Taking the analytical mystery out of sponsored content - Knotch's new approach

Barb Mosher Zinck Profile picture for user barb.mosher October 17, 2016
Sponsored content is now an established part of the media landscape. That doesn't mean it's easy to measure. How did the content perform against competitors? Did the content impact brand sentiment? Barb Mosher Zinck explores these issues and more with a new startup, Knotch, which is focused on sponsored content analysis.

Sponsored content, or native advertising depending on who you talk to, is fast becoming a popular approach to reaching out to key audiences on publisher sites. But getting sponsored content right is tricky, you depend on the publisher to demonstrate they are the right site to work with and to give you accurate information on how your content is performing. But they can’t tell you how your content performs in relation to your competition, nor can they provide sentiment data for your content (is it hitting the mark?).

A new startup can tell you all of this and more. It’s called Knotch, and I took some time to chat with its CEO and co-founder, Anda Gansca, about how it works and why this technology is so important.

Our conversation started off by going back to one of my earlier columns: Data alone doesn’t provide insights, data connections do. Gansca is a math and computer science nerd who believes that you get something beautiful when the two come together. Gansca has always been passionate about the data space and knew she would create something, but she spent some time working in venture capital before she finally found her way to being an entrepreneur who could capture real sentiment data and provide the right insights to act on it.

Sentiment data is a beast to get right

Sentiment data - how someone actually feels about something. Gansca said it’s hard for people to verbalize how they feel about something let alone have a machine do it. Getting it right posed both design and data challenges.

From a design perspective, how do you engage an audience and get them to share how they feel. Also, regardless of how you ask a question, to get engagement, you need to ask it contextually. So you need to have your question be around the content and non-disruptive (kill the pop-up right?).

Gansca said that embedding a question around editorial content gives you a response rate of 15-20% whereas asking the question later (like in a follow up email, or a pop up) only provides a 1% response rate.

Then she looked at where could you apply your question so that it serves both publisher and audience. Branded content was her answer (sponsored content). Gansca said understanding the impact of the content on the audience (things like attention and sharing, but also if it changed how they felt) was critical.

So to bring you up to speed, that’s one part of what Knotch does. It places a question at the end of a branded content piece that asks the reader how they felt about the content. At the same time, there’s some quantitative analytic data collected on the backend - demographics, browsers, time spent on page, all that good stuff Google Analytics gives you.

The power lies at the intersection of qualitative and quantitative data

Gansca said that Knotch’s tagline is “not just data.” Because it’s not enough to collect all this data, it’s how you bring it together to give you valuable insights into how your content is performing, how it’s performing compared to your competitors and what you should be doing differently.

The analysis and insights Knotch provides came around a year ago when Knotch shifted its focus to helping marketers with branded content (it will move into different content in the future, but this is its niche for now). Knotch provides marketers who work with branded content with three things: measurement, knowledge, and wisdom.

Measurement is all about bringing the data - quantitative and qualitative together and giving you reports on how your content is doing overall, with each publisher and for each piece of content specifically. There’s an engagement report that captures sentiment (assuming they answered the question at the end of the content item), traffic, social and audience. For audience, you see basic demographics, but also demographics and sentiment together. You can also select one or more articles in real-time and compare their performance.

Knowledge tells you what the competition is doing. Knotch is like Google for branded content. It has an intelligent machine learning scrapper that goes out, and it scans the Web for branded content (typically identified by a special image) which it then indexes. Gansca said the scraper was very complicated to build, but that this was an area of great growth and possibilities. You can select brands and see relevant competitors and their branded content, or you can pick a trending publisher and see who does branded content with them.

Wisdom is the final piece, where it all comes together, Gansca said - how does your content do against everyone else, or what’s your share of real estate? Not only does this component tell you how you are doing, but it also makes suggestions about where you should be putting more content. It tells you things like your content lifespan or content half life, the best performing content and the top five best performing content items, your most used themes vs. a competitor’s and much more.

In addition to these core components, Knotch does create custom components for customer’s specific needs.

How Knotch makes it happen

To make this work, Knotch has partnerships with publishers such as Business Insider (check out this example), Vice, Quartz, Politico, and others. The publisher places what Gansca referred to as a “unit” on their web page at the end of a branded content article. The unit acts as a trojan horse, Gansca explained, gathering qualitative and quantitative data (which is why they can correlate the two).

Why would a publisher do this? First, know that Knotch doesn’t sell to the publisher. It sells to brands. So when Knotch goes to a publisher to get its unit placed on its client’s branded content, the publisher doesn’t pay for it. It does provide the publisher two things - it gives them the immersive experience they want to build (get rid of the painful intercepting surveys), and it helps them look good (assuming they are good at branded content).

Knotch is young. It currently works with about twelve brands in a range of markets, but to get an idea of who would use it, GE was its first client. Knotch recently picked up US$10 million in Series A funding to continue its work.

My take

Gansca said something to me that made perfect sense (even if it does promote her service) - can you trust publishers to give you self-reported data on your sponsored content? You probably can for the most part, but it’s basic reporting data - basic analytics. And there’s a lot more to understanding how content performs than traffic stats. You’re also trusting that the publisher understands analytics well and knows how to both read and report on them accurately.

There are three big value points in Knotch’s approach in my view. Assuming most readers engage with the question and answer it, then the sentiment data is extremely valuable. Second, the competitive knowledge can’t be underestimated. You need to know where the competition is and where your audience is. You need to be in the same places.

And finally, all the data in the world is fine, but you need actionable insights - what should you do next? Making that obvious takes a lot of work off a marketer’s back and lets them focus on creating great content.

Image credit - Business People Working Technology Devices Concept © Rawpixel.com - Fotolia.com.

Disclosure - Diginomica's business model is a variation on the sponsored content model described in this article. We'd like to think it is a significant variation - in a good way - but that's for readers to decide. Diginomica has no financial ties to Knotch and there are no Knotch trackers on our web site.

A grey colored placeholder image