When Babylon Health listed in New York in 2021, the telehealth company was valued at $4.2bn. Within a year later the stock price collapsed and it later filed for bankruptcy. The company’s British assets sold for only £500k . Babylon’s investors, like the Saudi Public Investment Fund who led a series C of $550m in 2019, and Palantir (and others) who invested $230m in the stock market listing in 2021, lost all their money.
Our firm does a lot of commercial due diligence in health tech. While we did not work on or for Babylon, we thought that it would be instructive to dive through what diligence investors could have done before committing so much capital in successive funding rounds.
There are many lessons to be drawn from the demise of Babylon Health. We have highlighted three key ones:
1. Understand the technology
2. Understand the unit economics
3. Beware of “blitzscaling”
The above three points are the result of an outside-in analysis and are not an exhaustive list. Nonetheless, our analysis should allow investors to learn from one of the biggest digital health failures of all time.
Understanding the unit economics
Babylon Health was founded in 2013 and grew to a $4.2bn valuation following multiple funding rounds and finally a listing on the New York Stock Exchange (Exhibit 1).
The company went through a number of different phases and adopted several business models. In its early years, it attempted to sell its services directly to the end user via a pay-as-you-go model. But the economics did not stack up.
For a retail business to work, the customer acquisition costs (CAC) needs to be less than the lifetime customer value (LCV – the sum all the profits from each transaction over the course of a customer’s membership).
Though we don’t have the exact figures for Babylon, c. £150 to acquire a retail customer is par for the course. All those Google and Facebook advertisements to get someone to download an app do cost quite a lot, not including advertisements on the London tube. Hence, the company would need to make more than £150 over the customer’s lifetime with them.
In Babylon’s initial pay-as-you go model; however, the company was charging £25 per appointment[i]. If we assume they paid GPs about £20[1] that gives us a £5 profit per visit. Further assuming that patients made an average of 4 appointments per year this gives us a yearly profit of £20. Hence, it would take 7.5 years to repay the CAC. This is assuming no patient churn and excluding any overhead costs.
Possibly realising the difficulty in making this retail model work, Babylon subsequently pivoted to distributing their product through private medical insurers (PMIs) and through the NHS.
In the UK, Babylon’s main insurance contract was with BUPA. Though we do not have the details of this deal, we understand that this represented a fee-for-service contract. We further assume that this business was (or had the potential to be) profitable since eMed were willing to spend £500k to acquire it.
Contrary to the deal Babylon likely had with Bupa, in the NHS, primary care is reimbursed on a capitation fee basis with GP practices receiving an average fee per patient of £155 annually (2022).[ii] Babylon claimed it received £93 per patient (in 2019) due to the younger population signed up at its practice.[2],[iii] Furthermore, Babylon initially offered 24/7 access, though it did not get extra funding for this.
A senior figure in the Babylon leadership team who we spoke to says GPs at Babylon were expected to see five patients per hour, and would get paid around £100 per hour, hence £20 per patient. So, if the patient requested five or more consultations per year, the business would lose money under this model also. Further, in a model where GPs were paid an hourly rate there is little incentivisation to work more efficiently.
Increased access to care also typically results in increased use of care whether that care is truly necessary or not. Babylon claimed GP at Hand patients averaged 6-7 appointments per year, twice as many as patients with standard GPs the company claimed.[3],[iv] “In the NHS, it is much harder to get to see a GP. If you make this barrier lower, then you are more likely to have people come in for very minor things… [occasionally] we had patients who would call the service 3-5 times a day” says a senior figure at Babylon.
Though Babylon did provide downstream savings for the system through lower acute hospital costs, these did not end up benefiting the company’s bottom line.[v]
None of the above issues are specific to Babylon and any GP service will face similar challenges. However, telehealth providers have two additional costs that regular NHS GPs do not have in the form of marketing to attract new patients, and R&D to develop their tech platform. Furthermore, telehealth services are largely commoditised, making differentiation and ability to command higher prices difficult.
