely from the start of the project. They still have weekly calls to solve problems, refine the solution, and expand the data architecture. 

Oliver says, "Working with the Professional Services team has been great. It's really more of a collaborative relationship, both parties are learning new things. The team has really gone the extra mile to find solutions to all of our challenges. We still have weekly calls to go through any fixes, ideas, and new plans."

Props’ previous data architecture

Props had built a complex structure within Google Sheets to support its dashboards in Looker Studio. Multiple interconnected spreadsheets were powering each client dashboard, and maintenance required lots of resources.

Props had to manually enrich and transform the ad data to include AI insights into client reports. At this stage, Props was working with fewer clients because scaling was difficult.

Oliver says, "We were so accustomed to the granularity of data that Supermetrics provides for Google Sheets and Looker Studio that our bar was very high when looking for a data warehouse solution. We compared multiple data pipeline providers but kept running into dead ends with other solutions lacking the necessary data. Ultimately, it was an obvious choice to opt for Supermetrics and BigQuery.”

Props' new data architecture

Centralizing Props' marketing data was at the heart of the solution. Professional Services recommended BigQuery because of its ease of setup and low operating cost. It also integrates seamlessly with Google's cloud services, including two of the AI tools that Props use.

The new structure looks more complicated, but it has made Props reporting much simpler. Manual processes have been removed from the equation, and maintenance is nonexistent.

 

Here's how the new data architecture works:

1. The Supermetrics for BigQuery connectors extract data from multiple ad platforms and automatically move it to BigQuery.
2. Ad data is written into BigQuery and transformed: the data is enriched with data points from the Props platform and the three different AI softwares.
   1. The custom connector for the Props platform pulls in data about content, content creators, and the two AIs—Natural Language AI and Humantic personality analysis AI.
   2. The image recognition software, Vision AI, was more difficult to integrate into the architecture, so Professional Services built a separate integration to connect it to BigQuery.
3. Once the data is transformed in BigQuery, the complete data set is delivered from BigQuery to the Props platform, where clients can view their performance metrics in their workspaces. Client workspaces include dashboards and information about campaigns and creators.
4. Tableau is used only for internal reporting.

 

Oliver says, "Even though we used Supermetrics for Google Sheets and Looker Studio at the time, we hadn't realized the extent of other products and services Supermetrics offers, such as data warehouse solutions. Their engineering expertise enabled us to create a whole new level of reporting and insights for our clients."

The image recognition AI categorizes ad images, while the Natural Language AI recognizes unstructured text within ads. Props use the Humantic personality analysis AI to categorize its content creators to better match them with client brands.

One of Props' biggest challenges was enriching its ad data. With thousands of ads, it's impossible to manually apply ad categorizations. Supermetrics connectors solved the problem because they extract image-level URLs from ad platforms.

The Props team built the content marketing platform within Bubble, a no-code app platform. There was no standard solution for connecting a Bubble app to BigQuery, so Professional Services built a custom connector to move data to and from the Props platform.

The Supermetrics engineers also blended Props' Google Analytics data with its paid ads data using dynamic UTM tags. Previously, Props could see time-on-site per creator, and now it has instant access to that information for individual ads.

Building a new AI-powered content marketing platform

Watch the webinar recording to learn how Props uses BigQuery and image recognition to get deeper creative insights.

The new Props platform gives a deeper level of creative insights with a drastically improved time-to-value. Props customers have gone from spreadsheet reports and Looker Studio dashboards to see all their content and results in one platform.

Everything on the platform flows through BigQuery, which makes dashboard load times very quick. Props also customize each workspace based on the main KPIs of the customer.

Oliver shares, "Time-to-value has been one of the key improvements we've seen with Supermetrics and BigQuery. Our data flows are automated, and our clients have data available on demand. With the AI integration, we went from monthly reports to instant insights. It's a huge difference."

With the new data stack, Props can deliver the full value of its AI capabilities to clients. Before the BigQuery integration, Props would manually include image recognition insights in clients' monthly reports. That made the AI component only a nice-to-have feature since clients couldn't take immediate action based on the insights.

Oliver says, "Without Professional Services, we’d had to grow our team significantly to build our current data stack and to upgrade our client reporting. With Supermetrics, we've been able to focus on improving our AI capabilities and growing our client portfolio."

Confidence to scale

The confidence that Props has in its data has grown immensely. Its previous data architecture couldn’t support the growing business or its future visions. Now, data is an advantage rather than a bottleneck.

Previously, if Props' spreadsheets broke down due to excessive amounts of data, the system would pull the data again and recreate the dashboard, meaning nothing would work. Now, if there's an issue with one day's data, the dashboards can still revert back to yesterday's data.

Oliver says, “Imagine a client-side, C-level executive logging onto the dashboard for the first time and finding an empty dashboard saying ‘Data is not available’. That’s not the first impression you want to make.”

The new architecture removed the manual work that went into managing unstable spreadsheets and dashboards. With a solid data stack, Props can trust its reporting is up-to-date and accurate.

Being able to scale efficiently has been a turning point for the business. Onboarding new clients used to be a time-consuming process. Now the product works well from the start without requiring many checks and fixes for each new client.

In addition to ready-built dashboards, Props now has a self-serve reporting option for clients. They have a siloed section in BigQuery that customers can access to pull data into their internal dashboards. This gives clients an added level of flexibility in reporting.

Oliver shares, “Our clients can now access data directly through a siloed space in BigQuery. This allows them to integrate our data directly into their existing dashboards. Clients really appreciate the transparency. It generates a lot more trust than sending over a spreadsheet.”

Next on Props’ radar is to expand its offering to self-serve clients and advertising agencies. In addition to expanding its user base, Props is working on improving its image recognition by combining multiple custom models rather than using a general one.

If you’d like to learn more about our Professional Services team, have a look at the types of projects we work with. The Professional Services team can build highly customized solutions for customers. If data warehousing is more relevant for your business right now, you can request a data warehouse demo, and we can walk you through the options.