Everybody is talking about AI
We’re suddenly hearing about AI everywhere we turn: Generating dialogue through natural language processing, quickly turning a text prompt into a vibrant image or even video, or even automating programming to build new experiences.
As this technology becomes more accessible to people, AI moves beyond data analysis to become a collaboration tool, helping to elevate thinking, imagination, and effectiveness in people’s work. It’s no wonder we’re also seeing people’s increasing interest in AI play out on our platforms, with conversations around AI rising rapidly over the past year.
On Instagram, conversation around artificial intelligence has risen by 173% y-o-y(1).
AI is not new
In simple terms, AI is a powerful tool that draws on large amounts of information and data, often to solve a large problem or answer one or many complex – and sometimes interdependent – questions.
And while AI may be part of a current hype cycle, the reality is that this technology has been around for a while. Actually, AI has been studied for over 70 years, since the development of The Logic Theorist, a program designed to mimic the problem-solving skills of a human that was presented at a summer conference at Dartmouth College, New Hampshire, in 1956(2).
Since then, AI has progressed from being able to process basic logic and lists of information (AI 1.0) to being able to handle tasks we now take advantage of every day (AI 2.0). This includes helping us to find another route to our destination in traffic, suggesting alternative wording in our writing or internet searches, or even adjusting and colour-correcting the photos on our phones.
A global tipping point
When asked “what’s next?”, our CMO, Alex Shultz, often says, “The future is already here, it is just unevenly distributed,” which means the next big thing is already here; it just has not yet been experienced by the masses. It’s maturing in small pockets around the world.
This was true of AI in the early days, but this technology is now available to everyone around the world. All businesses, in every region, and every vertical, can now take advantage of this powerful technology and should be planning ahead for its future application across the marketing spectrum.
A lot of the conversation around AI portrays this technology as a replacement for the human element, delivering ultimate efficiency. But the sentiment here is misleading and incomplete. Efficiency can save time, but time savings alone is not necessarily an advantage. It’s what you can do with faster and clearer learnings, enabled by AI technology, that is giving marketers new opportunities in their work.
The real step change that AI presents for marketers is to better realise their potential by accelerating, not automating, their creativity. AI allows marketers to learn from and pivot strategies with confidence. Achieving better performance and higher returns.
In fact, marketers around the world, whether they’re in positions of leadership, strategy, or execution, have already incorporated AI into their work and are using it in a variety of ways.
From automated customer data integration to bridging online and offline experiences to programmatic media buying, AI gives people across the marketing ecosystem the power to do more; to accelerate, and enhance how they create and work.
AI uses for marketers worldwide, Aug 2022, % of respondents
Automate customer interaction - 90%
Automate data integration - 89%
Personalise the customer journey across channels - 88%
Process automation - 88%
Resolve customer identity - 87%
Bridge online and offline experiences - 87%
Drive best offers in real time - 77%
Programmatic advertising and media buying - 75%
Predict customer/prospect behaviour - 74%
Improve customer segmentation/lookalike audience modelling - 74%
How AI with Meta is benefiting marketers
To create better results for businesses that use our tools, we have three focus areas of AI investments that work together:
- Infrastructure, to support larger and more advanced models
- Modelling, to provide more accurate predictions to deliver the most relevant ads to people
- Products and tooling, to optimise for higher-value business outcomes for our advertising partners
And spanning across all of this is our commitment to Responsible AI and its guiding principles.
We’re investing in foundational infrastructure and AI technology to improve the performance and efficiency of our ads system, allowing for more parallel computations, which allows larger, more complex AI models to run more efficiently.
Now, you may be asking: okay, but what does that mean for me? Well, it means that we’re more quickly able to learn which ads work and which don’t, and then better predict which will be more relevant to each person. That results in more conversions for our partners, at lower prices and ultimately increases return on investment for advertisers.
While our ads ranking system has used AI to show ads to the right people at the right time for many years, we've been focused on substantially increasing our investments in AI to improve our system's performance and efficiency.
Club Med observed lower cost-per-conversion by adopting Advantage+ Catalog Ads solution
Problem: Club Med has a strategy to personalise and automate conversion campaigns, optimising for performance.
Solution: It adopted Meta’s Advantage+ Catalog Ads solution to dynamically match its catalogue of resorts to the intended destination of potential travellers.
Impact: 58% lower cost per incremental initiated check out vs BAU; 41% lower cost per entry in booking engine vs BAU.
To help account for recent changes across the digital advertising ecosystem, Meta is creating AI models that can help fill in the gaps when data is partial or missing. These models help to improve delivery and measurement systems, without needing to rely as heavily on individual-level data.
Think of it as a loop: AI and machine learning help build better models → those models feed into delivery/measurement → improving delivery/measurement then makes the AI smarter/better about what is working (i.e. finding the right people for an ad, understanding approximate # of conversions) → this leads to even better models → [cycle repeats].
