CASE STUDY

Podsights Partners with 
Deepgram to Pinpoint Podcast 
Advertising Attribution

Deepgram allows Podsights to extract maximum value from their data on behalf of brands.

The Landscape

Podsights proves to brands that podcast ad spend actually leads to new sales.

As of 2021 there are over two million published podcasts in the US alone and just over a hundred million podcast listeners. Naturally, brands have long been eyeing the advertising potential of the format. However, as recently as five years ago, brands were incredibly hesitant to leverage podcasts despite their obvious and rapid growth potential. According to Matthew Drengler, Director of Partnerships at Podsights, “Brands weren’t willing to dip their toes in the podcast media space because there were really only old school, unconnected attribution methodologies out there.” Traditional methodologies of podcast advertising attribution include things like coupon codes and vanity URLs. But vanity URLs are only visited by about 20% of podcast listeners, and coupon codes, while they certainly get used on brands’ websites, are not reliable indicators of listener engagement. Often shared online by listeners on publicly available lists, coupon codes may be used by shoppers who may have never listened to the podcast.

Neither of these methods of tracking campaigns presents an accurate picture of ROI on ad spending. But Podsights has changed all that. Six years ago, Matthew Drengler says, “Ad spend on podcasts was $170M. This year spend hit $1.3B.” What got brands to take the plunge into the murky waters of podcast ads? The answer is simple: Podsights.

The Challenge

Finding a better podcast attribution solution that directly related podcast engagement to sales.

Podsights did not set out to solve the biggest problem in podcast advertising attribution and become the de facto platform in the space. Initially they were building a research tool. You can still access their incredible database on the relationships between brands, publishers, and podcasts. In the process of building this impressive database, the problem of advertising attribution was discovered. Podsights quickly realized that they needed a better method of attribution. But there were serious challenges: working with hundreds of publishers, each in turn working with dozens of podcasts, which collectively have millions of listeners across multiple platforms, and hundreds and hundreds of brands whose ad campaigns run on multiple podcasts at once. The sheer array and interplay of data made validating and scaling a podcast based ad attribution solution seem impossible.

Throw in the various ways ads are inserted. Some are dynamically inserted, meaning they are recorded and approved by brands ahead of time and dropped in as separate sound files at programmed ad breaks. Embedded ads that are read live by the host or edited into the recording of the podcast itself. For these types of ads, brand staff have to listen to every single podcast episode on which their ad is run to ensure compliance.

Because they had built this research tool that collected all of this podcast data, they pivoted to helping brands and companies analyze their ROI on podcast advertising.

The Solution

Tracking with Pixel-Based Attribution and STT technologies.

Podsights delivers a data-rich alternative to vanity URLs and coupon codes. Their pixel-based attribution offers a results-based dashboard for podcasting campaigns.

“What Podsights does is provide what’s called ‘digital style,’ or pixel-based attribution for the podcast media space. Effectively, what we’re doing is calculating the metrics that brands and agencies can use to create an accurate, apples-to-apples comparison between podcast media and all other forms of advertising.”

The way it works is this: Podsights places a piece of code, called a “pixel,” on both the podcast download side and on the brand site side. When each of these links is clicked, it sends data to one of two disparate databases. One database contains all the identifying information on the podcast listeners, and the other tracks the identifying information of the brand’s website customers, as well as their actions, such as adding a product to their cart or making a purchase. Podsights’ analytics can then tie the information from those databases together in meaningful ways, creating a detailed and accurate picture of the relationship between podcast listeners and brand customers. This is very similar to digital click-thru rates and sales for online advertising. Podsights can demonstrate to a brand that a podcast directly lead to a sale or site visit.

Of course, the conclusions Podsights can draw from the correlations are far more valuable if you know with certainty that podcasters are actually reading ads and reading them correctly. This is where Deepgram comes in with our highly accurate STT solution.

Matthew Drengler, Podsights

With Deepgram’s accurate and fast speech-to-text solution, We’re the Google Analytics of podcasts.”

Matthew Drengler

Director of Partnerships, Podsights
podcast-analytics

The Results

Podsights is the Google Analytics of Podcasts

Direct attribution from Podsights, according to Drengler, means “Now brands can look at our metrics and start saying things like, ‘Hey, 50% of my podcast media budget drove 50% of my sales and oh, by the way, here are the dollars that drove that. So we’re answering that old question, ‘How much, and exactly which part of my budget is driving success?’”

Of course, to place pixels on ads and to ensure that ad-reads are compliant, Podsights needs a stellar STT solution. That’s where Deepgram has been instrumental. Deepgram’s STT has helped Podsights provide brands with better insight and auditing capability on ad-reads. In the text-based ad world, tracking impressions is easy. Speech is much more complex. By working With Deepgram’s STT, Podsights is able to provide transparent impression reporting in real-time, and track campaigns across multiple providers on one centralized dashboard, and perform automated air checks. Also, when your ads are speech-only, you have to know who said what and when. With the help of Deepgram’s rapidly evolving diarization, Podsights can attribute words to each speaker in a recording, so that brands can tell who is reading their ad. Naturally, Deepgram STT outputs include time-stamps and confidences for a fine-grained analysis. Podsights has leveraged a host of cutting-edge technologies in order to track exactly how an ad campaign is pacing and the number of households it is reaching.

Drengler notes, “With Deepgram’s accurate and fast speech-to-text solution, we’re the Google Analytics of podcasts.”

data exploration

Looking Ahead

Deepgram is Poised to Make Podsights’ Mission of Deep Data Exploration into One of Unlimited Innovation

Currently, attribution is Podsights’ main product, but with the help of Deppgram’s continually evolving speech-to-text solutions and tailored models, they are growing into something even bigger and better. The possibilities of redaction, automatic brand name recognition, and language/content-based ratings for spoken word are already being discussed by Deepgram and Podsights’ developers. With things like sentiment detection and context analytics just on the horizon, brands won’t be able to simply screen for “bad words,” but will be able to identify “bad context.” Podsights hopes to be able to help advertisers decide which podcasts align with their brand values and target audience’s values.

Podsights is also developing a tool they call “Advisor,” which will allow opt-ins from all publishers to send in their show data, so that Podsights can populate shows to match the audience targeted by specific brands. Advisor will be able to leverage user behavior as an indicator that a particular podcast will serve a particular brand based on that demographic or psychographic’s listening habits and shopping behavior.

Matt Drengler is excited about what lies ahead in their partnership with Deepgram. “Even with all of our success so far, I can say that the benefits are not yet to where they will be after we are fully leveraging everything from Deepgram.”

View Red Box Case Study