“Data Driven Thinking” is written by members of the media community and contains fresh ideas on the digital revolution in media.
Today’s column is by Andrew Shebbeare, Founding Partner & Chief Strategist at Essence Digital.
No doubt about it: Direct response marketers bankrolled the real-time bidding revolution, while brand dollars keep fuelling growth on the reservation side. That is set to change. Specialist exchanges and supply-side platforms open new doors to media that is biddable, targetable and capable of moving a brand.
Last month I wrote about the fallacy of chasing the easiest wins and optimizing yourself into a corner. This time I’d like to offer up some of the practical lessons my colleagues and I have learned about using RTB to influences attitudes, not just drive actions:
1) Know what you’re buying. Some of the shiniest inventory out there is accessible through exchanges, public or private. So is the very worst—the buys that are capable of tarnishing your brand beyond repair. You may have tolerated this when you were selling diet pills; you can’t when you’re trying to create an icon. Get yourself a solid ad verification platform, use it to validate brand safety, to make sure your ads are getting seen, delivered in the right footprint and to the right audiences. Don’t even think about running a brand campaign without this.
3) Statistics matter. Measuring brand lift is hard. Measuring it in such a way that you can really optimize in-flight while avoiding jumping to false conclusions is even harder. Bad practice is all too common in the industry. So polish up on your stats, or find yourself a buddy who can help. The relationships between sample size, baseline attitudes, accuracy of read, granularity and confidence are complicated, but you need to get a grip on these levers if you are to design successful experiments.
4) Proxies win prizes. No matter how good you are at stats, you won’t measure brand lift at the grain you’d like or as quickly as you’d like. It’s a fact of life: Perception change takes time, and change across populations is an analog outcome. Getting granular reads is also costly; you need to strike a balance between insight and measurement expense. Knowing that the readers of this column will get itchy feet waiting days or weeks to know whether a campaign is driving results, I recommend looking to proxies for success to guide your investment along the way. Here are some suggestions:
- Ad viewability: If your ads don’t get seen, they won’t work.
- Out of footprint: If your ads get seen in the Falkland Islands, that’s where you’ll get brand lift.
- Audience: You guessed it. If your ads are getting shown to 4 year-olds…
- Early brand metrics. It’s easier to drive ad recall than message association, and it’s easier to drive message association than purchase intent. If your message is getting seen and getting through, at least it has a hope of working. Because those “early” metrics tend to move faster, you will get significance faster and optimize sooner.
- Engagement. Generally speaking, if people are watching your video all the way through, expanding your ad unit, viewing content or singing your company song, that’s a good thing. An ad with 10% engagement is probably better than one with 0.1%, and you should optimize in that direction. Just be mindful of tip #5…
5) Beware the pitfalls of engagement.
- Don’t forget reach. Engagements are valuable, but they are also scarce. Forced exposure delivers reach where engagement channels can’t.
- Be wary of preaching to the converted. Some formats or placements might get lots of engagement because you’ve found a niche that loves your brand. Engagement rates will look great, but brand lift won’t follow—you’ve just found a higher baseline.
- Keep your measurement unbiased. I’ve seen comparisons of brand metrics for people who did and didn’t engage with a rich media ad. This is not a fair comparison. People who engage are different than people who don’t. We should compare identical exposed populations. If engagement is adding value, we should still see the difference.
6) Stay in control. You’re going to need a control group. I tend to be wary of modeled controls; your mileage may vary. I prefer randomly holding out an audience at the point of ad delivery. You can do this with cookie-based holdouts or with panel-based technologies. For all the threats to the cookie and the shortcomings across devices, cookie-based ad effectiveness is still a powerful signpost that is exceptionally flexible.
7) Bonus is bogus. Control-exposed studies using in-ad attitudinal surveys are great. The downside is they also are prone to design errors that introduce bias.
Advertisers commonly ask for “bonus” reservation-based inventory for survey recruitment. This inventory is often scraped from the bottom of the barrel, bearing little or no resemblance to your actual media buy in position, frequency or even audience (we’ve checked!). Advertisers also look for publisher help retargeting to recruit within their sites, leaving recency, frequency and context up to them. Yet any method that relies on multiple different recruitment methods between sites introduces bias. Say Site A has light but daily visitation, while Site B has hours of weekend browsing. Recency and frequency will be completely different, and you can’t compare your reads. If you serve a companion survey and a video creative on the same page, you’re going to get some mighty impressive ad recall!
With RTB, we can take fine-grained control of survey recruitment. Through retargeting—making sure you make all the right disclosures and opt-out notices—you can build a robust control / exposed recruitment base and normalize frequency, recency and context across the board. Better still, you can adjust bids and budgets to make sure you have a critical mass of respondents in each test cell.
8) Frequency: RTB’s unsung hero. Changing attitudes is hard. It can take many exposures to drive a message home. In a reservation-based world, shooting for a given frequency drives a ton of wastage—it’s hard to make sure each incremental impression is getting delivered to underweight audiences and not just making the “frequency tail” longer. Marketers miss the importance of this when they look at averages; your frequency mean could be 4, but 1% of your audience might be wolfing down 50% of your impressions. That’s a big waste, and it’s pretty boring for that minority. With RTB, you can fine-tune your delivery, capping across media and over long periods, bidding up on lower-frequency audiences and dropping bids on the “tail.” Better still, you can tap into cross-channel synergies. We see time and time again that Effect of Video + Display > Effect of Video + Effect of Display. If you’re buying a lot of reservation-based video, use RTB to drive the message home and get more brand lift for your budget.
9) Context counts. Brand marketers are in the habit of thinking hard about context. Every ad buy gets thought about. This habit comes with a dual purpose: to get your ad somewhere you want it, and to reach the specific audiences that hang out there. With RTB, you can split these two things. You can buy audiences and/or context. My advice is not to neglect the latter. In the right setting your message will benefit from the medium that carries it. Ideally, RTB should do more than match traditional brand planning. We should be segmenting audiences and creating connections that add value for each of them, in context.
10) There’s no such thing as Real-Time Branding :-). Serious point: What I mean is be patient, and don’t expect all the answers to come from data. This stuff isn’t at the point where an algorithm can build you a brand. You still need a plan of your own. Be absolutely clear about your communications strategy and stick to it. Don’t get blinded by the data; measure what matters and focus on it.
You’ll probably want more than biddable inventory on your brand plan. I know I would. But hopefully this list is helpful in making the most of RTB opportunities in tandem with your reserved buys.
I’d be grateful for any other contributions; this article certainly won’t be exhaustive and we’re still learning every day.