《LUMA Partners ：2025年全球数字生态状况报告（英文版）（90页）.pdf》由会员分享，可在线阅读，更多相关《LUMA Partners ：2025年全球数字生态状况报告（英文版）（90页）.pdf（90页珍藏版）》请在三个皮匠报告上搜索。

1、LUMA is pleased to present our 2025 State of Digital reportIn 2024,we talked about the macro landscape continuing to be a rollercoaster and how such uncertainty hampered M&A activity.We talked about AI becoming a utility,like a railroad.We mentioned governments increasing efforts to oversee Big Tech

2、 within both the US and Europe.We noted that the performance of public company Ad Tech revenue growth was improving substantially.Lastly,we laid out a framework to contextualize the market:“Survive,Revive,Thrive”-survive the depths of 2023,return to growth in 2024,and then get set to Thrive in 2025.

3、Well,we got a few things right.The macro environment did improve,interest rates came down,inflation came down,and the stock market responded positively in 2024.AI became absolute table stakes for marketers.Best of luck to those 6%of marketers with their next job.Revenue growth continued to be strong

4、 quarter after quarter and so far,with the publicly announced earnings in Q1 of 2025,growth continues to look healthy.As predicted,M&A activity rose substantially with numerous transactions across Ad Tech and MarTech.The government continues its crackdown,securing case victories against Google and A

5、pple,and appears deeply engaged in pursuing further wins against Big Tech.Europe has several antitrust initiatives as well.And of course,the most consequential case for this ecosystem is the Ad Tech lawsuit against Google.We would argue that every potential constituency benefits if Google is ultimat

6、ely required to divest what is arguably its lowest-growth,lowest-margin,and most troublesome business,including Google.Were seeing new entrants ramping up.Some companies are accelerating their efforts through acquisitions or by expanding their initiatives in advertising,while others are well-positio

7、ned to move into the space.Notably,OpenAI just hired Fidji Simo,formerly of Instacart and Facebook.Given her deep advertising experience,that move seems particularly telling.So,the Thrive was happening.Ad Tech,MarTech,and the Nasdaq were all up substantially until February.The announced US tariffs o

8、n Canada and Mexico and then all countries added significant uncertainty to the markets,erasing a year of public market gains.We would note that the markets have returned significantly over the past several weeks leading up to DMS.Lets think about what the impact of a draconian tariff regime might b

9、e.Yes,theres the shock and awe of the market reaction,but perhaps the biggest concern is the hit to business confidence.We are encouraged by the recent tariff deals,but if those dont further materialize,substantial tariffs could disrupt the supply chain,which may ultimately impact advertising spend.

10、On the positive side,all the major trends AI,CTV,commerce media,and media fragmentation remain firmly in play.And historically,major negative disruptions,have often accelerated them.Every year as we prepare our State of Digital we reflect on the most important trends impacting the ecosystem.Often th

11、ese end up being driven by specific channels(open web,CTV,Commerce Media,etc.),with discrete trends driving each.This year,we observed that in fact there are three core and intertwined forces that are driving the entire industry forward,regardless of channel:the Proliferation of AI,the Shift to Outc

12、omes,and Evolving Media Consumption.Evaluating the potential applications of AI to advertising,we tend to think of them within a hierarchy.At the base-level you see daily tasks and functions,such as workflow,audience data,measurement,and media buying.One could describe AI as software that optimizes

13、these existing tasks primarily for efficiency.The next layer involves AI software designed to enhance both efficiency and effectiveness across content,ad creatives,and communication.Next up is a layer which represents a new consumer interface that has the potential to completely reconstruct search a

14、nd internet navigation.Lastly,is the unknown unknowns.What are powerful AI solutions that we havent quite thought of yet?With respect to the LLMs,the major platforms have largely been subscription-based business models.We believe its inevitable that they will lean into advertising and commerce due t

15、o their massive user engagement,rich intent signals,and high personalization potential.These companies raised hundreds of billions of dollars and will need to demonstrate that they can monetize all that time and attention.Were seeing early movers and believe next year well be talking about the subst

16、antial initiatives of these platforms as they expand into these new revenue streams.As it relates to productization of AI in Ad Tech,we have seen these tend to manifest in two forms:CoPilots(humans completing tasks with the assistance of AI)and Agents(humans delegating tasks to AI).One of the bigges

