Skip to Content

Analytical Chemistry

Theranos and the Limits of Salesmanship

I haven’t been covering every step of the Great Theranos Unraveling, partly because it’s been on every news site there is. But the company (in the person of Elizabeth Holmes) recently disclosed yet another new blood testing device, in a presentation at the American Association for Clinical Chemistry meeting. That might seem an audacious venue, given the company’s clinical chemistry difficulties. The talk itself was notably long on claims and short on details (and does not seem to have gone over well). But there’s even more to the story, according to the Financial Times:

Several scientists on the committee responsible for organising the meeting opposed the decision to invite Ms Holmes for fear that it would lend credibility to her company, while others decided to boycott the presentation in protest. They believe the association invited Ms Holmes to generate publicity.

Dr Andy Hoofnagle, a member of the organising committee, said he and several of his colleagues had “fought really hard to prevent” Ms Holmes from appearing but were overruled by the AACC president, Patricia Jones.

“I’m removing myself from the committee and don’t intend to pay my dues next year,” he added, in effect announcing his resignation from the association.

The article goes on to quote a number of very disgruntled AACC members, but it’s impossible to say what fraction of the membership was upset. Patricia Jones herself is quoted as estimating about ten per cent, so we can take that as a lower bound. The decision to invite Holmes’ presentation probably looks even worse in hindsight, but the people quoted in the article were opposed before the first slide went up on the screen. Still, she had the perfect venue to actually break out some real data and restore some scientific credibility, but would seem to have spent too much of her time in sales-pitch mode.

That’s definitely the wrong way to pitch a talk at a meeting like this. The people who go to meetings on this topic have been bombarded by advertising for new instruments and assays; they can recognize when someone’s trying to sell them something. At any scientific meeting, you can see people walking out when something starts to sound like a “vendor talk” instead of a scientific presentation. At its worst, it feels a bit like having a conversation with someone and having them start to ask you if you’ve considered buying Amway products. Holmes surely realized that she would be facing a very technically oriented audience, many of whose members were skeptical (or openly hostile), so taking this approach really shows poor judgment. The only way Theranos is going to convince anyone of anything is with data – lots of data, and as much of it generated by independent third parties as possible. Talking about the great new chapter in the company’s history – look at this new machine! – is not going to cut it.

41 comments on “Theranos and the Limits of Salesmanship”

  1. Anon says:

    Lost for words that Holmes could have such poor judgement not to learn from her past mistakes. Everyone makes mistakes, especially when starting out. But those who don’t learn fast should be dropped and avoided like the plague.

    1. bhip says:

      Sorry to disagree but Holmes (& her ilk) do not consider the high sizzle/low steak model to be a mistake-she just got caught. Similar lofty claims with little-to-no data has made many, many people filthy rich (Hello Sirtris Pharmaceuticals!!) & will continue to do so simply because outlandish, “transformational” ideas are more likely to gain traction with VC/investors than solid, boring old science…

      1. SedatedFMS says:

        Surely you don’t mean that vulture captial funds want the instant successes with minimal risk promised by charlatans like Holmes rather than the long term, higher risk associated with actual science? The FMCG approach to drug discovery has worked so well.

        1. Anon says:

          Well as long as there’s a more stupid buyer out there willing to pay even more for this kind of crap, who can blame them? It’s like a game of pass the parcel, the trick is not to be the one holding the bag of excrement at the end.

    2. Mark DeWitt says:

      The original sin of Theranos is that it’s a biotech that was run like a tech startup. In tech (the silicon and code variety), the -idea- is the thing. Having a visionary, charismatic founder in a black turtleneck is a great way to lead a company into a new direction. Turning the idea into a reality from an engineering standpoint has a very high probability of success. But in biotech, the -execution- is key. Ideas are cheap, because most pan out for reasons unforeseen. Data is the only currency worth anything. Biotech executives know this, VCs know this, scientists know this, but the general public (including the tech VCs that lost all their money on this boondoggle) does not.

      The sad fact is that Mrs. Holmes doesn’t understand this, in part because she has no actual experience in the industry; only a year of grad school and then CEO of this sham of company. I agree with the previous commenter that the investors (and the board) are mostly to blame for letting things get this far. Holmes literally doesn’t know any better.

      1. tangent says:

        “In tech (the silicon and code variety), the -idea- is the thing. […] Turning the idea into a reality from an engineering standpoint has a very high probability of success.”

