The online advertising world or the SEM industry is very compact yet a complex one. Search advertising or search engine marketing as some would prefer to call it, are being mainstreamed in myriad ways with the use of artificial intelligence (AI) and machine learning algorithms. And these changes have a significant, if not a critical influence on how you approach your search advertising.
Without a contention, Google AdWords and Facebook Ads take the major chunk of the search engine marketing market. In this post, we will look at how AI has influenced both these platforms, where they are headed, and how you can make the most of your campaigns, given that every business likes to get more out of the advertising buck.
Google AdWords – Getting more value
Google has been slowly incorporating AI into its AdWords framework as a result of which there is a greater scope of opportunities. They could be getting one up on the competition, or saving time for strategic thinking by automating low-level tasks. Let us look at a few ways in which you can make more from your AdWords campaigns.
Weeding out poor performing ads
The first scenario, clicks rolling in but no sales or conversions is a disastrous scenario for any campaign. This means you are leaking money without any ROI.
The second scenario, getting bids on an advert but not clicks, your quality score [QS] goes for a toss. End of the day, your ROI takes the fall.
Imagine the time taken to filter out these ads, analyze them and pausing them. If you are running a campaign for your business, it is easy, but in a scenario where you are handling multiple accounts, you will be spending all your time here without any time to create strategy.
How about using a machine learning [ML] algorithm to weed out these ads and pause them automatically? This will help in keeping your ROI intact as well as take care of QS.
Getting into the finer detail, your ML algorithm should be able to:
- Abandon or pause the ad before it affects the ROI. Based on previous performance factors, it should be able to estimate potential gains and losses using statistical inference.
- Analyze at a molecular level. The ML algorithm should be able to factor in other individual segments like mobile traffic, non-revenue producing browsers, times and days that see a poor performance and ad variations before it pauses the entire ad outright. With this analysis, it can close or fine-tune the non-performing factors one after another with realtime analysis.
“Automation and machine learning are a big emphasis on Google currently and our industry as a whole. The more we as practitioners can leverage these tools, the more time we can dedicate to higher-level strategy and other account growth opportunities.” – Josh Brisco, Sr. Manager, Retail Search Operations @ CPC Strategy
Using dynamic ads as a part of your SEM strategy
Dynamic search ads are an ML-based tool offered by Google AdWords as a part of the platform. These ads generate automatic headlines that can capture the attention of the searcher. All you need to do, is to create and upload the list of landing pages that you want to generate dynamic ads for. The tool will identify searches that it deems fit for your landing pages and generate ad content using the phrases from your landing pages. Sounds easy, but imagine doing it manually, if you are running multiple campaigns.
Using ML, AdWords also generates ad suggestions. Ignore them at your peril because these are generated using prior performance data. Incorporating them after due diligence will boost your results.
Dynamic ads can also be created using a custom ML algorithm that should incorporate:
- External factors such as days and times.
- Mix and match the audience, imagery and copy with multivariate testing using self-learning or evolutionary algorithms.
Used with due diligence, they could save your ROI and time.
Using automated bidding for best results
Using machine learning algorithms to cap your bidding, is becoming popular today. Bidding is a very important mechanism in search engine marketing – keeping your bids low means losing out on opportunities, and keeping them high means an ROI sacrifice.
Google Adwords comes with an automated bidding interface, however, it lacks the intelligence of maximizing your ROI. Maximizing ROI would require the input of certain external factors like seasonality, consumer trends, demographics, purchase behavior, and customer lifetime value amongst others.
A good ML algorithm for automated bidding must:
- Be able to estimate the price range of each ad based on previous bids.
- Calculate and factor in the click value from each click depending on the previous click data. For this, it should factor in the website data and this is a challenge. If you have multiple landing pages and keep removing or adding new ones, previous data that is fed to the algorithm could be bad or zero. Therefore, use the algorithm for aged landing pages only.
- Identify the bidding landscape changes and results and adapt quickly to them instead of assuming that past performance guarantees future performance. This could be a big challenge for want of niche data.