17、t opportunities in applying AI to any sector is bringing efficiencies to the workflow/human services element and advertising is certainly no exception.Advertising,as an ecosystem,is heavily service-oriented and people-driven,with specialized teams handling planning,targeting,buying,optimization,and

18、reporting.While these teams may communicate,their work is often siloed,with plenty of manual processes throughout.One can easily imagine a world where AI re-shapes this dynamic,one thats not only highly disruptive to existing team structures,but also incredibly powerful in streamlining how work gets

19、 done.By deploying intelligent agents across these traditionally siloed,human workflows,the entire advertising process could become significantly more integrated,efficient,and effective.One can easily imagine a world where AI re-shapes this dynamic,one thats not only highly disruptive to existing te

20、am structures,but also incredibly powerful in streamlining how work gets done.By deploying intelligent agents across these traditionally siloed,human workflows,the entire advertising process could become significantly more integrated,efficient,and effective.Were all familiar with the Jeff Bezos quot

21、e that has become emblematic of Amazons strategy,“Your Margin is my Opportunity.”In a world of AI intersecting with Advertising/Ad Tech,we see services becoming the focal point for disruption and disintermediation.Said otherwise,“Your Services are my Opportunity.”Data is critical to AI,and we have b

22、een talking about this trend of third-party data shifting to first-party data for the last five years as the ecosystem prepared for a world with significantly less 3rdparty data.After five years of being told by Google that cookies are going to go away imminently,in fact they are not.Despite this ch

23、ange in Googles position,the ecosystem has moved on,and whether cookies are here or not,just does not matter to the ecosystem anymore.This trend towards first-party data is not reversing,despite Googles renege on cookie deprecation.Platforms have already realized the immense value of first-party dat

24、a.Organizations are increasingly learning how to harness both the data they own and the signals they can capture more efficiently.The momentum is still strong,and we continue to move steadily in this direction.The convergence of AI with the broader move to first-party data couldnt be better timed.Wi

25、thout AI,this shift would be far more challenging,particularly as we shift toward leveraging more probabilistic signals in measurement and ad buying optimization.Probabilistic methodologies can be incredibly complex and historically slow,however when combined with AI,were seeing what feels like a tr

26、ue probabilistic renaissance.As you move up the AI hierarchy,you reach the layers where benefits to efficiency and effectiveness become critical and Creative sits right at the center of that.On the production side,Creative has traditionally been a highly manual process,but with the rise of Generativ

27、e AI,theres tremendous potential to streamline and scale content creation.More important is what happens beneath the surface,where AI paired with Creative Data can drive greater effectiveness.AI enables rapid iteration,precise contextual alignment,and personalization at scale,ensuring that Creative

28、not only looks good but truly resonates with the right user in the right moment.Another trend weve been seeing is a shift in how brands communicate,moving closer to where consumers are already spending time vs.driving them to a landing page or site.AI will play a critical role in enabling this trans

29、ition.Traditionally,brands have engaged with customers across various touchpoints through paid,owned,and earned media.Think search,social,push notifications,SMS all typically designed to drive traffic back to one central hub:the brands website,where they have full control over the message and experi

30、ence.With the rise of AI,brands now have a growing opportunity to engage in true one-to-one conversations with customers where they are on Social and Messaging platforms.This means not only collecting valuable information and delivering personalized content,but in some cases,also transacting.The suc

31、cess of Manychat is emblematic of this trend.LUMA advised Manychat on its recently announced$140mm Series B financing led by Summit Partners.Manychat powers these interactions for brands,creators,and prosumers to communicate their audiences on the platforms those consumers spend time(social apps and

32、 messaging).The verdict is clear:sometimes new trends appear as shiny objects,only to fade without delivering real impact.AI,however,is not one of those trends.It is a true game changer.In fact,Mark Zuckerberg went even further,suggesting that AI-driven productivity could make advertising an even la

33、rger contributor to GDP,transforming a relationship that has remained stable for decades.There remain open-ended questions around the potential disruptive impacts of AI on the industry.IP credit who gets the credit for an AI summary that utilizes IP content if the publisher didnt cut a license deal?