        Where “very high” might be 5% chance — most startups with a great idea flop — but I agree with your point that execution is still easier for tech than bio. Bio appears to have even more unknown unknowns.

        The big killer in tech is the product idea, whether people actually want your groundbreaking thing. Most successes come after one or more ‘pivots’ i.e. realizing your product was wrong. Is product also a 95%-fail type of gate for bio, on top of bio’s tougher execution gate? It seems product is not as bad there, though Theranos for one does seem to have a real product problem, what their whole hypothetical thing is good for. A charismatic leader can be a negative if they like to persuade more than to be persuaded.

        1. Mark S. says:

          Quote: “In tech (the silicon and code variety), the -idea- is the thing. […] Turning the idea into a reality from an engineering standpoint has a very high probability of success.”

          Where “very high” might be 5% chance — most startups with a great idea flop — but I agree with your point that execution is still easier for tech than bio. Bio appears to have even more unknown unknowns.

          I’m not the person “tangent” replied to, but I think there’s a misunderstanding between what these two folks mean by successful execution. As I read it …

          “Mark deWitt” was referring to successfully building the product and putting v1.0 out in the marketplace. That’s something that almost always succeeds. Clearly in current-era drug discovery, successfully building the product is a very, very long shot.

          The _next_ hurdle, as “tangent” points out, is whether the successfully-engineered product is a success in the marketplace. And yes, here’s where IT-type tech encounters it’s Great Filter.

          IANA expert on bio drug / device / test sales results, but it seems the truly breakout products that offer obvious gains in previously intractable spaces tend to be very successful. The me-toos offering minor incremental gains for major increased cost? Not so much.

    3. xplo says:

      Holmes isn’t a scientist, she’s a marketing gimmick (“marvel at the biomedical genius who’s also a hot young blonde!”), and to the extent she contributes anything, she’s an entrepreneur. For a dedicated entrepreneur, crushing failure that destroys your reputation isn’t a warning, it’s just a sign saying to try it again, maybe twice as fast or backwards, and preferably on a new group of suckers, because one of these times it’s going to take off.

      Hell, considering how long her microfluidic dog and pony show persisted the first time – despite legions of scientists and people in related fields shouting FRAUD! – she probably thinks of Theranos as an imperfect success, and considering the amount of money I understand she made on the venture, she might be right…

  2. Magrinho says:

    She is true to her nature and does not have a broad skill set.

    It’s almost Darwinian – when the conditions require substance over sizzle, she is not well adapted.

  3. Earl Boebert says:

    In the course of my career I have played (with decidedly mixed success) a technical Jiminy Cricket to a variety of marketing Pinocchios. What I learned from this is that the really great salespeople are mythomaniacs. You could strap any of them to a theoretically perfect lie detector during their pitch and it would register “no deception.” That’s what makes them so dangerous in fields that involve the fate of human beings.

    1. Mark Thorson says:

      My parents used to say “The best salesmen are easily sold.”

      1. Crocodile Chuck says:

        “”Good propaganda fools the people who read it. Great propaganda fools the people who make it.” — Dan Neil, The Los Angeles Times, 2007

  4. RTW says:

    It was really poor judgment by Holmes to present the research. Why wasn’t her PI that was responsible for the research presenting it? It would have definitely given much more credibility to the work. Goes to show its all about Elizabeth and her cult of personality. Would Pfizer’s CEO standing up in front of say AACR and presenting the results of their latest phase II clinical trial of one of the oncology portfolio drugs? Give me a break! Ian Read might be better able to pull it off at least he actually holds a degree in Chemical Engineering from Imperial College! Holmes quit her CE degree and is a Stanford dropout!

  5. Eric says:

    Isn’t it the most parsimonious explanation? It’s not the wrong choice of emphasis or wrong presenter or wrong audience. It’s that there is no steak, only sizzle, and this talk was an attempt at what Theranos has left — “Look! New sizzle! Different size pan!”

  6. CFaulk says:

    During her talk:
    “We know there’s a lot of questions about the past,” Holmes said, “and in the appropriate forum, we’ll address those.”
    I wonder what that forum might be, or whether it exists.

    Maybe we’ll have to wait until she has a memoir to sell us.

    1. Isidore says:

      Hmm, perhaps by “the appropriate forum” she means a courtroom.

    2. Emjeff says:

      It’s the same forum where Hilary Clinton will address her email server scandal. It’s called “Never”.