Hence, any providers that are dependent on healthcare professionals and face a capitation reimbursement model will need to focus on their margins by either 1) building workflow or cost efficiencies that can counter both a potential increase in patient adoption and an increase in costs, or 2) adopt a loss leader strategy and leverage its services to sell other products or services.
In contrast to the above, Swedish telehealth provider Kry (known as Livi in the UK and France) adopted a scalable model when entering the UK and did not require patients to sign up to them as their main provider. Hence, they did not bear the same risk of customer ownership as Babylon.[vi]
Though we have not been able to identify the financials of Livi’s first contracts, the evaluation of a Livi pilot done in 2021/2022 by Manchester Health and Care Commissioning (MHCC) cited fees of “£27 per appointment, with a service efficiency of 4 appointments per hour …an hourly rate of £108 and a sessional rate of £432 for the interpretation service (16 appointments)”.[vii]
For six consultations, which Babylon claimed to have per patient, the above would result in £162 in revenue for Livi, which is significantly above the £93 which Babylon claimed to receive per patient in 2019.
Even with Livi’s more scalable business model, Livi’s CEO Johannes Schildt states it is only now reaching profitability in all its markets and ongoing growth would not have been possible without implementing technology.xi
It is apparent that both Babylon and Livi recognised the importance of leveraging technology to create efficiencies that would allow you to scale primary care, but ultimately, fundamental differences in their unit economics contributed to almost unavoidably different outcomes.
Beware of “blitzscaling”
Wired reported that “[founder and CEO] Ali Parsa was obsessed with “blitzscaling”—the kind of entrepreneurial hypergrowth popularized by LinkedIn cofounder Reid Hoffman.” Further claiming that “The company went on uncontrolled hiring sprees, ex-employees say, and teams were often working on overlapping projects.”[i]
We analysed comments on the employer review website Glassdoor from the time prior to Babylon’s $550m series C, which would have been available to anyone doing due diligence (Exhibit 2). The comments were telling and summarised in the table below. 25% of employees mentioned a lack of overall strategy. One employee said that Babylon was “over-hiring with no clear direction or objective”. Another employee described Babylon as “bursting at the seams yet still hiring like crazy”. Whilst Glassdoor reviews are often left by disgruntled employees, this should have been a red flag, and investors should have followed this up by conducting interviews with former employees and management for further insight.
EXHIBIT 2 – EMPLOYEE COMMENTS PRIOR TO BABYLON’S SERIES C (AUG 2019)
Below is a selection of comments. The n number are the actual number of people referencing these issues (out of 108 total reviews).
| Lack of clear strategy (n=27/108) | Blitzscaling (n=34/108) | Limitations of Babylon’s Technology (n=16/108) | High churn rate (n=8/108) |
| “The goals change every minute and you are just expected to run with the changes, no time to even draw strategies” | “It’s a company that has tripled in size in the last 9 months alone” | “Some people working on AI don’t really do AI but APIs instead … rushed implementation sometimes” | “The best engineers I’ve ever had the pleasure of meeting have been replaced by the most expensive contractors” |
| “Senior management have no coherent plan, and change their mind from week to week about priorities” | “Very fast growth (100 à 1300 employees in two years)” | “AI Engineering Team is not about AI at all, it’s casual backend engineering. The only “AI” part of it is calling the APIs of ML services that are maintained by data scientists” | “Most of our best Python developers left in the last year of so for greener pastures” |
| “The constant confusion whether Babylon is a startup or a growing corporate company, due to the clashes of hires from startups and Apple, Uber, 111. Is it a tech company, is it not? CEO and the managers are not aligned in the ideologies and it shows” | “HR currently has 50+ people a month to bring into certain departments. There is nothing in place to ensure these people are up to standard, that there is capacity for them to actually have a desk or training in place” | “Edge cases are never expected to happen and usually every project is only thought on the happy path situation… the communications between front end and backend teams is very poor… There is no data driven decision being made at all” | “I have never seen so many people getting fired so quickly. From the most junior to the most senior people, I have seen so many losing their jobs. Literally, every day at least one person was losing their job because someone decided that they are not fit for the role they were hired for” |