The investments Meta is making in AI modelling are also helping improve ad formats like Shop Ads, Lead Ads and Click to Message. For example, when an action takes place across the Meta platform after seeing an ad, like filling out a form to generate a lead or starting a conversation with a business, AI helps better determine who is more likely to take those steps. This information also helps improve the efficiency of those campaigns, meaning the advertiser can spend less to achieve a similar outcome, given the AI models learn more quickly.
Société Générale reduced its acquisition cost by 19% through deploying Advantage + placements
Problem: The French bank Société Générale wanted to optimise its performance campaigns.
Solution: It tested Advantage + Placements to see if improving the liquidity can drive impact of cost acquisition by running an A/B test comparing BAU and Advantage+ Placements.
Impact: Compared to BAU, 19% lower acquisition cost, 22% lower CPM, and 6% incremental reach for the same budget.
Products and Tooling
We are building AI-powered tools like the Meta Advantage suite, which is our portfolio of automated ad products. These AI-powered solutions enable automation across every step of the ads creation process and help businesses to connect with the right audience, in the right place and at the right time and helping marketers achieve better performance in their campaigns.
Marketers are gaining the automation Advantage - driving performance with fully automated campaign solutions (like Advantage+ shopping campaigns and Advantage+ app campaigns) or choosing individual automated campaign levers depending on the job they need to do (like Advantage+ creative and Advantage+ catalogue ads). We build flexible solutions so advertisers can benefit from automation where it best serves their marketing strategy.
For example, in a recent test with 31 advertisers, we found that, on average, Advantage+ Shopping campaigns improved return on ad spend by 32% compared to business-as-usual-only campaigns.
And in another study of 16 advertisers, we found that those who’d used Advantage+ App Campaigns saw a 9% improvement in cost per app action on average.
And finally, in an experiment with 118k global businesses, we found that, on average, those who used standard enhancements for Advantage+ creative saw 14% more incremental purchases per dollar spent.
In all of these instances, the AI was able to help advertisers achieve their goals more effectively and cost-efficiently.
Bricomarché used Advantage+ shopping campaigns to recruit new customers
Problem: Bricomarché explored the business opportunity behind Advantage+ shopping, to recruit new customers.
Solution: AB test to see if Advantage+ shopping could complement Advantage+ catalogue ads for broad audiences.
Impact: 5% reduction in cost per purchase; 21% reduction in cost per add to basket; 23% additional impressions.
Our commitment to Responsible AI
AI will continue to revolutionise how Meta connects people and allow for even more individualised content. But, it is important that we work proactively to mitigate unintended consequences of technological advancement in this space.
Our Responsible AI efforts are propelled by a cross-disciplinary team whose mission is to ensure that AI at Meta benefits people and society in a variety of critical ways. And we are also consulting and collaborating with outside experts and regulators to help ensure that systems are designed and used responsibly.
Our Responsible AI team has several key principles that guide and inform advancements in AI. Namely, these are(5):
- Building diverse datasets and powerful tools for more inclusive AI products
- Protecting privacy while addressing fairness concerns for different demographic populations and mitigating unfair differences
- Innovating to improve fairness in ad delivery and ensure that people of all backgrounds have equitable access to information about important life opportunities
- Generating responsible associations and mitigating problematic content associations that reflect or reinforce biases and bigotries embedded in social and semantic data
- Giving people more control over AI-driven feeds and recommendations
- Developing new methods for explaining our AI systems and how they work
- Testing new policy approaches to AI transparency, explainability, and governance, e.g. through our Open Loop governance programme that connects people across governments, tech companies, academia, and civil society to help develop forward-looking policies around the development and use of new technology.
Now is the time to explore where AI fits in your journey
Meta’s focus and investments in AI will continue to drive innovation. The result of this investment is industry-leading AI that's driving better experiences, engagement, and business results today across our apps. It puts Meta in a position to create a sustainable business model for future innovations in AR, VR, and the evolution of social connection.
So now the question is: how do you want to embark on your AI journey?
AI will lead a productivity and capabilities boom this decade. And now, it’s your turn to reinvent how you work and how you imagine, how you fuel your AI opportunity and give it the freedom to perform. The time is now, to engage with all the potential of this new era of AI-powered performance and possibility.
Pioneers | Conversations on our Industry’s Future is a thought leadership series, brought to you by Campaign and Meta, exploring emerging trends and cutting edge insights on the changes that will shape the future of the marketing and advertising industry for years to come. Click here to see more from the series.
SOURCES AND CITATIONS:
(1) Culture Rising 2023 report. Instagram data, year-on-year growth, Sep-Nov 2022 vs. Sep-Nov 2021
(2) “The History of Artificial Intelligence” Harvard University, The Graduate School of Arts and Sciences, Aug 2017
(3) Community Standards Enforcement Report, Q4 2022
(4) Q4 Meta Earnings, Feb 2023
(5) “Meta’s progress and learnings in AI fairness and transparency” Meta AI. Jan 2023