Many businesses have the process all wrong. They select an advertisement type, such as social media ads, video, display banners or a channel like Facebook, before they even have a goal. However, the best way to go about it is to define a specific KPI and then determine the type of advertising that will allow you to reach that KPI. Be it brand recall or purchasing during the same visit, the type of advertising you should be using will depend on the desired outcome.
Match Your Model To Your Funnel Stage
The marketing funnel is more than an infographic. It’s a helpful roadmap for where to spend your ad dollars.
At the top of the funnel, you’re purchasing awareness from strangers. Naturally, CPM (cost per mille, or thousand impressions) is where you want to be. You’re paying for eyeballs, not clicks or conversions, which is exactly what you want when you’re trying to generate recognition. The right metric will be reaching x frequency, not conversion rate.
Further down the funnel, the right metric becomes more important. The more someone has already expressed interest by clicking, the better CPC (cost per click) is for you. The farther down the funnel your goal is (purchase, sign-up, download, etc.), the better CPA (cost per action) is for tracking ad spend against results.
The problem most businesses have is that they’re running a CPA funnel to cold traffic, where no recognition exists. The result is abysmally low conversion rates and an obviously high cost per acquisition. That’s not a creative problem. That’s a model problem.
Use Your Historical Data To Find Pricing Model Arbitrage
If you’re already locked into running campaigns, your click-through rate data isn’t completely useless here, either.
In fact, a high CTR on your creative actually means that CPM campaigns might cost less per click than CPC campaigns. The math is simple. If you’re paying a fixed rate per thousand impressions and your CTR is strong, your effective cost per click is far lower than you would pay bidding directly in a CPC auction. They take the risk on pricing, while you benefit from your better-than-average creative.
Again, this is where testing comes into play. Don’t assume that CPC is the more profitable model for traffic-focused campaigns. Run the math against CPM, using your actual CTR history, and you might be surprised at the results.
High-Volume Formats And Why Reach Still Matters
There is a conflict between advertisement formats that stand out and those that do not compromise user experience. For instance, full-screen interstitial ads grab attention but may annoy users if they are not timed well. On the other hand, native ads blend with the content but that may cause them to be overlooked. It is hard to say which ones are better because it depends on the tolerance level of your audience and the goal of your campaign.
If your business aims at quickly reaching your audience and maximize your brand exposure, then you should consider using high-volume formats such as pop up ad networks. They can offer such high volumes that conventional display ads cannot compete with, especially if you are trying to make your target market more aware of your brand. Additionally, the cost per placement is much lower, and the large number of impressions allows you to test different versions of your ads quickly.
The downside is that these ads are less targeted. High-volume networks work to get your ads seen by as many people as possible which is not a problem when you are running a brand awareness campaign or your product has a broad appeal. However, this is not ideal for B2B products with very specific target customers. In these cases, it is better to invest your budget in more targeted networks.
Niche Networks Versus Broad Platforms
Large social media platforms may grant you the scale and targeting capabilities you desire, but they also tend to be more costly and saturated. Niche-specific networks, such as those that are industry-focused, interest-based, or cater to particular geolocations, often generate higher-quality leads at a lower cost per impression because of less competition.
This isn’t about turning your back on the big guys, though. It’s about testing the long-tail and comparing the results based on the return on ad spend (ROAS) you’d expect from any marketing activity.
Also, the more sources you have sending traffic to your site, the more integrated your strategy needs to be. First-click, last-click, and even linear attribution models capture data that makes the case for top-of-funnel targets.
Build For Scalability From Day One
The optimal business advertising model isn’t necessarily the one that currently works best within your existing budget. Rather, it’s the one from which you can derive the greatest benefit as your spend increases, without having to overhaul your operation to support it.
Programmatic buying through a DSP gets you automated, impression-by-impression auction bidding based on real-time performance data. Because the system continually decides how much each impression is worth based on the likelihood of a conversion, your spend automatically shifts toward better-performing placements and creative, and away from weaker-performing ones. Also, ads win out less frequently on low-value inventory, essentially getting cheaper as your algorithms get better at identifying the right audiences. These all save you budget that you’d otherwise lose to waste or suboptimal spending.
Manual blocks of guaranteed inventory, PMPs on premium inventory, or direct relationships with individual publishers can sometimes still outperform the open auction available through your DSP. If you haven’t reached your scale potential, though, these usually don’t make sense given the added costs of manual operations and likely the reduced reach and targeting capabilities. Essentially any programmatic spend level is high enough to at least experiment early and see if you can get the machine to scale with an acceptable level of margin.
The hybrid approach works because you’re solving two different problems. At the top, you need volume and efficiency. At the bottom, you need precision and intent. Force the same model across both and you’re optimising for the wrong thing at every stage. Keep them separate – different metrics, different logic, different benchmarks – and your overall ROAS reflects that discipline.