| “Hiring a lot of people with no clear direction or objective. They think mass hiring means growth” | “Product organisation needs time to mature to the state where it can create a collaborative and transparent process for what products we develop and why” | “[You are] expected to just “build fast” i.e. do things quickly even if we do not think they were scientifically rigorous or up to the standards we would have applied in our previous lines of work” | “Retention of employees: The onboarding is a mess that truly has not improved for the past two years. They also come to see people leaving in bulk, which again forms their opinions about the company” |
| “Lack of C-suite strategy” | “Bursting at the seams and still hiring like crazy” | “Engineering practices are poor” | “Turnover is really high” |
| “Complete lack of direction in so many different fields” | “Due to the speed at which the company is growing, we seem to be running out of space… cases where people had limited desk space or had to work from ping pong tables” | “Clinical safety is difficult to guarantee due to the way the technical stack is developed … Poor architecture over the entire platform, lack of quality and design documents do not happen” | “Highest turnover rate I’ve ever seen” |
| “Management is entirely reactive, and there is very little if any commercial strategy” | “Cramped office space, hiring too fast (i.e. +10 people per week) and too low quality hires” | “Denied the opportunity to write reliable tests or reliable architecture by lack of time, discouraged by active disheartening” | “Product managers come and go, causing lack of direction” |
| “Management changes daily as does the direction and mission of the business” | “Extremely fast growth and start up environment there is an issue with a lack of meeting rooms and space” | “With so many engineers the product is still below average – look & feel and functionality” | |
| “Chaotic at times and it sometimes feels the company is trying to do to many things at once” | “Babylon is going very fast, on the verge of frantic… the speed of a start-up has largely been maintained, but now with over 150 engineers” | “Not too much effort seems to be placed on ensuring these solutions remain well-integrated going into the future” | “Replac[ing] constant stream of leavers” |
| “No vision … mission is nowhere to be seen. Instead there is a pressure to be fast and first” | “Technical debt is HUGE” | “Systems used are outrageously outdated” | |
| “Hiring to “blitzscale” | “Legacy code makes our life difficult and a lot of times we have to fix legacy bugs” | ||
| “Complex legacy systems that could do with improving or rewriting” |
This “daily change in direction and mission”, as one employee put it, likely impacted product development leading to both too many products in its portfolio and those products not being sufficiently iterated upon after release.
Anecdotally these pressures on the product development also contributed to a high churn rate. On Glassdoor, employees mentioned “most of our best Python developers left in the last year” with the “best engineers … replaced by the most expensive contractors”. Another employee wrote that: “Product managers come and go, causing lack of direction”.
Beyond personnel, Babylon also had a broad product portfolio which included a symptom checker, chatbot tools, an app platform, appointment booking, workforce management, video consultations, prescriptions, and referrals to other care pathways. The breadth of products and rapid growth might have been encouraged by investors and partners, but ultimately meant the company could not spend the time needed on its core products.[i] “Instead of focusing and iterating, we instead just went on to develop newer products… We were pursuing too many areas instead of iterating on the core service” says a senior figure at Babylon. It is pertinent to note that the part of Babylon that does remain is its core ‘GP at hand’ service.
Babylon also expanded rapidly to many markets and at its height it was present in 17 countries by 2021 (Exhibit 3).[ii] There are numerous reports however of Babylon not properly understanding its new markets[iii],[iv], or underestimating the challenges in adapting their business to enter them.
Babylon’s product needed to be localised for different languages and epidemiological models. Hence, a headache and fever, which in the UK would most likely be a cold, in Southeast Asia or Latin America could be dengue fever, a much more serious condition.[i] This obviously required significant efforts on behalf of the company.
Babylon tended to build its technology in-house rather than make acquisitions. However, this combined with the rapid scaling of the company is likely to have created issues. Hence, engineers on Glassdoor also claim to have been hampered by “fixing legacy code” on “outdated systems”, being pressured to “build-fast” and “denied the opportunity to write reliable tests or reliable architecture”. One person commented that “clinical safety [was] difficult to guarantee due to the way the technical stack [was] developed” (Exhibit 2).
