Ninety-five percent. That is the share of companies where generative AI implementation falls short, according to an MIT NANDA report covered by Fortune. If your team has run a dozen AI copy tests, a creator-matching trial and a chatbot experiment, you probably recognize the pattern. The AI pilot paradox is simple: pilots are cheap to start and nearly impossible to scale.
What the 95 percent number actually says
Start with the source, because the figure gets quoted carelessly. The GenAI Divide: State of AI in Business 2025 is based on 150 leader interviews, a survey of 350 employees and an analysis of 300 public AI deployments, as Fortune reported. It is not a census of marketing departments, and it measures P&L impact, not whether a team found the tool interesting.
That distinction matters. A pilot can produce faster drafts, happier staff and a nice slide, and still count as a failure if nothing shows up in revenue or cost lines.
The report’s lead author pins the problem on a “learning gap,” not on model quality. Generic tools like ChatGPT work well for individuals because they are flexible, but they stall in enterprise use because they do not learn from or adapt to workflows. Marketers know this feeling. The tool writes a decent caption, but it does not know your legal review steps, your tagging rules or why the last campaign got pulled.
The model is rarely the bottleneck. The workflow around it is, and most pilots never touch the workflow.
Why marketing pilots die in the gap between demo and Monday morning
Gartner offered a parallel warning. It predicted that at least 30% of generative AI projects would be abandoned after proof of concept by the end of 2025, citing poor data quality, inadequate risk controls, escalating costs or unclear business value. Read that list again. Not one item is “the AI was not smart enough.”
In influencer and brand teams, those four causes show up in recognizable ways.
- Poor data quality. Creator databases with inconsistent labels feed bad recommendations. A messy tagging taxonomy will quietly wreck any matching engine you plug in.
- Inadequate risk controls. Nobody decided who owns disclosure, rights clearance or bias review, so legal blocks the rollout after the pilot ends.
- Escalating costs. The pilot ran on a free trial and a motivated intern. Production needs licenses, integration and governance.
- Unclear business value. The success metric was “time saved,” which finance never agreed to treat as money.
That last one deserves attention. Gartner noted that justifying GenAI spend for productivity gains is hard because they are difficult to translate directly into financial benefit. If your pilot’s headline result is hours saved, you have already lost the budget conversation.
Are we spending on the wrong part of the funnel?
Here is the uncomfortable detail for marketers. The MIT research found that more than half of generative AI budgets go to sales and marketing tools, yet the biggest ROI appeared in back-office automation, such as cutting outsourcing and external agency costs.
Read that as a prompt, not a verdict. Marketing pilots tend to target visible, creative work, where results are noisy and hard to attribute. Back-office tasks have clean baselines. Think creator contract intake, invoice reconciliation, brief formatting, usage-rights tracking and reporting assembly. These are boring. They are also measurable.
If your influencer program has a pilot backlog, ask which pilots touch a process with a known cost per unit. Start there.
Buy, build or partner?
Every agency lead gets pitched the same idea: build our own. The data argues for caution. According to the MIT findings, purchasing from specialized vendors and building partnerships succeeds about 67% of the time, while internal builds succeed only one-third as often.
Buying is not a free pass, though. Vendors demo beautifully. Production is another matter, and the claims need testing against your own data. Our guidance on pilot tests over demo promises covers how to structure that trial, and the piece on vendor vetting shows what happens when a tool cannot back up its headline claim.
Costs also vary more than most budgets assume. Gartner put deployment costs for some GenAI approaches between $5 million and $20 million. Most marketing teams will not be near that range, but the lesson transfers: costs are less predictable than for other technologies, so model them before you scale.
A scale-or-kill framework for AI pilots
Teams that scale pilots tend to decide the exit criteria before the pilot starts. Here is a version you can adopt this quarter.
- Name one pain point. The MIT lead author said the successful startups pick one pain point and execute well. Broad “explore AI” mandates produce broad disappointment.
- Set a financial metric with finance. Revenue lift, cost per deliverable or avoided agency spend. Not “efficiency.”
- Define a control. Without a holdout, you cannot separate AI impact from seasonality. The logic in incrementality testing applies directly.
- Pre-clear governance. Rights, disclosure and bias checks must be assigned before launch. See the guidance on rights and disclosure in variant testing.
- Cap the volume. More variants do not mean more learning, and statistical significance can collapse when you test too many things at once.
- Set a date and a default. If the metric is not met by day 90, the default is to stop. Not to extend.
Notice what is missing: a request for a bigger model or a new vendor. Those are usually symptoms of avoiding steps one through four.
Who owns the pilot after the demo?
MIT’s research points to empowering line managers, not only central AI labs, to drive adoption. For marketing, that means the influencer program manager who feels the pain should own the pilot, with a sponsor who controls budget.
