As we enter Q4 of 2024, AI is pushing further into the mainstream, broadening its impact on the world of digital marketing.
In this roundup, we discuss the latest AI developments and tools shaking things up, from LinkedIn’s new AI-powered advertising assistant to the future of paid media in generated responses.
LinkedIn and AI
Accelerate - LinkedIn's new AI advertising tool
After a soft launch in North America in Q4 of 2024, LinkedIn is preparing to roll out their new AI-powered advertising tool.
This new option for marketers is called Accelerate - and it promises to do just that.
Accelerate is designed to boost efficiency across all areas of advertising on LinkedIn, from designing the creative, targeting an audience, and reporting.
What would take hours to accomplish manually, Accelerate can produce in 5 minutes due to the tool’s broad scope. However, as with any AI, it requires human guidance.
How does LinkedIn's Accelerate work?
To begin, you feed Accelerate the URL for the product you wish to advertise - then it gets to work on your behalf.
Using AI capabilities, Accelerate analyses your website, your LinkedIn page, and previous marketing accounts to make initial recommendations. It then uses your customer data to mock up creatives and zero in on a target B2B audience.
Once the constituent elements of a project are in place, you review Accelerate’s efforts and make any necessary adjustments before publishing.
Finally, Accelerate compiles performance reports ready for sending to relevant parties.
Impact and limitations
Early use has proven effective for some B2B marketers in North America. For instance, Sean Johnston of Closed Loop reported “Accelerate campaigns far surpassed the lead conversion performance” of their client’s best-performing manual campaign.
But it hasn’t been an entirely smooth ride for Accelerate up until now.
In conversation with Digiday, an anonymous US ad agency confided that their clients decided against using Accelerate due to limited objective and targeting options.
LinkedIn has been improving their tool since the initial release, but it is possible Accelerate might not be ready for targeting highly specific business niches when it becomes available for UK marketers - which should be any day now.
LinkedIn and AI content
Despite a slew of AI facilities emerging on the platform - from AI generated recruiter messages to AI-powered profile building - a new algorithm report shows LinkedIn has a complicated relationship with AI content.
The report revealed that, on average, AI-generated posts see:
- 30% less reach;
- 55% less engagement;
- and 70% fewer clicks
That said, there have been some 2024 case studies that seem to contradict these results, so there this is clearly some nuance to account for. As it often does, success with AI content on LinkedIn comes down to the level of human oversight.
Using AI with a light touch can be useful in scaling your post schedule or just for resolving a spell of writer’s block. But if your audience can identify your content as generated, expect to see engagement drop.
So, before you begin using AI for LinkedIn content, we recommend studying up on the dead giveaways of a generated post. Look at enough of them and some patterns should start to jump out at you(1).
The most common structure of AI generated posts is:
- emoji/icon - Title followed by simple CTA or basic elaboration - emoji/icon
- A short textual passage (sometimes formatted as bullet points)
- A sub header followed by more text, numbered points or bullet points
- A sign off CTA that promotes media content
While you shouldn’t avoid using handy structural tools like bullet points altogether, it pays to be mindful of how they’re implemented — and always try to add your own personal flair to your posts.
It’s also worth bearing in mind that LinkedIn will now be labelling AI generated images in feed(2). So if you’d rather not have that association, best stick to images you’ve made or commissioned, or use free stock images found on sites like Pexels and Unsplash.
Read LinkedIn’s guidelines on AI content for more information on how to approach it responsibly.
SearchGPT - early performance overview
SearchGPT has been live for a couple of months now, available to a user base of 10,000. One of these 10,000 users happens to be a member of the SE Ranking team, who conducted a study to see how SearchGPT compares to Google and Bing at this early stage:
The research involved searching 36 keywords with varying intents and seeing how SearchGPT’s results compare to Google and Bing’s SERPs.
Here are the key findings:
- URL count: SearchGPT provides 10 to 18 URLs per query, averaging around 13, while Google and Bing deliver hundreds across multiple pages.
- Content citation preference: Unlike Google, SearchGPT does not cite user-generated content (UGC) forums like Reddit or Quora.
- Google traffic: 26% of SearchGPT results receive no traffic from Google, indicating potential advantages for smaller, lesser-known brands to build visibility, similar to the way other answer engines like Perplexity AI do.
- Domain duplication: SearchGPT may cite a single domain multiple times in a response. For instance, a query about retirement strategies could include several citations from Investopedia. In contrast, Google typically avoids showing duplicate domains in its top 10 search results.
- Wikipedia usage: SearchGPT includes Wikipedia in only 2.7% of its responses.
- SERP design: SearchGPT's SERP is less cluttered than Google's, displaying only URLs and their metadata without additional SERP features like videos or map overlays.