34、Agentic governance what guardrails do we put on the agents that are going to(autonomously)complete tasks and workflows on behalf of ecosystem constituents?Search/Web Disruption what are the implications for the publisher ecosystem should the top of the funnel shift entirely?Lastly,the Innovators Dil

35、emma how do incumbents embrace this technology while also ensuring the viability and success of their existing businesses?There is a narrative that suggests this conflict in Ad Tech between AI-First companies(David)and the scaled Ad Tech incumbents(Goliath).A pervasive sentiment is that these AI-Fir

36、st companies are going to the render the incumbent companies useless.Yes,theyre going to be disruptive but suggesting they will kill incumbents ignores one of the single most powerful forces in business:inertia.Weve spent a lot of time thinking about what truly drives success for AI in this space.At

37、 its core,you need the AI engine or product itself,which in turn requires data the lifeblood of AI.While these two components are essential to get started,as noted,in Ad Tech,they dont mean much without access to both Demand and Supply,and ideally at scale(this is where inertia comes in).Scaled Dema

38、nd and/or Supply access are crucial for effectively applying and leveraging AI,as they create a continuous feedback loop that fuels further data into the system,adding value to the AI engine.There are unique aspects to all components of this flow,from various demand or supply side access points,to u

39、nique AI models,and data strategies that range from owning unique,valuable data to providing strong data processing/cleaning built for AI utilization.It becomes even more interesting when you introduce agents into the equation.Agents are only effective if theyre fully integrated into the ecosystem.T

40、hey need to be able to interact with the underlying systems and adjust the necessary parameters to be effective.However,if you have AI or agents running critical advertising functions,both demand and supply need to trust that the platform is running in the best interest of the customer and is workin

41、g correctly.Gaining trust and already having scale is where these coinciding forces become truly powerful.Scale brings in a wealth of data,which feeds the ecosystem and drives greater impact.Were already witnessing this with major platforms like Google,Apple,and AppLovin,where their shift toward AI

42、models,even at scale,has enabled them to significantly accelerate growth.While there may be“trust”concerns broadly related to these platforms,for the marketers leveraging the systems,there is currently significant trust in driving results.Given these evolving dynamics,the only option is to lean in.O

43、ur advice as it relates to AI:Build Fast or Buy Fast.You need to get to market quickly,because if you dont,your competitors will.AI has the potential to substantially shorten the startup lifecycle.Traditionally,building infrastructure,sales teams,and a go-to-market strategy could take anywhere betwe

44、en 5 to 10+years.However,AI startups can accomplish this in a fraction of the time and with significantly less capital.A great example of this is Perions recently announced acquisition of Greenbids,an AI-enabled media buying optimization business that launched in late 2022.The Outcomes Era is here a

45、nd represents an improvement to the historical proxy-based approach.Weve been talking about this for over 10 years.The advantages of migrating toward a performance model are twofold.First,it helps mitigate issues like fraud,bots,and viewability.When youre only paying for the outcome,those concerns b

46、ecome less relevant.Second,it creates a much stronger business model by combining scale and automation which boosts operating leverage.Back in 2018,we discussed a fundamental shift taking place.Gamers were among the first to understand the relationship between CAC and LTV and apply it to marketing.S

47、oon after,app marketers and then direct-to-consumer brands adopted this outcome-based approach.We predicted that eventually all marketers would follow suit,treating performance advertising not as a discretionary expense,but more as a cost of goods sold.So,why is this shift accelerating now?Its a com

48、bination of factors:a changing data landscape,increasing advertiser demand driven by CMOs pushing for accountability and efficiency,publisher need for new demand sources,and most importantly the enablement provided by AI.Much of this transformation was triggered by the collapse(or anticipated collap

49、se)of third-party data.This shift forced many of the top advertising platforms to adopt new strategies,which ultimately moved them off bottom funnel deterministic measures and forced a new paradigm tied to outcomes.This was most evident in some of the mobile app platforms,who were directly impacted

50、by the loss of IDFA as Apple shifted to iOS 14.The near term impact on companies like Meta,AppLovin,and Google was significant.But they adapted quickly by pivoting to AI and embracing more probabilistic,deep-learning models to predict outcomes and measure based on Conversion APIs(CAPIs),rather than

51、simply relying on deterministic 3rdparty tracking.Historically,deterministic measurement was the default.It was simple and convenient,focusing on bottom-of-the-funnel metrics like last-click attribution and(MTA)models.These approaches worked well for platforms that could support them,but they also c

