PanelBot
Automated Advertising Annotation
Intelligent Content Processing for Market Research Companies & Media Agencies
What is PanelBot?
PanelBot is an automated data extraction platform for media like videos, images, and HTML pages collected from market research panels and other first-party sources.
Using Computer Vision and Optical Character Recognition (OCR) technology, PanelBot recognizes and annotates brands, products, and advertising that appear in panelists’ recorded browsing activity, reducing transcription costs by 70% or more.
It provides the fastest way to go from unstructured streams of digital ads, sponsored social media posts, and eCommerce pages to actionable content intelligence and market research.
TL;DR
You: bring your own panel data (BYOP).
We: automatically scan and tag brand-relevant content that your panelists have viewed.
Media We Transcribe & Label
We identify market intel in almost any type of Paid, Earned, Shared or Owned (PESO) media. In addition to raw image, video, and gif files and URLs, PanelBot also scans URLs, PDFs, audio files, photographs, and other content.
Video & Rich Media
Video: .mp4, .mov, .wmv, .webm, .avi, .avchd, .flv, .f4v, .mkv, .mpg and more.
Audio: .mp3, .wav, .aif, .m3u, .ogg, .vorb, .aac, .wma, .3gp, .aa, flac and more.
Other: URLs
Images & Documents
Images: .jpeg, .jpg, .png, .gif, .bmp, .tiff, .raw, .aeif, .heic, .eps, .webp, and more.
Other: .pdf, .eps, .ps, .docx, .xls, .xlsx, .rtf, .csv, .tsv, .txt, and more.
Benefits
Beyond Transcription & OCR
Traditional OCR software only scrapes text and external content.
Since 99% of digital content that matters to marketers is image or video-based, this is a big blind spot for brands and their data providers.
PanelBot Artificial Intelligence (AI) generates as many as 80 Ad Tech-relevant KPIs for every media asset it scans. This list is constantly expanding and we can customize our output to fit your specific requirements.
Our data output answers two key questions in clear, objective detail:
What market intelligence does the media asset contain?
Including in-content (imagery, text and audio): Brand Names. Logos. Trademarks. Products. Celebrities. Pricing. Deals and Offers. etc.Where is the asset from and what can we know about the content publisher?
Including: Platform. Advertiser. Industry. # Followers. # Shares. etc.
Contact us for the full list of KPIs and inquire about pricing.
How it Works
-
Step 1: Upload
Integrate PanelBot with your market research panel database or select media to upload. We can also scrape digital content on your behalf.
-
Step 2: Process
PanelBot ingests and processes the data. Depending on the format, we pre-process media, deduplicating it, removing any PII, autocropping, and more.
-
Step 3: AI Annotation
30+ neural networks seek and tag occurrences of brand and advertising intelligence and pull relevant data from across the web.
-
Step 4: Refine
Operators may review, refine, and approve the labeled panel data. This step is optional.
-
Step 5: Export & Sync
All detected and discovered data is normalized and organized in your reporting format.
Fast and accurate panel data attribution is critical.
Understanding consumer behavior in 2021 can feel like a crapshoot.
Decoding the complexity of where and how customers find inspiration and validation before buying products – YouTube? Facebook? TikTok? Spotify? Instagram? TV? Peloton Classes? – is one of the reasons the Big Data Analytics market is expected to grow 28.9% annually to $68 Billion in 2025.
Market research panels are one available solution to this challenge – especially with marketing attribution and measurement becoming more and more challenging.
Millions of people are willing to exchange data (such as screenshots and screen recordings of digital ads and other in-app browsing and purchasing activity) for dollars. And with reliable first-party data in hand, marketers can develop sophisticated profiles of their customers and market.
But even if collecting this data is easy, annotating and analyzing it is whole other matter. That’s why we built PanelBot.