{"id":11233,"date":"2026-09-29T14:17:50","date_gmt":"2026-09-29T14:17:50","guid":{"rendered":"https:\/\/justfineinfotech.com\/mit-harvard-data-show-how-seo-and-paid-media-managers-become-directors\/"},"modified":"2026-09-29T14:17:51","modified_gmt":"2026-09-29T14:17:51","slug":"mit-harvard-data-show-how-seo-and-paid-media-managers-become-directors","status":"publish","type":"post","link":"https:\/\/justfineinfotech.com\/fr\/mit-harvard-data-show-how-seo-and-paid-media-managers-become-directors\/","title":{"rendered":"MIT & Harvard Data Show How SEO And Paid Media Managers Become Directors"},"content":{"rendered":"

The SEO, paid media, and digital marketing managers who become directors and executives over the next few years will be the ones who can prove AI pays for itself while keeping their teams intact. Two pieces of research, one from the MIT Technology Review and one from the Harvard Business School, explain why that combination is becoming the promotion case<\/p>\n

I recently told a joke about a grocery store between Harvard and MIT. A student wheels a cart holding 15 items into the 10-items-or-less lane. The cashier looks at the sign, looks at the student, and sighs. \u201cYou must either go to Harvard and can\u2019t count, or go to MIT and can\u2019t read.\u201d I now take it back for both schools<\/p>\n

MIT Counts The Cost<\/h2>\n

David Rotman\u2019s article in the MIT Technology Review analyzes the bet behind every AI tool in your marketing stack. Hyperscalers will spend about $750 billion on data centers this year. Gary Gensler, who ran the SEC under President Biden and now teaches at MIT Sloan, puts total AI revenue at $150 billion to $200 billion<\/p>\n

Jessica Wachter<\/a> of Wharton, a former SEC chief economist, asked how fast hyperscaler earnings must grow to justify roughly $1.1 trillion of spending through 2027. Her answer is a factor of 2.7 by 2030 to break even. If that productivity does not show up, she and her coauthor call the buildout the \u201clargest misallocation of capital in history.\u201d<\/p>\n

Search marketers have a direct stake. Rotman reports that Alphabet had revenue of nearly $120 billion last quarter and still posted a free cash deficit of about $5.9 billion, its first since going public in 2004. My inference, not Rotman\u2019s, is that a company funding AI answers from a thinner cushion than it has ever had will keep changing how those answers look and who gets cited<\/p>\n

The productivity gap Wachter needs is not visible yet. Rotman cites a survey of about 6,000 executives in four countries. Around 90% reported no productivity gain from AI over three years. Kevin Indig<\/a> has traced where those hours go inside marketing teams. Workday\u2019s research, as he reports it, finds that for every 10 hours AI saves, companies give back about four hours, fixing weak output. BetterUp Labs and Stanford Social Media Lab<\/a> found that 41% of workers received AI \u201cworkslop\u201d in the previous month, and each instance took nearly 2 hours to sort out.<\/p>\n

I wrote \u201cHow to Calculate the ROI of Influencer Marketing Campaigns\u201d five years ago, and you can use the same formula to calculate the ROI of AI. I\u2019d also argue that any AI efficiency claim in a marketing plan is unproven until someone shows the hours on both sides of the ledger<\/p>\n

Harvard Shows Where The Jobs Are Moving<\/h2>\n

Ana Elena Azp\u00farua\u2019s article, \u201cEnhance or Eliminate? How AI Will Likely Change These Jobs<\/a>,\u201d in HBS Working Knowledge summarizes research by Harvard Business School professor Suraj Srinivasan<\/a> and his coauthors, and it shows where that ledger points for your team. After ChatGPT\u2019s public launch in November 2022, job postings for occupations heavy on structured, repetitive tasks fell 13%. Postings for occupations built on analytical, technical, or creative work, the kind AI can enhance, grew 20%. The researchers reviewed postings from 2019 through March 2025 across nearly all U.S. vacancies, using ChatGPT to sort more than 19,000 tasks across more than 900 occupations.<\/p>\n

The skill lists moved too. Postings for automation-prone roles listed 7% fewer skills. Postings for roles with high augmentation potential asked more often for AI skills such as prompt writing and using AI tools. The largest declines came in finance and technology. Srinivasan\u2019s advice to companies is to treat generative AI as an \u201caugmentation tool rather than merely a cost-cutting measure\u201d and to fund reskilling. The study covers the short-term U.S. picture only, so the authors say longer-term and non-U.S. effects remain uncertain.<\/p>\n

The article does not say where SEO, paid media, or digital marketing roles land. My reading, and it is an inference to ground truth in your organization, is that the structured and repetitive parts of our jobs sit on the exposed side. Think recurring reports, bulk title tag, and ad copy production, bid adjustments, and search term cleanup. The judgment work sits on the enhanced side. Strategy, testing design, measurement, and cross-team persuasion are where a manager becomes a director.<\/p>\n

The Tension Executives Will Bring To The Room<\/h2>\n

Put the two studies side-by-side. Rotman reports that the surveyed executives expect to lift productivity by growing sales while cutting staff. Srinivasan\u2019s research argues that firms get more from AI when they build human and AI collaboration and retrain people for it. Your budget meetings will sit between those two positions<\/p>\n

I would promote the manager who can bring both. Show the productivity gain in hours and outcomes so it survives an executive\u2019s question, and show the plan for the people whose repetitive tasks the tools absorb. Rotman adds a further reason to be careful. If AI improves productivity by destroying jobs, public backlash could block the very investments the hyperscalers are counting on. I have argued that AI brands are winning consideration and not trust, and job cuts make that gap wider.<\/p>\n

How To Apply This To Your Strategy<\/h2>\n

If You Run SEO And AI Search<\/h3>\n

Keep an AI hours ledger for each workflow. Record the hours spent prompting, checking, and fixing output, and set them beside the rankings, citations, and assisted conversions the workflow produced. Compare your rework rate to Workday\u2019s four in 10, and take any gap to your director as a finding. Move the freed hours toward the work Indig says gets crowded out by internal AI tools, which is publishing, earning mentions, and strengthening the brand that search engines and AI systems recommend. Track citations across several assistants every month so one platform\u2019s change shows up in your own data.<\/p>\n

If You Run Paid Media<\/h3>\n

Ask your platforms and agencies which AI features in your campaigns are bundled or discounted today, and get pricing terms in writing. Judge automated campaigns on incrementality with a holdout test, not only cost per result, because programmatic taught us that efficiency alone can hide waste. Write down which paid media tasks are structured and repetitive on your team, and take the executive team a plan that moves those hours into creative testing and measurement.<\/p>\n

If You Run Digital Marketing More Broadly<\/h3>\n

Build workflows so a cheaper model can be swapped in, since Rotman lists cheaper, good-enough models and small local models among the competitive threats to frontier ones. Keep your prompts and a small set of test tasks in files you own, and run one smaller model against your current tool on a real task. Then build a one-page AI scorecard for your next budget review with hours saved, hours given back, the outcome metric that moved, cost per workflow, and the fallback model. Add a reskilling line to your budget request covering AI literacy and working alongside AI, which Srinivasan\u2019s research says employers now want in augmentation-prone roles.<\/p>\n

Nobody can say when the AI bubble bursts, and Rotman does not try. The manager who can show the hours, the outcomes, and the plan for the team will hold a case that works whichever year it happens. MIT can count. Harvard can too. So should we<\/p>\n

More Re<\/p>\n