52、reated significant problems.They introduced perverse incentives,were easily manipulated,which often encouraged behaviors that didnt align with true performance or long-term value.In short,it wasnt an ideal or sustainable way to operate.UPDATE IMAGE TO ONLY HAVE DETERMINISTIC HEREMeanwhile,probabilis

53、tic methods have always been possible,though they were difficult to adopt at scale because of their complexity,cost and time intensive-nature.They required pulling together large,disparate datasets,building complex models,and extracting insights and recommendations,all of which was slow,expensive,an

54、d(most critically)disconnected from the day-to-day media buying and optimization process.AI enables probabilistic approaches to become faster,more accessible,and more integrated into real-time media buying.As a result,methodologies like MMM and Incrementality are becoming far more practical and scal

55、able.Thats why were seeing a surge in companies focused on this space and a broader shift from static insights and recommendations toward predictive measurement and predictive buying.As these systems begin to iterate and learn together,we move toward the next frontier:Predictive Outcomes.Were alread

56、y starting to see the traditional boundaries between data,measurement,and media execution begin to blur.Over the past decade,these functions were largely separated data platforms supplied execution and measurement companies,measurement firms handled attribution and analytics,and media platforms exec

57、uted campaigns.But thats changing.As companies re-evaluate how they use their data,theyre realizing theres significant value,not just in analyzing performance after the fact,but in pushing data and signals upstream,into the pre-bid and predictive phases of the process.This creates opportunities to d

58、rive the optimization and collapses some of the distinctions between measurement and execution.Weve seen a strong trend of companies converging across the Ad Tech stack.Execution platforms are looking to add more data and measurement capabilities,while measurement companies are moving closer to opti

59、mization.In just the last six months,several deals reflect this shift.Broadly speaking,the shift toward outcomes is reshaping how we define performance in advertising.Traditionally,performance was seen as bottom-of-funnel,focused on low-cost inventory,short-term actions,and deterministic measurement

60、.Meanwhile,brand advertising sat at the top of the funnel,emphasizing awareness with high-value media and relying on proxies like reach and frequency.Whats exciting now is the opportunity to use AI and probabilistic methods to bridge that gap.With the combination of AI and Probabilistic methods,we c

61、an start linking brand signals directly to outcomes,using predictive capabilities to make brand buying more outcome-driven.Ultimately,this means advertisers can measure impact based on actual outcomes,identify which KPIs are predictive of success,and shift from buying impressions or audiences to buy

62、ing against KPIs sometimes with guarantees aligned to business results.Long-term,this could shift all ad spend towards a cost of goods sold vs.a discretionary expense.With the shift toward outcomes,the dynamic between CFOs and CMOs is evolving.CMOs are under mounting pressure to demonstrate ROI,a tr

63、end underscored by a recent IAB study showing that 70%of Brands are feeling pressured to demonstrate ROI.When connecting spend to outcomes,CFOs and CMOs can come together to align on how best to leverage advertising spend to drive real business results.Evaluating the channels that are leaning into O

64、utcomes,Mobile App stands out as the early adopter of performance monetization and user acquisition models.There are two critical growth channels in Digital Advertising that are similarly leaning into Outcomes in a big way:Commerce Media and CTV.Commerce Media,with its close ties to the transaction

65、is inherently outcome focused.Meanwhile,CTV,which historically is historically very brand/awareness focused,is increasingly becoming a space for innovation and growth towards Outcomes.Everyone in the CTV ecosystem is leaning into this opportunity,both traditional TV players and CTV first platforms.W

66、ith respect to Commerce Medias role in the ecosystem,you cant measure outcomes without visibility into transactions you need to see the commerce happening.Thats fundamentally the goal of advertising,and the closed-loop attribution enabled by Commerce Media signals is incredibly powerful.Whats change

67、d over the past few years is the permission structure:companies that have access to these transaction signals and customer relationships are now realizing theyre allowed to start advertising directly.These dynamics have enabled Commerce Media to evolve into a massive industry,generating around$60 bi

68、llion of spend in the U.S.alone.To date,much of this energy has been focused on O&O solutions,whereby retailers and other commerce centric consumer platforms have added advertising to their own user experience.However,revenue isnt evenly distributed.The majority has aligned to the top players,such a