      Handling this the way the press has does women in science no good at all. If Theranos had a male CEO and has a similar scandel happen, I guarantee that that CEO would never present anywhere again. Yet, because Elizabeth Holmes is an (attractive blonde) woman, she is being given a pass.

      1. M says:

        Actually, we’ve done the experiment with male CEOs and other wealthy types–let them fail spectacularly, overhype achievements and scam investors, even convict them of felonies. The data suggests they, too, can keep going. People admitted to the club of the wealthy live by different rules.

      2. tangent says:

        What examples do have in mind? What’s the least-bad scandal for a male CEO that got him consequences of the type you suggest?

  7. kjk says:

    Theranos is one example of “no, you can’t program biology Silicon-valley style”. But the current drug-discovery paradigm is also failing. It relies on grunt lab-work (both academia and industry), which is expensive and wastes brilliant minds. And we can’t get enough throughput to find much meaningful data, so end up publishing mostly statistical artifacts as shown in “a call for better mouse studies”. Biology is too complex to program, but most of the work isn’t. Automating the boring parts, the goal of Transcriptic, Emrald, etc, seems the best way to increase throughput and let biologists be biologists.

    1. Ryan George says:

      Exactly, the key is automation and increasing throughput. A platform that would allow biologists to design and execute experiments in the cloud is going to bring about a new way of doing a lot of biology. Making it microfluidic will be the next step to increase throughput and drop costs. Right now everyone has a problem where they think microfluidics is all for medical applications, but the real innovations will happen first in automating the mundane tasks all of the innovators are stuck doing. Medical microfluidics will spin off of that when they find some interesting phenomena happening while studying something and realize it has applications in medicine.

      1. John Wayne says:

        Okay, I have to call BS on the following two statements:

        1. “A platform that would allow biologists to design and execute experiments in the cloud is going to bring about a new way of doing a lot of biology.” Experiments within biology are hard enough to run in real life; running them on a networked computer (what people now call ‘The Cloud’ for some reason) is daydreaming. The computer would need to understand biology to be predictive. No. Not even close.

        2. “Making it microfluidic will be the next step to increase throughput and drop costs.” This is an interesting idea, but not without it’s challenges. Talk to anybody who has miniaturized an assay and you will learn that biology doesn’t scale up or down without complaint. The things you have to do to make that work often run counter to the in vivo relevance of your results.

        It turns out that dropping a bunch of currently popular ‘disruptive’ terms only distracts people from what must be done.

      2. Oliver Hauss says:

        Sorry, but you can’t engineer around basic statistics. Making it microfluidic means there are certain parameters you will not get a precise measurement on, period. Because to get precision, you need events. In your microfluidic drop, you have a limited number of events. If you have an actual concentration of two cells per microliter in the patient, you may get no, one, two, maybe three or four events. With very limited probability five. How do you know which is closest to the true patient value? You can’t. Because the coefficient of variation is so large that you will routinely get values that are below or above the actual value, and in this concentration range, that means a significant percentage off.

        But we’re talking here about measuring one single parameter. The situation is even worse when you take that drop, dilute it and then split it up to measure a host of different parameters. Some white blood cell populations will routinely show a reading of zero under these circumstances, simply because no cell of that type made it to the pertinent channel, even though they are present in the peripheral blood.

        And any low concentration clinical chemistry parameter will have similar problems with detectability. Microfluidics is suitable only for higher concentration parameters, as lower concentration ones will have too high a coefficient of variation to be reliable.

  8. SteveM says:

    As a side note, Continuing Medical Education (CME) and clinical conferences for years have been corrupted by crony-pharma gaming that pays M.D.s to directly or indirectly pimp for their products while pretending objectivity.

    Yet nobody ever walks out on the speakers and everybody politely applauds the propaganda at the end of a presentation.

    Too bad the AACC response against Theranos is not representative of the larger medical community.

    1. Emjeff says:

      Not true at all; CME and CPE standards have been considerably tightened over the years. To deny past abuses would be foolish, but to say that things are continuing as always would be equally so.

  9. biotech scientist says:

    I haven’t followed this very intensely, but it seems to me some of this has to be on the VC. I don’t work actively in VC-biotech dealings, but I have been around a few, and those all had NDA’s set-up, and disclosure was expected. Am I looking at this incorrectly?

  10. J. Peterson says:

    Word has it the conference organizers played the Rolling Stones’ song “Sympathy for the Devil” over the PA while waiting for the Theranos session to start.