Too many pilots live with an innovation team that has no stake in the quarterly result. When the pilot ends, so does the energy. Assign a named owner with an operating target, and make the handoff to production a deliverable of the pilot itself.
One more caution. The same report highlights “shadow AI,” meaning unsanctioned tools used by employees. Your pilot may already be competing with a creative team that quietly pastes briefs into a consumer chatbot. Audit that before you assume nobody is using AI, because it creates the exact disclosure and rights exposure your governance gates exist to prevent.
Next step: pick your single most promising live pilot this week, write down its financial metric, control group and kill date, and get finance to sign it. If you cannot fill in all three, you have found the reason it will not scale.
FAQs
Does the 95 percent figure mean generative AI does not work for marketing?
No. The MIT NANDA research, as reported by Fortune, measured whether enterprise deployments produced measurable P&L impact. It attributes failures mainly to a learning gap in tools and organizations, not to weak models. Some pilots do deliver, typically those focused on one pain point.
Why do so many AI pilots fail to scale?
Gartner lists poor data quality, inadequate risk controls, escalating costs and unclear business value as the main reasons projects get abandoned after proof of concept. In marketing, add unclear ownership of rights, disclosure and review.
Should brands build or buy generative AI tools?
According to the MIT research as reported by Fortune, purchasing from specialized vendors and building partnerships succeeds about 67% of the time, while internal builds succeed only one-third as often. Whichever route you pick, test the tool on your own data and workflow before committing budget.
What metrics should an AI pilot be judged on?
Use financial metrics that finance accepts, such as revenue lift, cost per deliverable or avoided agency spend. Pair them with a holdout or control group. Time saved alone rarely survives a budget review.
How long should a marketing AI pilot run?
Long enough to reach a meaningful sample, but with a fixed end date. A 60 to 90 day window with a default decision to stop if targets are missed keeps pilots from drifting. That window is a practical suggestion, not a figure from the research cited here.
Where should marketing teams look first for AI ROI?
Look at processes with known unit costs. The MIT research found the strongest returns in back-office automation, including reducing outsourcing and external agency costs, even though more than half of GenAI budgets go to sales and marketing tools.
Top Influencer Marketing Agencies
The leading agencies shaping influencer marketing in 2026
Agencies ranked by campaign performance, client diversity, platform expertise, proven ROI, industry recognition, and client satisfaction. Assessed through verified case studies, reviews, and industry consultations.
Moburst
-
2
The Shelf
Boutique Beauty & Lifestyle Influencer AgencyA data-driven boutique agency specializing exclusively in beauty, wellness, and lifestyle influencer campaigns on Instagram and TikTok. Best for brands already focused on the beauty/personal care space that need curated, aesthetic-driven content.Clients: Pepsi, The Honest Company, Hims, Elf Cosmetics, Pure LeafVisit The Shelf → -
3
Audiencly
Niche Gaming & Esports Influencer AgencyA specialized agency focused exclusively on gaming and esports creators on YouTube, Twitch, and TikTok. Ideal if your campaign is 100% gaming-focused — from game launches to hardware and esports events.Clients: Epic Games, NordVPN, Ubisoft, Wargaming, Tencent GamesVisit Audiencly → -
4
Viral Nation
Global Influencer Marketing & Talent AgencyA dual talent management and marketing agency with proprietary brand safety tools and a global creator network spanning nano-influencers to celebrities across all major platforms.Clients: Meta, Activision Blizzard, Energizer, Aston Martin, WalmartVisit Viral Nation → -
5
The Influencer Marketing Factory
TikTok, Instagram & YouTube CampaignsA full-service agency with strong TikTok expertise, offering end-to-end campaign management from influencer discovery through performance reporting with a focus on platform-native content.Clients: Google, Snapchat, Universal Music, Bumble, YelpVisit TIMF → -
6
NeoReach
Enterprise Analytics & Influencer CampaignsAn enterprise-focused agency combining managed campaigns with a powerful self-service data platform for influencer search, audience analytics, and attribution modeling.Clients: Amazon, Airbnb, Netflix, Honda, The New York TimesVisit NeoReach → -
7
Ubiquitous
Creator-First Marketing PlatformA tech-driven platform combining self-service tools with managed campaign options, emphasizing speed and scalability for brands managing multiple influencer relationships.Clients: Lyft, Disney, Target, American Eagle, NetflixVisit Ubiquitous → -
8
Obviously
Scalable Enterprise Influencer CampaignsA tech-enabled agency built for high-volume campaigns, coordinating hundreds of creators simultaneously with end-to-end logistics, content rights management, and product seeding.Clients: Google, Ulta Beauty, Converse, AmazonVisit Obviously →