69、s Amazon and Walmart.These two platforms are heavily focused on owning this space from a first-party technology perspective.They have shifted from primarily making money on the margins for selling goods to consumers to one where the technology/platform they provide to brands is becoming the primary

70、financial driver of their business.The remaining players account for roughly$9 billion in spending.Many of these companies entered the space with great enthusiasm,but that excitement is beginning to wane as they face the realities and challenges of operating as media companies.The mass fragmentation

71、 of the long-tail of Commerce Media networks,paired with a lack of standards,low degrees of optimization,and reliance on third-party vendors makes it an ecosystem ripe for(and in need of)aggregation.Consolidation of mass fragmentation becomes more important as you think about the bigger opportunity

72、in Commerce Media taking that data and moving it off platform.To drive outcomes at scale,you need to be tying commerce signals to all these different channels,not just on O&O Commerce Media.Extending data to the Walled Gardens,CTV,and the Open Web enables Omni-Channel outcomes at scale,and the major

73、 players have been doing this.Amazon has done this from a first-party technology approach.They have their own DSP and are investing heavily into growing their market presence there.They have multiple CTV assets across hardware,distribution,and content.And very interestingly in the last few years,the

74、yve started driving a lot of partnerships with the major social platforms,even as they are becoming more competitive with those Walled Gardens emblematic of how critical Commerce Signal is.Others in the industry are following Amazons lead.The largest companies with access to consumers are announcing

75、 initiatives inventory and data partnerships to aggregate their scaled,unique data and bring it to bear across the ad ecosystem.Rather than acquire the technology,most are doing this via partnerships,though there are a few exceptions,such as Walmarts acquisition of Vizio to attain O&O capabilities i

76、n CTV.CTV is moving toward Outcomes.Just consider whats happened over the past few years as framed by an Economics supply and demand chart.With viewers shifting toward SVOD,there was initially a shortage of premium ad inventory compared to linear TV.Fast-forward to 2022-23 when the preponderance of

77、SVOD vendors leaned into ad-supported models to drive greater monetization on top of subscriptions.Proliferation of FAST as a category and these AVOD tiers,especially Amazon Prime Video,contributed to a deluge of inventory.Where we sit today,CTV is no longer a supply-constrained channel.With continu

78、ed shift of eyeballs to CTV(from Linear)and the rise of platforms like YouTube and Prime Video,more and more streaming platforms are leaning into performance as an opportunity drive revenues by reaching a new category of advertiser.Where people spend their time shows a significant and growing concen

79、tration within both Walled and Hedged Gardens.Historically,Social,Streaming Distribution,and Subscription CTV were consumed and perceived(by the ad ecosystem)as separate,siloed ecosystems.That is changing.These platforms are aligning and increasingly focusing on video entertainment,all supported by

80、advertising.Content from these platforms spans short-form and long-form,though they are increasingly intertwined and directly comparable,with some companies,like YouTube and Spotify,operating across the entire spectrum.The popularity of short-form video has skyrocketed driving platforms such as YouT

81、ube,Meta,and TikTok that did not have clear competitors to begin competing directly.Weve also seen an increasing convergence around long form,video podcasting at the same time the medium becomes increasingly credible in the media landscape.In fact,the recent election has been dubbed the“podcast elec

82、tion”due to the significant influence of this content.These too are bleeding into short-form,with podcast clips emerging as growth format in short-form videos.Overall,theres a clear convergence happening around this new form of media.Part of this trend has been driven by the platforms themselves,as

83、they increasingly focus on increasing consumer time spent and gearing them towards entertainment over“social”.If you consider Instagrams current use case,its no longer primarily a“social platform”where you connect with a network of friends.It has shifted to focus heavily on an Interest Graph,which h

84、as more to do with what youre watching than who you are connected too.In fact,in 2024 just 7%of peoples time on Instagram was spent viewing their“Friends”content.User-generated content has also made its way into the Living Room,with YouTube leading in share of TV viewership according to Nielsen Gaug

85、e data.In fact,TV is now the primary device for consumption of YouTube content.We talked years ago about how the definition of TV would be recategorized.The recategorization was historically driven by supply constraints and the inclusion of other Premium Video delivery platforms,but today is driven

