skip to main content

pharma

the reflex buy

the biggest line on a media plan usually goes to the surface the buyers grew up with, while the audience is somewhere else. the 2026 midterms show the same habit in public, with the spending on the record, and pharma promotions run on it too.

the plan nobody argued for

your 2027 media plan is probably most of last year's plan. tv, streaming tv, display ads, and the brand website where all of it points. chances are nobody argued for it. it was approved because it was there the year before, and the biggest lines on it are the ones everyone in the room grew up buying. those lines are table stakes. they are on every plan, so they say nothing about this one. a plan says who the audience is, where they are in 2027, and why each line reaches them better than the next one would. that is the reflex buy: the old standby gets the money, and nobody asks the questions a plan exists to answer. is it still working? are we meeting people where they are now, or where we were? does another dollar on the old surface get us a script, a referral, or a patient asking for one?

for 1 in 3 adults, an answer about their health has already come from a chatbot. that is rock health's 2025 survey of 8,000 us adults, and it is double the year before1. some of those questions were probably about your drug or device, and whatever the tool answered got there before your spot did. ask who at your company has read that answer.

people are in chats, on social, on their phones and on youtube. the spot and the billboard reach them in the hours they are somewhere else. was the plan that still leads with those surfaces argued for in 2026, or carried over? the old standbys were the lines the room already trusted.

where the hours went

where people spend their day is on the record, and it changes by age. in august 2026 streaming took 49.6% of tv viewing and broadcast 17.5%; youtube alone took 15.8%, a record for it2. the split follows age. viewers 65 and over spend about 75% of their tv time on linear tv, the channels on a schedule. viewers under 35 spend more than 60% of theirs on streaming3.

by generation, live tv climbs with age and youtube and tiktok fall4. in a june 2026 survey of adults 18 to 66, the oldest group, ages 62 to 66, gives live tv 2 hours 8 minutes a day; gen z gives it 1 hour 6 minutes. gen z gives youtube 2 hours 25 minutes a day and tiktok 1 hour 21 minutes; the oldest group gives them 53 minutes and 22 minutes.

the newest surface splits by age the same way. by pew research center's february 2026 survey, an ai chatbot has been used by 66% of adults 18 to 29, 61% of those 30 to 49, 42% of those 50 to 64 and 23% of those 65 and over5. that is reach, not time, and for any use, not only health, it is already about half of all adults.

none of this says the spot is dead. adults 62 to 66 still give live tv more than 2 hours a day. it says the answer changes with age, and a plan has to know how old its patient is.

the reflex, with a review step

the pharma plan is the reflex buy with a review step attached. medical, legal and regulatory review, mlr for short, approves the words for a spot, a banner and a brand page. it takes weeks, and it is why a claim about your product can be trusted. the chatbot's answer never went through mlr. nobody bought it as a media line, so there is no process for it. the october essay on this site asked whether pharma has a ux problem: the verified words sit behind brand sites and review cycles while people get their answers somewhere else6. a media plan that puts the money where those words cannot go is the same problem, with a budget attached. ask whether it is a comms problem too, and who has told the room where those answers now come from.

the midterms, with the receipts attached

campaign spending is public, so the 2026 midterm elections, held on november 3, show the same habit with the money on the record. political advertisers spent nearly $12 billion this cycle and put about half of it on broadcast tv7. republicans spent about $5.5 billion to the democrats' $5 billion across the cycle and lost the key races in 4 of 5 battleground states8.

2 races show where the dollars failed. in the texas senate race the losing side spent upwards of $160 million on broadcast and the winner about a third of that7. in michigan the winner spent less on broadcast than the loser, 2-to-3, and of the digital spend sensor tower tracks, put about 94% on facebook and instagram79.

the bigger broadcast check also won in georgia, north carolina and kansas7, so its size did not decide any of these races on its own, and where it lost, it had been bought the way a reflex buys. asked why, a republican digital strategist told the trade press: "we had the money, so why not throw it wherever we can?"7 a cleveland media executive called it "egregious": a reach land grab with no cap on frequency8. the head of a political ad firm said it plainer: "running the same ad in front of people 60 times just has diminishing returns."8

on the buyers' own account, the dollars that failed were the ones spent on reflex: reach bought because the budget was there, frequency with no cap, the same spot in front of the same people again. none of it was a secret. nielsen's split by age, above, was published in 2025, before the 2026 buys were made3, and the money went on broadcast anyway. the strategist's own answer says why. the budget was spent without a strategy. the test for every line on the 2027 plan is whether anyone in the room can say why it is there and who it reaches that nothing else does.

discovery

a media plan used to start with discovery. discovery asks 4 things: who the audience is, where they are now, what they do there, and what last year's lines returned. the first 2 are answered by public surveys. the last 2 are a request to your agency for its own reporting. the public half is not hard to find. the linear tv audience is real, and it is old, so the spot still reaches the over-65 audience and fewer of the people under it. but the feed reaches the older audience too. facebook, the surface the michigan winner bought, reaches 70% of us adults and a majority of every age group, including 59% of adults over 65, by pew research center's 2026 survey10. youtube reaches 84% of us adults and 65% of those over 65, by the same survey10.

if the 2027 plan was built without those 4 answers, ask who built it and when they last did discovery. does anyone do discovery anymore?

before the next buy

3 things before the 2027 plan is signed, in this order.

