Planning for a personalized advertising campaign can be daunting, but it doesn’t have to be. Implementing a step-by-step process for how the campaign will come together will save everyone a lot of time and effort, ensuring that your company gets the most out of data-driven advertising.
Step 1 – Identify Key Data Signals
The brand team needs to explore the available data signals and identify what is relevant to the brand. This involves trying to identify possible correlations between consumers of the brand and the data signals. The question to ask would be “Does the data signal’s value, or its presence or absence, influence the engagement with or purchase of the brand?”
Step 2 – Identify Specific Variables and Granularity of Signals
After identifying key data signals, the next step is to identify how granular the data signal should be. For example, in terms of weather as a signal, is the brand sensitive to temperature, precipitation (e.g., whether it is snowing), or just overall weather conditions? Similarly, for sporting events, the brand may have more affinity to specific sporting events than to others. For example, luxury brands have a higher affinity to sports like golf and tennis, while more mainstream brands may look at sports like football and baseball since those games are enjoyed by most people.
Step 3 – Identify Trigger Conditions
The next step is to figure out what specific conditions of the data signals should trigger delivery of a specific creative or messaging. This is ideally done after steps 1 and 2. It could be done by creating an all-inclusive list of data signals and their associated variables, in a spreadsheet.
A creative agency typically creates such spreadsheets. It may develop assets for each of the trigger conditions after it has built this spreadsheet. This is an important step as it can get quite chaotic and confusing if the whole picture of what triggers to use and assets to build is not defined up front. It is also important to develop creative messaging for a default condition that does not meet any of the rules. Crafting trigger conditions also requires some thought because it is possible to end up with very few impressions delivering personalized creative if the trigger conditions are narrow.
Step 4 – Ideate and Define Creative Canvas and Variable Elements
At this point, the creative team comes in and starts to think of the creative canvas on which to deliver personalized creative and messaging. Usually this involves defining what parts of the ad will remain constant and what parts of the ad will be personalized to the user. In the simplest case, the only thing changing in the ad may be some copy. However, the creative team should consider defining other elements in the ad to be variable, such as a skin that reflects weather, an animation showing the team colors for the sporting event, or maybe skins or animations showing the geography being delivered to.
This is a fine balance because the more variable creative elements are, the bigger the matrix will get, and the longer the approvals will take. However, the other extreme is an ad that is simply not interesting enough to get attention; in fact, consumers may not even be aware that the ad is dynamic and personalized to them.
Step 5 – Produce Creative
Creative production is the process of building out all the variations of the ads and messaging. If the above steps have been diligently followed, this process is likely to go much more smoothly since the data signals, trigger conditions, and creative messaging for each would all have been decided on and approved.
Production can be very challenging if it’s done using a technology platform that is not designed for dynamic ads. The key here is that with creative production and ad serving platforms that are not designed for dynamic ads, each creative variant will have to be built as a separate ad. This could create significant issues and delays in production since each ad has to be separately developed and put through a QA process.
Step 6 – Identify Key Metrics to Be Measured
At this stage, it is usually a good idea to identify the metrics that will be measured in order to determine the success of the campaign and also to optimize creative. In modern dynamic ad platforms, the measurement process is much more dynamic, and all variable components of an ad are automatically recognized and recorded.
Key metrics for performance-oriented dynamic ad campaigns are usually click-throughs, view-throughs, and conversions. On the other hand, branding campaigns tend to focus on other metrics, like engagement time, interactions, and video views.
Unless personalized ads are highly interactive, they are often best measured using brand lift studies. Often users see an ad with relevant messaging, and it causes the brand to register, but the users may not click on or otherwise interact with the ad. A brand lift study can usually get a measurement of impact on the brand in terms of recall.
Step 7 – Identify Optimization Criteria
Most personalized/dynamic ad campaigns can benefit significantly from optimization. For campaigns that are more branding oriented, there may be multiple creative concepts or data signals being used to vary creative messaging, and so it would be useful to know which ones are the most effective. For example, if a campaign has weather-based creative and creative tied to local sporting events, it may not be clear which one is performing better, and being able to look at various metrics related to both data signals/rules will allow for optimization to produce better engagement.
Step 8 – Define Dynamic Data Signals
Certain dynamic data signals cannot be determined or programmed up front as there needs to be human involvement to interpret and come up with messaging or creative elements for the data signal. This process needs to be defined up front so the brand can take advantage of it. For example, a brand that wants to look at top Twitter trending topics and figure out what messaging to publish will need to set up a process to do so.
This is time-sensitive information; that is, if it takes more than a few weeks to publish the trend, it may no longer be a trend by the time it gets published and therefore may no longer be relevant or interesting. Brands that want to leverage these kinds of dynamic data signals should consider a newsroom-like process and team, where the team can quickly assess the news or event and then decide on messaging for it, create any needed assets, and push out the updates to the ad via the dynamic ad platform.
Step 9 – Preview the Ad and Get Creative Approvals
In general, the system being used to produce and deliver ads must be capable of showing in preview mode all the possible creative variations as well as filter by certain triggers, rules, and creative sizes/types to allow for narrowing into a specific set of creative for approval.
The question of whether someone will approve each and every permutation of a creative is really one that the brand/creative and media teams need to evaluate and explore, both in terms of necessity and feasibility. With campaigns that may have several thousand creative variations, it might be a process of sampling what works best rather than one of approving every single creative combination.
Step 10 – Handle Final QA and Launch
Dynamic and personalized advertisements require a somewhat different approach to QA than traditional static ads. The QA process has to replicate the dynamic data signals, rules, and ad serving in a test environment in order to ensure that the correct creative variants are being rendered for each data signal and rule combination. Here again, modern dynamic ad platforms have extensive capabilities to use preview tags and pages to conduct extensive testing to ensure that all the signals, rules, ads, and reporting work. These platforms also have extensive facilities to test rules and signals to troubleshoot and discover issues well ahead of campaign launch.