86、by a convergence of these platforms.When YouTube is the most watched app in the living room and consumers spend hours watching clips on their phones,it shifts the consideration dynamics between“UGC”and“primetime”and forces a re-definition of what should be considered“premium”video entertainment.Unsu

87、rprisingly,the time,intent,and attention devoted to these channels have driven massive ad spend.Today,Social Platforms and CTV Distribution channels collectively account for over$300 billion in advertising spend.As all these platforms evolve into forms of video entertainment,Creators become more cen

88、tral to content consumption.Were seeing a convergence in format,but at the same time,a fragmentation around content production.On Instagram,the top five accounts collectively reach over 2 billion followers,and on YouTube,the leading personalities generate weekly audiences that surpass many TV shows.

89、However,when you take a step back and look at what Creators are earning,the numbers are small from an advertising perspective.In 2025,Influencer Marketing spend is projected to be just over$20 billion.Despite the massive scale of the platforms and Creator audiences themselves,creators are collective

90、ly seeing a magnitude less than the platforms.Historically,weve viewed platforms through which Creators reach their platforms as Walled Gardens controlling the customer touchpoints,owning the data,and dictating what users see through their algorithms.But increasingly,Creators do not see these platfo

91、rms as Walled Gardens.Instead,they view them as distribution channels.As content formats converge,Creators are operating across multiple platforms.As we discussed last year,Creators are actively guiding their audiences across platforms to capture more data and deepen engagement.This convergence is a

92、lso forcing the platforms to become more stable,as they need to capture the mega creators and their respective audiences.Creators hold significant influence over consumer intent,particularly among emerging buyers.This is evident in the fact that Gen Z and Millennials discover most of their products

93、through these platforms and are increasingly buying products on Social.Unilevers new CEO recently announced that in 2025,the company aims to adopt a social-first advertising model with 50%of spend on social platforms(up from 30%)and a 20 x increase in the number of creators they work with.Notably,th

94、is comes from a CEO who was previously the CFO,underscoring that one of the worlds largest advertisers sees Creators as critical drivers of real business outcomes in todays ecosystem.Historically,the challenge with Creators has been the difficulty in scaling the channel,leading to confinement at the

95、 very top and bottom of the purchase funnel sponsored content at the top and affiliate marketing at the bottom.The landscape is also highly fragmented,with non-traditional formats that are hard to compare to conventional ad buying,thus separating the budget from other channels.Each campaign demands

96、bespoke creative,making the process complex and resource-intensive,and buying has typically been limited to individual platforms.With AI and the maturation of this ecosystem,those constraints are beginning to change.AI is now enabling better targeting,more effective measurement,and improved workflow

97、 management.Additionally,creative production is becoming easier,as is evaluating it to ensure it aligns with brand goals.These dynamics unlock the opportunity in Creator bridging the top and bottom funnel and the potential to capture a larger share of ad budgets(i.e.,shifting Creator from the Influe

98、ncer Marketing budget to the much larger Paid Media budget).Historically,the Walled Gardens have captured most of the value from growth of the Creator ecosystem,as brands spent into the platforms and creators pushed content into the platforms.As Creators increasingly work across multiple platforms,w

99、ere entering a new era of fragmented media especially in the age of AI.While information is available everywhere,the real attention and time are largely focused on these platforms that are increasingly becoming points of distribution across which(1)Brands can tap into Creators audiences and(2)Consum

100、ers can follow and engage with their favorite Creators.This evolution presents a massive Ad Tech opportunity to enable the shift of Paid Media dollars to the Creator ecosystem at scale,leveraging AI to deliver Outcomes across the highest attention platforms and most trusted purveyors of content.Lets

101、 summarize the implications of these three converging paradigms.Number one,change is about to accelerate dramatically.Weve discussed this in previous DMS sessions,but now it feels like it is shifting into a higher gear.Legacy assumptions are being discarded,and we need to approach many of these para

102、digms with a fresh perspective.Ultimately,the focus on outcomes will be a positive to the ecosystem,advertisers,and publishers.These development should finally allow a separation of wheat from the chaff,addressing the long-standing over-fragmentation of the Ad Tech landscape and driving better a business model for winning intermediaries.Lastly,we believe that inorganic growth will be essential to winning in this evolving environment(a shocking conclusion from investment bankers,we know).Thanks for reading.We welcome your feedback.