By Ken Weiner, CTO at contextual intelligence company GumGum
Sharpen your pencils: the back-to-school season returns in full swing this September. With total back-to-school ad spend reaching $32.5 billion last year, reported here on Deloitte – a 16% YoY hike – the 2022 curve will continue apace, driven by a new wave of buoyancy in digital retail (and a booming sideline in classroom tech).
From Nike’s #Best1stDay social media callout to Target’s “School List Assist” feature, most brands are well-oiled in the art of standout marketing campaigns: but this year’s digital landscape is particularly hard to navigate. The recent SCOTUS rulings, the tragedy in Uvalde, Texas, along with the war in Ukraine and an ongoing global health crisis, make for a febrile online climate.
This volatility means it’s more important than ever to fine tune the notes of brand suitability, for a pitch-perfect playbook. Ad tech alone is not enough: marketers need to educate themselves on how exactly content is being classified and targeted in an erratic, fast-moving space. Here are the key points to think about in finessing your safety strategy before the bell rings, and a new school year begins.
Scrutinize Your Tech Partners – Especially on Video Content
Like concrete floors in a Renaissance-era building, most keyword blocking devices are simply a poor fit for the nuance of online safety. For brands, the impact of this mismatch is twofold: most obviously, a keyword tool that cannot accurately scan a video’s content in its entirety poses a clear risk to brand safety.
A recent Kellogg’s UK ad documenting the touching story of a boy’s return to school after COVID-related school shutdowns, for example, was a marketing masterstroke. But the campaign tagline, “Whatever you do, we do breakfast” would jar badly if, as a web banner, it ended up next to an article about childhood obesity.
On the flip side, however, is the chance that vendors – overcompensating with blunt blocklist tactics – will miss opportunities to align with brand-suitable content. A blanket block on the word “shoot,” for instance, may protect against gun crime, but it also rules out perfectly safe content. For instance, a campaign promoting NFL Sunday Ticket to student football fans may be blocked from anything containing the phrase ”basketball shoot.”
To get the balance correct, brands need to be asking the right questions of their ad partners and digging further to discover what kind of targeting technology they’re using. Is it automated, and if so, how? Does their approach complement your objectives?
This scrutiny is particularly important when it comes to mostly unregulated – but highly engaging – back to school video environments such as gaming and CTV. Video metadata is only as incisive as a publisher makes it, and CTV carries the added complexity of live streaming content. So, it’s vital to know how – and if – a vendor’s tech is sophisticated enough to go beyond metadata to understand child or teenage video content and add extra precautions.
Ensure Keyword Targeting Is Responsive and Robust
Ad tech operators who rely on keyword blocking alone should be a red flag, but – since most companies will have an element of keyword targeting in their arsenal – it makes sense to build a picture of how this adapts to developing events. Updating keyword lists in response to disasters such as a school shooting is a basic step, but brands still need to reach out to their vendors proactively when these situations arise.
The smartest safety option is for a vendor to contextually scan and analyze a web page in real time. Theoretically, this could be achieved with human curation, but it would be a hugely impractical, and labor-intensive approach. So, a solution that uses the human-like perception of AI is ideal; with Natural Language Processing to understand the semantics of tone and sentiment, and Computer Vision to break down images or footage on a frame-by-frame basis.
Machine learning is a prerequisite of sophisticated and scalable brand safety, but it isn’t a panacea in itself. Brands should also understand how providers are training their programs to avoid biases and keep pace with developing vocabulary – such as “bubble” meaning COVID-related support group, rather than bubble machines for kids. This type of razor-sharp attention to detail will allow AI systems to scan for brand-suitable pages more efficiently in back-to-school placements.
Compare and Contrast Vendors
Content taxonomy is another area where more accountability is called for in the targeting world. The GARM brand suitability standard does a decent job of laying out the subtleties of brand safety; but many people misinterpret the severity of its risk levels versus content context. If you’re advertising a college friendship app, for example, running that promotion in an article that mentions beer in a fictional or historical context is far less risky than footage that details college alcohol abuse.
It’s also worth bearing in mind that a vendor that verifies its own brand suitability targeting leads to a false sense of effectiveness. The two sides of the targeting coin – the technology itself, and its certification – should be independent, just as classification needs to be accurately labeled and universally adhered to.
Fueled by new-gen tech and a boom in digital retail, the back-to-school season will be more profitable than ever this year: but brands that thrive will need to do their homework. Marketers should be unafraid to shop around, ask difficult questions and compare ad tech; potentially by running tests on their own manual sample of web pages. It’s this forensic deep dig that will make the difference between suitability targeting that sounds good; and tech that truly delivers.