So you planned your programmatic campaign to the last details. Before you roll out the campaign, you would have -
- Identified your Audiences
- Set the Targeting parameters accordingly - Here is a definitive guide to audience targeting in programmatic advertising
- Finalized the Message
- Uploaded Creatives - We compiled a detailed guide to available creative types and their role in Programmatic Advertising
- Calculated the Optimal Bid
- Set up the Campaign Flow
- Established the KPIs
Then you go live with your programmatic campaign.
Your campaign starts to receive a good number of Impressions and Clicks.
But you are not satisfied!
You ask yourself, "what should I do in the campaign so that it fulfills the goals (KPIs) I want to achieve, and gives an even better Return On Ad Spend (ROAS) to the business?"
Let us try to answer these Whats, Hows, and Whys
Programmatic provides a vast reach to marketers and advertisers. Technically, everyone on the internet can be reached through programmatic advertising, but not everyone will belong to your target pool.
Whether you're using Kritter's Demand Side Platform or any other DSP available in the market, the fundamentals of optimization remain the same. Marketers, media planners, and advertisers generally employ one of the two broader strategies for targeting their prospects -
The first approach is when starting, they set up the campaign with a limited budget covering a broader pool of audiences with limited targeting controls. Gradually they introduce the targeting restrictions based on what set of audience is interacting better, working toward the KPI. That's the top to bottom approach.
The second approach is the bottom-up approach, where you introduce all kinds of targeting to get to a very narrow audience and slowly loosen them to get the desired results. It might be hard to keep track, and campaigns will end up spending more than expected. Hence this is not recommended.
Optimization helps bring forth the audience with the highest intent to buy your products or services or interact with your campaigns.
Where To Start, Which Module To Tweak First, What To Include Or Exclude
Before we dive in, let us tell you that optimization is a systematic procedure. It should not be done haphazardly. One should not push all buttons at once. Instead, apply the adjustments in phases and wait. Take your time to evaluate the effects individually.
Also, there is no right time to optimize. It all depends on the data that the campaign gathers. Ideally, optimization should be done the second or third day after going live after seeing enough data in reporting. Doing analysis on the data sets and applying those insights too frequently or infrequently - both are not advised. One wrong setting at the wrong time may result in burning unnecessary Impressions on people not interested in your products & offerings, spending thousands of dollars.
So instead of striving to create a Fail-Safe system, it is good to have a "Safe Fail" system where even when a campaign fails to achieve the desired goals, it does so at a negligible/minimal cost.
Now let us understand the basics of campaign optimization.
- Before starting to optimize, check the options available at the Traffic Source i.e., the Ad Exchange. Make sure that you have set or activated the inventory needed
- After making the campaign live, give it some time to accrue enough data, so you're able to make a decision via a data-driven approach
- Apply High-level Campaign Optimizations and check the impact (positive or negative)
- Apply further Optimizations at granular levels based on previous analysis
- Repeat 3rd and 4th step, until you get the desired results
Changing targeting preferences too often will leave you confused without providing enough insights to make data-driven decisions. So, always allow enough time, spend, and data between changes.
The First Set of High-Level Optimizations
Programmatic ads take time to attain the delivery/spend pacing. Initially, there will be a loss of revenue (therefore, the budget cap) or lag in delivery. But when you have enough data in the reports, you can start optimizing -
# Removing a Supply Source or Site (Blacklisting) could be the first thing on the checklist. Ensure that the ads are showing on placements that are most relevant and avoid delivery at the sites and apps that don't go well with your brand.
Kritter's programmatic platform allows access to different inventory sources, with varying levels of quality and performance. Because programmatic auctions are carried out in milliseconds, it is impossible to control where your ad will show up. So, you can check reports, segmented by exchanges, and see what exchanges are serving your ads. Highlight the worst-performing exchanges and remove them from targeting. If the largest source of inventory is the worst performing, it may be worth reevaluating your targeting.
# Device/OS - There have been cases where desktop traffic has been consuming much of the daily budget when realistically, mobile traffic has been over-performing on all the KPIs compared to traffic from other devices types, or vice versa. Use the available data to create a test to break out Mobile vs. Desktop Vs. Tablet traffic and move budgets towards the better performing device types.
# Hour / Day Parting - It allows users to follow user behavior and increase budget efficiency to serve ads during the times of day and days of the week when the desired audience is most likely to click or buy.
# Frequency cap - It is done to avoid ad fatigue. It is essential to continually adjust the frequency and time between impressions that users need to see an ad to be influenced. Begin testing different frequency caps, like one impression per user every 2 hours or five impressions per user per day, or any other combination that you deem fit. This will significantly help in achieving good CTR and ensures that the campaign spreads to a broader group of audiences.
# Geo-Targeting - Definitely a great way to improve performance is targeting by geography. Not every part of the country will respond to your ad or even your product in the same way. When looking at the Geo-wise reports, make sure to drill down to not just state, but as deeper as the major cities and markets.
After testing out the first set of optimizations, evaluate both the positive and negative effects of the settings that you chose - If there are no positive effects, roll back the changes to earlier settings or investigate what is causing the adverse effects and remove it from the targeting. Maintaining a simple Excel document to track campaign tweaks will help you see what changes produce what results.
Let's say one or more tweaking helped in removing irregularities, or it gave average results. Now to get better results, move towards granular optimization.
The Second Level of Optimizations
# Bidding adjustments - At Kritter, we have our proprietary algorithms that can apply to your campaigns to reach specific CTR or CPA goals. You can choose to bid manually with a fixed CPM rate or let the system decide the cheapest cost for the inventory.
Controlled spend is crucial. Campaign Managers should decide what a realistic CPM bid to maintain both scale and performance is. The bid needs to be adjusted for a better Win Rate.
# Site Analytics - Look at the top spending sites to see if they are meeting or exceeding the KPIs. Take out domains report with the average bid price. Identify which exchanges or supply sources are providing these domains at what price. It may so happen that Exchange A is sending WSJ.com at $1.50, whereas Exchange B is sending the same at $0.75. Consequently, exchange plus domains is a combination that you should always consider exploring.
# The 80-20 Rule - This is an experimental setup to explore new optimization ideas for the campaign while keeping the budget and the broader performance under control.
This method is similar to A/B testing your landing pages, emails, or other campaigns.
Here, advertisers run two sets of Campaigns simultaneously - the first campaign will be an experiment with 20% of the budget, and the second campaign will be the actual performance campaign with controlled settings getting 80% of the budget.
The experiment campaign would inherit all settings from the control campaign except one where it keeps one of the parameters open to a broader audience.
Let's say you want to optimize your campaign for domains. You want to whitelist all the domains that are relevant to your products & messaging and, at the same time, help you reach your campaign KPIs. You employ your experiment campaign to discover such domains that provide good CTRs while later adding them to the whitelist of your control campaign.
Initially, the control campaign will have fewer domains in the whitelist, and it will affect the campaign delivery, but with more experimentation, the list size will increase, and it can be used for similar campaigns in the future.
Similarly, this method can be employed to add more geographies, demographics, device types, and other parameters that perform well with your ads into the primary (control) campaign.
This blueprint is intended to help you, how you go about the process of optimizing your campaigns to achieve better results. It is worth noting that some optimization techniques can have an impact on scalability!
Since there is no strict set of rules for optimization, if a particular technique works well and helps drive the campaign towards its goal, then keep the settings as they are.
Also, note that these steps or techniques to optimize would vary depending on the KPIs of your campaign. Typically, a Branding Campaign will have different levels to tune compared to a Performance Campaign since the goals for both are different.
So yes, keep track of all your changes. If you see a bad performance, do revert to old settings.