A new report from ZoomRx is based on interviews with oncologists and looks at aided recall about drug features being discussed with a Rep. The top features that oncologists recall aren't efficacy:

When efficacy messaging does not include data (the report says "nearly 3 in 10" don't) then efficacy effectiveness and recall both fall.

This chart shows how often a topic leads the conversation, in other words is "order 1" in the list of topics. The number at the base of the bar charts is the average position of the topics in conversations. The low numbers here would seem to indicate that each conversation only has a few topics which is why you get no numbers above 2 (seems reasonable but you would have to ask the report authors -Ed.). 

This report from ZoomRx goes deep on oncology messaging so will be of interest to anyone with that indication.

A new report from Pipedrive is based on "1,000 respondents across a range of age groups, industries, and employment statuses." The company says that the respondents are "professionals" which are implied to be knowledge workers. There are many other types of "professionals" but we won't get into that (thank you -Ed.).

There's a gap between "occasionally" and "rarely" for those who use AI once a week but with these categories 63% say they use it multiple times a week or more often and nearly 40% say they use it daily.

LLM tools started as strictly text and data and those are still their top usage modes. Visuals appear at number five, coding at the bottom, and audio / video aren't even on the list.

This result is just humorous on the face of it. We don't know if there was context in the survey that made this answer more likely, but 23% of respondents said they use AI but don't want to get AI from others.

Also, 44% say they never use AI for communications. Respondents may be considering this only as written materials as AI can be used to research information that is then communicated by people. Regardless, there is still a stigma to AI-generated content in messaging... unless it gets good enough that audiences can't tell.

Oura filed its S1 with the SEC to register for an IPO on September 3. In that filing it revealed some market data that, presumably, is accurate. Here are some tidbits.

When looking at the members of the Oura app, the ages skew Millennial. The gender split is 72% female, 28% male.

The Oura app membership doubled from Q2 2025 to Q2 2026, but that growth is slowed down from previous periods. When we look at monthly growth, and month-over-month increases, we see a definite "let's IPO" bump and then slowdown.

No hardware product will ever have a 100% hardware-sales-to-app-membership ratio but Oura looks like it is heading towards 50%. We charted Oura's growth in both based on the filing and an initial hardware figure of 2.4 million coming into Q4 2024.

It will be interesting to see how the IPO does (not financial advice -Ed.).

A substantial report from Similarweb and Statista looks at the shopping journeys of consumers in the US, UK, and Germany. It's general ecommerce but there are some interesting findings for health marketers.

Almost four years since ChatGPT was released publicly, 49% of Americans say their internet browsing habits have changed. This is a substantial shift for a large market. Even the oldest group, 55-64 are at 44%.

Looking at the health-related categories on the three country charts (they vary by country) we see that the US has the largest share but a lower ranking than the others. The UK and Germany both have health as the second largest growth.

The report looks at the consumer journey from a few different angles. While only 11% of consumers are currently using AI while shopping, the number is growing and usage is concentrated in the middle, consideration and comparison phase.

With 73% of pharma managers either running or planning to run DTP programs, this journey is becoming more important for 2027.

A new report from the Video Advertising Bureau (VAB) looks at how podcast spending is increasing. This may seem strange at first as podcasting is considered an audio medium, but 88% of weekly podcast listeners consume podcasts via video channels such as YouTube. 

The 55+ group has reached 44% in monthly podcast listening and other age groups are over 60%. No matter your target, there is now a podcast market available.

When we look at the industries at the top of the investment ranking (presumably from the US) we see categories that typically demand considerable research before purchase. This may make them more aligned with pharma than, say, CPG.

Forecasts by Emarketer show podcast spending growing inside a growing channel (digital audio). 

These numbers are a small fraction of overall US media spend but the trust that comes with host-read ads on podcasts is high and worth a look for 2027.

Array surveyed 253 hospital and health system leaders and 261 clinicians to determine the state of behavioral mental health care presumably in the US though that isn't stated.

Nearly all, 95%, of the clinicians surveyed say they are "considering reducing their hours or leaving their role due to burnout related to behavioral health patient volume or complexity."

Virtual care can help, and 84% say that it is actually better than face-to-face care.

Nearly all respondents said that AI will eventually save them time. When asked how that will likely happen, workflows topped admin.

OpenEvidence included some data with its recent announcement of a partnership with an oncology center in the US. 

These numbers, especially the one from the NOHARM study, are not the same as previously published. See the Wire story on NOHARM to find the top performers (hint: its AMBOSS and Doximity).

A case study at Northwell Health showed the benefits of using AI in subspecialty matching for patients. The benefits included:

The increase in speed to a specialist for patients can make all the difference for their outcomes.

The FDA is considering a three-part approach to evaluating patient-facing health AI. The direction emerged from an August 28, 2023 meeting between a senior FDA official and Amazon, which presented an LLM benchmark designed to evaluate health AI agents.

The components are:

The benchmarking and comparison monitoring sound great, but no details were given about how they might be approached.

When looking at streaming on CTV devices, Netflix is the non-YouTube platform with the most hours. 

This data does not include YouTube and don't believe any statements that it doesn't count. Data from APAC shows higher viewing rates and data from the US shows higher daily viewing than all streaming services.

A new study published in the Association of Computing Machinery (ACM) Special Interest Group (SIG) for Computer-Human Interfaces (CHI) looks at information overload and its predictors. It turns out that fragmentation of online sessions is a predictor of overload while total hours online is not. 

Above, we have put the total hours for each group in the labels to show that the highly overloaded group has average or lower total time (but higher fragmentation).

Also in this report, even if somewhat unrelated, is usage types by hour of the day. The colors are extremely difficult to tell apart but we have tried to highlight the main tasks.

Unsurprisingly, games, messaging, and social media are the top mobile use cases. Health is in the lists but it is the second-lowest task by time (after Gambling).

EDO put out a report on NFL ads hyping the results from the league and playoffs. In the report we found that three of the top 10 brands were pharma.

Also, page 10 of the report is all about pharma. The creatives that won are listed for anyone looking to do some research.