- Citations vs SERP: SearchGPT may reference sources in its generated responses that are not shown in its SERP. Conversely, Google's AI Overviews tend to cite pages ranking within the top 10 SERP positions.
- Localisation challenges: OpenAI’s search platform often stumbles when providing localised results; for example, mismatching currency to region.
- Informational focus: Currently, SearchGPT is primarily focused on informational intent and does not effectively handle navigational or commercial intent queries, offering detailed responses instead of directing users to destination links or products.
Will SearchGPT pull market share from Google?
Despite clear teething issues, SearchGPT has a lot of promise and could well be a serious Google challenger.
That’s not to say Google’s market share will go into free-fall as soon as SearchGPT is integrated into ChatGPT. On the contrary, it would likely be a very gradual redistribution of a small share segment.
However, as Google’s market dominance is so extensive, even a small shift would be a big win for OpenAI, giving the company a solid foundation to build from.
Paid media in generative AI
OpenAI advertising plans
While ChatGPT and SearchGPT remain ad-free for now, OpenAI has been open about the potential to introduce paid media into its products. It’s no longer really a question of if - but of when and how.
How ads might look in SearchGPT
In-chat native advertising
If OpenAI were to introduce ads in SearchGPT, one goal would likely be to blend them naturally into the conversational exchange with users. This could make ads feel less disruptive than traditional formats, improving the user experience if handled correctly.
However, this type of native advertising has its own challenges: if not done with transparency, it could undermine user trust, a critical component for platforms that serve as information hubs.
Another potential drawback is that integrating ads into real-time search answers might skew the information users receive.
To maintain credibility, OpenAI and other developers of AI search tools will need to ensure ads are relevant to the topic at hand and that they enhance, rather than dictate, the flow of conversation.
Multimodal advertising
SearchGPT’s multimodal capabilities also open doors for more dynamic ad formats. Besides text, there could be opportunities for image-based, video, or even interactive media ads.
Opportunities for advertisers
Caution is advised when entering any ad space for the first time - let alone an untested one - but placing ads in SearchGPT could present a valuable early opportunity for marketers.
With an emphasis on high-quality, topic-relevant ads, companies might find a unique space to connect with users in a less competitive environment before ad prices rise and the benefits become more widely known.
Perplexity AI advertising plans
We covered Perplexity AI in detail in one of our earlier AI and digital marketing overviews. If you didn’t catch that instalment, Perplexity AI is a citation-rich generative answer engine that searches high and low (not just the standard blogs and webpages) for information to satisfy user queries.
While still nowhere near as popular as ChatGPT, Perplexity continues to be mentioned alongside the biggest names in AI and earn endorsements from notable marketers.
Even with a comparatively small active user base of 15 million(3), the Perplexity team has decided that now is the time to introduce ads to their service, with rollout set to complete this quarter (Q4 2024).
How will ads work in Perplexity AI?
Options will include video ads served either above related questions (mobile) or along the side of the screen (desktop) and sponsored related questions.
The latter will trigger a conversational response as normal, however, the core message will have been approved by the paying advertiser.
To begin, ads must fall into one or more of 15 categories, including:
- Arts and entertainment
- Finance
- Food and beverage
- Health
- Technology
… The full list is yet to be disclosed.
It has been reported that Perplexity is opting for the CPM (Cost Per Mille) billing format, and the minimum spend will be something to the tune of $50 (£38.63).
How will Perplexity AI differentiate itself in the paid media market?
Most advertisers would see Perplexity AI’s small (comparatively speaking) user base as a major caveat. But the Perplexity marketing team is leveraging this to differentiate their ad space from the competition.
A slide deck Perplexity is currently circulating among US advertisers(4) includes the following statistics:
- More than 8 in 10 users have an undergraduate degree
- 3 in 10 users hold senior leadership positions
- 65% of users are in high-salary, white-collar professions
These stats show that the company is playing the quality over quantity card, emphasising the value of its users as consumers.
This messaging reframes the platform’s limited user base as a pre-targeting of sorts, bypassing the average consumer and helping marketers reach a concentrated pool of affluent decision-makers.
Being that Perplexity is positioned as the thinking person’s AI chatbot for more serious search tasks, it’s an intelligent angle that rings true. And, as mentioned earlier, the “price-to-play” is reasonable.
All things considered, the UVP (Unique Value Proposition) here is higher quality, higher margin leads for less.
Final thoughts
As AI continues to push digital marketing into new territory, it’s important to stay aware of the challenges and opportunities it poses in order to plan ahead and remain competitive.
Paid media for generative AI is beginning to take shape, but a lot is yet unknown. While early adopters may find unique advantages, the long-term effectiveness of these advertising models hinges on user acceptance and the quality of integration into existing platforms.
At TDMP, we stay on top of the latest digital marketing technologies, trends, and developments to provide our clients with the most relevant and effective guidance. Contact us today for expert digital marketing support.