At Madhive, we help our clients reach precise, high-value audiences for their campaigns through cutting-edge CTV advertising technologies — and as we like to say, the trickier the audience, the better. So we wanted to see if we could succeed with one of the hardest audiences to pinpoint, B2B marketers.
In the world of B2B advertising, the target audience tends to be extremely specific, which seemed like the perfect opportunity to put our tech and data to the test. We launched our first B2B CTV campaign, targeting people who could benefit from CTV advertising solutions and had the power to influence within their business.
Our campaign had three phases:
Throughout the campaign, we optimized our budget for conversion – not just impressions – and tested out some experimental AI technologies, too.
Now that the campaign has come to an end, we're sharing the insights and outcomes we experienced from this enlightening test. Hopefully, some of our learnings can help guide your own CTV advertising strategies, too.
Let's dig in.
We began building our campaign in the most obvious place: with first-party (1P) data. We took a list of known contacts, anonymized them, and matched them to IP addresses so we could deliver our ad to these households.
As is common in the B2B world, this list was fairly small, so we wanted to bulk up our higher-intent audience bucket. We decided to seed the campaign ahead of launch by announcing the test on Linkedin first, then retargeting any IP address that clicked through to learn more about the campaign.
The results were a goldmine for conversion. In every flight, the retargeting line item performed the best for conversion.
This approach was a clear winner in driving conversions, but the reach was, by nature, restricted. So while retargeting and 1P audiences drove 62.8% of all conversions, they only delivered 0.46% of total reach.
Knowing that our 1P list lacked the scale we needed, we turned to 3P data sources to build a unique profile for our target audience. We partnered with Datonics, blending job titles, industries, and income.
So how granular did we get? To start, we focused on job titles:
We also segmented by industry and job function:
Finally, we layered in income targets as well, to ensure we were more likely to hit decision-making audiences.
Datonics' 3P data acted like a compass, guiding us towards new potential viewers — and ultimately 2.7 million unique viewers across the entire flight.
We wanted to avoid a "set it and forget it" mentality with our CTV campaign, so we continued to evolve our targeting as we honed in on what really moved the needle.
As the campaign unfolded, we reallocated our budget to double down on what worked, so we weren't just throwing darts in the dark; we were recalibrating our throw based on where we hit the bullseye.
By shifting funds towards high-performing locations and giving a boost to our AI-driven initiatives, we saw our efforts bearing fruit in real time. For example, one important outcome from our optimizations was improved conversion.
In flight 1, it took 5 to 7 views of the ad before someone converted. After many optimizations, we improved that number; by the end of the campaign, audiences were converting after 3 to 4 views, illustrating the importance of adjustments to get the most from TV as a performance channel.
Picture this: an ad that mispronounces our name, yet still manages to hold its ground amidst its human-crafted counterparts.
That's what we learned when testing ad creatives. In the third flight of our test, we delivered multiple versions of our ad to see if slight variations on the messaging made a difference. We also included a version that was AI generated, to see if a computer could compete.
This toe-dip into the AI pool was to prove a point: AI in advertising isn't a distant future — it's here, and it's brimming with untapped potential. For smaller teams and budget-conscious businesses, this is akin to discovering a secret weapon. Harnessing AI for creative generation opens doors to innovative storytelling and engaging content, thus leveling the playing field.
While there are many wonky executions of AI generation, and sometimes uncanny valley results, our campaign proved that careful AI creatives can pass the test with viewers.
Of course, we didn't stop here when it came to AI testing for our campaign. We also used it to create next-generation lookalike audiences, as well as new profiles from scratch. We'll publish a dedicated blog on those experiments later this month; sign up for our newsletter below to get the results delivered to your inbox.
One of the really amazing side effects from running a CTV campaign is actually learning more about who your ideal audience is based on how it performs. This is often how small advertisers experience CTV – they don't have robust marketing teams, and much of their ideal persona is anecdotal, location specific, and/or assumed. Running a campaign actually teaches you how many of your assumptions are correct.
For example: When we launched this test, we assumed that cities would perform the best, since the business crowd gravitates to these more densely-populated areas. But in actuality, 19 out of the top 25 zip codes for conversion were actually small towns.
This made sense – our core audience is mature business professionals that likely live in suburbs outside the major hubs – but even if we instinctively knew this, we didn't have the data to locate them. That is, until we started planning the campaign and saw which zips had heavier populations with our targeting. Testing then enabled us to find what worked, and then double down on it.
Another surprising fact: we performed much better with east coast geos vs. west coast. The top 5 highest converting states were east coast, while California came in at #6.
Last, but certainly not least, we learned what our core audience was watching.
While many advertisers like to create a campaign with specific publisher parameters, we didn't know much about our audience's viewing habits beforehand and we decided not to set any limits. Instead, we focused on audience targeting, and set out to see where the Madhive persona is watching content.
The list differed slightly for impressions vs. conversions, but across the board, news and sports did well.
So what did we learn from this test, overall? No matter how specific your audience is, you can find them and target them efficiently — if you have the right tools in your arsenal.
We started our test with very little known data about our audience, and we used geotargeting, audience planning, third-party data, and AI ingenuity to make the blurry picture more clear.
While a B2B campaign is very different from what agencies and broadcasters are managing for their clients, the takeaway remains the same; in order to run effective, cost-efficient campaigns — particularly for advertisers with tight budgets and hard ROI goals — you need tools that help you make informed decisions during planning and execution.
