All display campaigns have different goals, but the mechanism and process for maximising performance against those goals is similar regardless.
Display performs an incredible job in any digital marketing portfolio. People often think about the multiple touchpoints that users go through as a funnel, with display at the top, search and social in the middle and retargeting at the bottom. Anybody who has looked at their customer's paths to conversion online knows that's not true. People take all kinds of different journeys, and the media you consider early-stage or late-stage are nothing of the sort.
Display does build awareness. It lets you get in front of people who have never heard of your brand, your product, sometimes even your product category. But repeat exposure builds credibility and recognition. Once somebody is familiar with your brand it can support sales, remind people that you're on their shortlist, and is an engagement point all throughout the user's path.
How do you balance this?
When you build a display plan you need to carefully consider your goals, beyond impressions. Reach, frequency and engagement all interact with each other to build a picture of exposure, credibility, and even sales.
The first step is to judge the effect on those metrics of each kind of exposure that users might have. If a user sees you on website A, or device X, or after having been to your website before... those situations all dictate a slightly different expectation of what you should expect as a result.
Write down the placements, devices and situations you expect users to be exposed to your ad in, and think about how valuable that is for each goal you have. Try to assign a score out of 10.
Channel by channel.
You need to do this for each different set of buys you're likely to make. A premium placement, a mobile network, a video buy... put together the list and score them out of ten. They will all behave in different ways, so you should write down the score achieved per pound spent
Then the frequency and reach come into play. Alongside each line, write down how many users you think you can reach, and how much money that will cost. That's the level after which you will stop accruing more points as you increase your spend. Now you have a good idea about your realistic point of diminishing returns for each buy you plan to make.
Plot the curves.
Draw a graph, with spend on the X axis and your 1-10 scoring system on the Y axis. Now plot each of those buys, so that you can see the points per pound, and the spend beyond which additional budget stops contributing to more points.
Each line will show you how far your budget can go. What you want to do is find your predicted budget and optimise against those curves. You can do this by simultaneous equations, or by a lagrange optimisation, or even just by colouring in on your graph.
The end result of planning this way will be similar to the traditional method where budgets are planned incrementally: max out the best inventory, then move onto the next. The difference is that by thinking holistically instead of incrementally you know how to adapt quickly to changes in budget. You also have the advantage that you can make better predictions about KPIs, and you don't have to plan based on which inventory is cheapest, but which has the best returns/price value.
You know that some of these buys will overlap. Mobile inventory tends to be very separate from desktop inventory. Make sure that you're as up to date as possible on cross-device tracking, de-duplication and attribution modelling. Feed the results of your chosen model back into your planning and re-plan at regular intervals. Being able to change those scores for next time is a huge opportunity to drive better performance to your goals.