Make use of existing platforms
What we discussed above could lead to an assumption that a custom ML algorithm is a necessity. However, we are speaking about marketing and hiring data scientists would be a waste of money and time when there are a platforms that could be leveraged to do the needful. Let us look at some:
- Trapica: The best part of this tool is that it can scale your campaigns by identifying the right audiences and matching them to creatives thus optimizing bidding.
- Acquisio: Built for Bing, AdWords, and Facebook ads, this ML platform helps you cut CPC and CPA while raising clicks and conversions.
- Frank: Works with Facebook ads and AdWords and connects to millions of publishers. Frank is automated to launch campaigns and optimizes them by channel, creative and target audience.
- Cognitiv: This tool uses deep learning and predicts the best spends using customization based on historic data for each brand, based on historical data.
Google has already announced its “AI-first” future when Sundar Pichai highlighted their efforts of expanding and integrating AI and ML capabilities across their products, 2 years ago.
Apart from the Google AdWords platform, choosing an alternative to create and control spending could be a challenge, however, knowing your requirements well and comparing it with the features of your chosen platform, will be a wise way to go about it.
Facebook ads – Improving campaign performance
Unlike Google, Facebook has a lot more relevant data about users. The recent scandals, albeit negative, are evidence to this fact. Concerns over private and personal data are not misplaced. Commerzbank and Mozilla have taken a negative stand over security concerns.
However, this goldmine of data that Facebook has, used with security and in a proper manner can result in much better insights about user preferences, tastes, and behavior. This will also help run relevant and personalized campaigns that can get better results, additionally, users can breathe easy without the generic ad bombardment every time they log in.
This filtered data is used by Facebook to help marketers create ad campaigns and the platform churns out a host of automated optimization options, helping marketers run effective ad campaigns.
Let us look at Facebook’s ML capabilities that can help you achieve this.
The importance of campaign objectives
The best part of the Facebook ad manager is that it wants you to specify a campaign objective, unlike any other platforms. Most other platforms give you objectives that are not specific to a simple goal. For example, one ad campaign cannot raise brand awareness, get you more reach or traffic and still get you a good ROI. One campaign for one objective is what works.
Courtesy – Marketingland
First up, while creating a campaign, you must understand what is to be achieved. The ad manager gives you thirteen options to choose from. Remember, campaign objective is not a cosmetic attribute. It will define the reach and placement of your ad. For example, if you choose “video views, Facebook’s AI algorithm will place the ad in front of people where there is a better chance of viewing, derived from user behavior.
Once chosen, the campaign objective cannot be changed post-launch. If I was doing it for my business, I would choose “conversions” and choose the sub-objective higher up as a CTA. This will help me keep my campaign, unique amongst all those ads looking for clicks.
Criticality of placement
Facebook looks at making money and this is the reason why most campaigns have placements offered this way. The more the ad serve, the more money Facebook makes, however, for marketers, choosing all placements will result in placement optimization through Facebooks’s AI algorithm. It is a win-win situation.
Courtesy – Marketingland
When chosen, the algorithm will decide where to place your ads considering the lowest CPM (Cost per thousand impressions). Placement optimization can save up to 20% costs when “Facebook newsfeed” and right-hand column are selected manually.
There are strategies that you can create using add-ons like Instant Articles, In-Stream and FAN, but brand safety becomes a concern there. Though there are filters available to categorize the placement, the newsfeed is a better choice considering the brand. A small price to pay, but worth it.
Optimizing ad delivery
If using campaign objectives was one part of the delivery mechanism, you still have options to further customize “Optimize for ad delivery”
Courtesy – Marketingland
Here, you will see options depending on the campaign objective you chose, and you can further focus on building a user persona to deliver those ads. “Value” is the latest addition to the list of choices.
Additionally, you can also use Facebook’s split-testing feature because testing is vital to your campaign.
Conclusion – Is AI SEM the answer?
AI Marketing is the future and has already turned into a mainstream feature in the search engine marketing or the PPC space. Micromanaging keywords and bids is no more a marketer’s job, it can be taken care of by simple algorithms, while strategy takes the center-stage for marketing success. Many industries are yet to include automation into their marketing and the first players to adopt AI/ML technology will see a quick boost.