1. read the answers. take the 10 questions patients and doctors ask most about your brand; the call center and the medical information team have the list. type each one into chatgpt, gemini, perplexity and google, the way a patient would, and read what comes back next to your approved words. it is a day of one person's work. tell compliance first, because they will want to see the answers too.

2. look at the numbers you already have. pull the click-through rate on the display line and the cost per completed view on the tv line, and put the nielsen and pew numbers next to them. would anyone in the room fund either number if the line were new on the 2027 plan? nobody wants to be the person who blew up the plan. if those are the numbers, the plan is already in question, and someone has to say so.

3. put the next dollar on the gaps. for each question where the answer came back wrong, thin, or from a competitor, write an answer in the words a person types, with the citations already in your approved materials, and put it on its own page of your site. step 1 will show you which pages got quoted, where the tool cites anything at all. expect plain pages that answer the question as asked, not launch pages. each of those pages is a new mlr submission, not a reuse, so put the review cycle on the calendar before the buy and give the copy an owner. keep tv on the plan for the audience nielsen says is there, and write that reason next to the line.

then put one question to the promotions and mlr team this week: where does our audience get its first answer about us today, and which of our approved words are on that surface? get the answer in writing before the plan is signed.

sources

  1. 1rock health, 2025 consumer adoption survey (8,000 us adults): 32% had used an ai chatbot for health information, double the 16% a year earlier. disclosure: rock health is a digital-health investor and advisory firm.
  2. 2nielsen, the gauge, august 2026 (released october 2026): streaming 49.6% of tv watch-time, cable 18.9%, broadcast 17.5%; youtube a platform-record 15.8% of tv. the gauge measures time on the television set.
  3. 3nielsen, "why marketers should not discount age" (2025), from nielsen national tv ratings, persons 2+, october 2024: viewers aged 65 to 99 spend 74.7% of their television time on traditional linear tv; viewers aged 2 to 34 spend over 60% of their tv time on streaming.
  4. 4attest, 2026 us media consumption report (august 2026): survey of 1,000 us adults aged 18 to 66, fielded june 2026. "boomers" in the report means ages 62 to 66 only, because the sample stops at 66. the daily averages include respondents who do not use a format at all, so they are the shape of the day, not a user's day; gen x sits between the two groups the essay quotes on every line (live tv 1 hour 41 minutes, youtube 1 hour 18 minutes). attest sells consumer research.
  5. 5pew research center, "how opinions and use of ai differ by age" (june 17, 2026): survey of 5,119 us adults, february 17 to 23, 2026. share who say they ever use ai chatbots like chatgpt, gemini or copilot: 49% of all adults; 66% of 18 to 29, 61% of 30 to 49, 42% of 50 to 64, 23% of 65 and over.
  6. 6clarai, "the answer lands before your website" (october 20, 2026): the earlier essay in this series, on why the verified words sit behind brand sites and review cycles while patients and clinicians get answers elsewhere.
  7. 7campaigns & elections, "did campaigns bet too big on broadcast in 2026?" (november 2026): paxton and allies "upwards of $160 million on broadcast", the talarico side "about a third as much"; el-sayed "outspent 2-to-3 on broadcast" in a race with "about $161 million in broadcast spending"; the adimpact cycle totals (about $5.84 billion broadcast, roughly half of nearly $12 billion; $2.61 billion ctv; $1.27 billion cable). campaigns & elections is a trade publication for campaign professionals; adimpact sells ad-tracking data.
  8. 8adweek, "political campaigns spent a near-record $331 per voter, but waste forces rethink" (november 2026): adimpact's partisan split (republicans $5.5 billion, democrats $5 billion), "still lost key races in four of five states"; maine at up to $331 per registered voter; a cleveland media executive on the mistargeting ("egregious", "a broad reach land grab coupled with overt aversion to frequency capping"); the chief executive of fullpac, a political ad firm, on diminishing returns ("running the same ad in front of people 60 times"). adweek is a trade publication; adimpact sells ad-tracking data.
  9. 9sensor tower, "2026 midterms ad spend" (november 2026): el-sayed's campaign put "~94% of ad funds into meta's facebook and instagram." the share is of the digital ad spend sensor tower tracks; it measures nothing about broadcast. sensor tower sells ad-intelligence data.
  10. 10pew research center, social media fact sheet, survey of us adults conducted january 30 to june 18, 2026: 70% of us adults use facebook; by age, 69% of 18 to 29, 78% of 30 to 49, 71% of 50 to 64 and 59% of 65 and over. 84% use youtube; by age, 95% of 18 to 29, 91% of 30 to 49, 85% of 50 to 64 and 65% of 65 and over.