  11. a. nonymaus says:

    “… At its worst, it feels a bit like having a conversation with someone and having them start to ask you if you’ve considered buying Amway products.”

    Worst first date ever…

    I’m surprised that Theranos is still trying to make a go of it. Isn’t the usual MO to strike the tent and then come out a bit later with a new layer of paint on the signs?

  12. ScientistSailor says:

    Maybe time to re-visit the thread on Cargo-Cult science?

  13. Isn’t this against the FDA ban that was placed against Holmes? See link in my handle.

    1. UudonRock says:

      She was banned from operating a lab, not from pushing junk on people gullible enough to buy it.

    2. Li Zhi says:

      WSJ reports her ban has 2 weeks before it goes into effect. (to give her & co. time to appeal).
      Given all the money being poured into medicine, and given the lack of rigour in the area – as measured by replication of published work, the existence of this kind of very visible corruption in the AACC organization should come as no surprise.
      I no more fault Lizzy for being true to form, than I’d fault a snake for biting the hand that feeds it. The people who should be criticized are the committee members (and their employeers) who chose to invite her without, apparently, evaluating the quality of the science she would be presenting.
      From what I have heard, she was announcing new packaging of old ideas/technology. Trash heap of history, indeed.

  14. Oliver says:

    This was clearly the old bait-and-switch. She should have been thrown out on a rail, while being tarred and feathered!

  15. Daen de Leon says:

    I work in the diagnostics field, and if anyone wants to waste half an hour looking up Theranos’s patents (or scant publications), it will send up enough red flags to have made Stalin happy for ten May Days in a row. I’ll save you the time: don’t bother.

    The thing with Holmes and Theranos is that sales mode is their default setting. They don’t do substance; they do *disruption*. Also, cast an eye over the names of rheum-eyed Cold War-era dinosaurs on their board, and you’ll get an inkling how such a know-nothing company could have ever wielded such political influence.

  16. David says:

    To be fair (said between gritted teeth) Kissinger and Schultz seem to have taken their money and gone.

    The CMS communications to Theranos are terrifying – this was a lab that appears not to understand concepts such as precision, accuracy, quality control, employment criteria, truth-telling . . .

    But it was all ‘disruptive’ so everything is OK

  17. Kim says:

    It’s because THERE IS NO DATA to submit. The limited amount of data they have shows inconclusive results and little to no accuracy or replication. They can’t show that to anyone! Just keep trying to silence your employees and former employees, Theranos.

  18. outsider says:

    Holmes has zero scientific training. She left college after one year as an undergraduate. She didn’t receive any graduate training. Her scientific training is therefore at an undergraduate, freshman level.

    Imagine an undergraduate presenting a 90-minute detailed presentation of data and serious science at the American Association for Clinical Chemistry meeting. LOL. Even the best undergrads I know are not capable of that, because they don’t have the relevant training and experience.

    Holmes clearly doesn’t understand science and how to evaluate scientific data, including and especially her own. As Feynman pointed out, in science, it’s easiest to fool yourself. One important aspect of scientific training is surrounding yourself with capable, skeptical colleagues, who teach you how to be skeptical of your own data. She never had that, it seems.

    From what I can tell from the outside, this is a person who was able to raise a ton of money from investors, but she didn’t know how to do science. Too bad for the VCs. They should stick with trained scientists.

  19. s says:

    NPR reported three UG of Stanford will tackle the problem of superbug. I wonder whether we will see more Ms. Holmes and Mr. Holmes from Stanford.

  20. Anonymous says:

    Genetic Systems, 1980-1986. Founded by David Blech, a 24-yo songwriter / stockbroker and Robert Nowinski (Hutchinson Cancer Res Center) must have been an early model for Theranos. GS was developing their miracle machine for bioanalytics, ChemWare. It didn’t work, either. Employees had to toe the line and not rock the boat or else. In 1986, Bristol-Myers-Squibb bought GS for $300 M. A few years later, they sold off the worthless assets for $20 M.

    Blech started other biotech companies, was eventually convicted of fraud and sentenced to 5 years (served as probation, due to a medical condition).

    The only question is: Did David Blech wear a black turtleneck sweater?

    Here are some links:

    1. tangent says:

      I was too young to have heard the Blech story. Sounds familiar… except all of the dollar figures are so quaintly small!

Comments are closed.