The strategy of building mainly in-house is in contrast to for instance, Livi who in 2019 acquired the Stockholm-based company Hello Shim, which developed the psychology apps Enjo and Shim, and integrated it into its app to offer mental wellness support.[ii] Livi later acquired the practice communication service MJog[iii] and VIX Digital, which specialised in designing digital solutions for the NHS.[iv]
Livi has also been more conservative in its international expansion being present in only Sweden, Norway, France and the UK (the company pulled out of Germany in 2022 and previously also operated in Spain).[v] As seen above Livi also initially partnered with the NHS instead of competing with it and when the company decided to open two physical clinics in France it did this in partnership with hospital care provider ELSAN.[vi]
Though Livi has had to go back on some of its growth investments, notably laying off around 400 employees in 2022, the company has been more cautious in scaling.[vii] Johannes Schildt, CEO and founder of Livi, stated in 2020 that “It’s a market-by-market exercise” and “There is a fair amount of customization… Different markets have different regulations, slightly different needs.”[viii]
Digital health is just not as scalable in the same way as other technologies. Blitzscaling might have been successful for Uber and Amazon, but healthcare is deservedly much more tightly regulated than ridehailing. Miss an Uber, you might be late; miss a diagnosis, and the consequences could be much worse.
Understand the technology
After Babylon claimed its chatbot beats GPs at a medical exam in 2018, three academics wrote a paper in The Lancet that “Babylon’s study did not offer convincing evidence that its Diagnostic and Triage System can perform better than doctors in any realistic situation, and there is a possibility that it might perform significantly worse.” [i]
This was one year before Babylon received a $550 million investment, valuing the company at over $2bn.[ii]
The authors in The Lancet were not alone. an NHS consultant oncologist, Dr David Watkins, also repeatedly flagged the flaws of the chatbot in triaging patients on Twitter (now X).[iii] Concerns were also raised about the company’s AI capabilities on Glassdoor, with one software developer stating that the “AI engineering team was not at all about AI … the only AI part of it [was] calling the APIs of ML services that are maintained by data scientists” (Exhibit 2). Further, in March 2019, the London-wide Local Medical Committees[1] expressed concerns over the fact that Babylon’s technology ‘has yet to be properly evaluated’.[iv]
Again, Babylon is unlikely to be unique in this case and all probabilistic triaging systems will face at least two significant challenges. Firstly, as a senior figure at Babylon summarises, all probabilistic models will be limited by “whether you have enough variables (such as risk factors and symptoms) built into the model … and how good the published evidence base is [that give a probability between the variables and a diagnosis]”. Secondly, there is a trade-off between sensitivity and specificity of your model i.e. deciding when and when not to signpost patients to seek out physical care.
The more restrictive a model is in sending people to seek care, the more likely it also is to miss something critical. Conversely, if a triaging system sends through patients less discriminately to avoid missing any critical cases, it starts losing its purpose.
From the studies that have come out, a senior figure at Babylon explains that AI triaging tools “do work very well but are more likely to over-refer to emergency services”. Understanding these inherent limitations and trade-offs of any triaging tool is a key starting point for investors when evaluating these kinds of digital health assets.
Remarkably, there was no regulatory scrutiny of Babylon’s chatbot, since neither the UK’s Care Quality Commission (CQC) nor the Medicines and Healthcare Regulatory Authority (MHRA) saw the chatbot as under its remit.[i]
It is possible that investors relied too much on the NHS and the company’s existing success as signs of quality assurance, rather than doing their own diligence.
“Big investors trusted Babylon and their AI because of those NHS contracts. They assumed the NHS had undertaken due diligence.” said Dr Watkins to The Times. In the same article, Per Brilioth, Managing Director and board member of Vostok New Ventures (VNV) echoed the importance of the NHS as a first client. VNV lost over $100 million when Babylon collapsed.[ii]
Conclusion
Though it is always easy to draw conclusion in hindsight, there were a lot of red flags for Babylon’s investors to find, had they looked for them. The sad story of Babylon Health shows how important it is for investors to properly diligence digital health companies.
[1] Financial Times: Babylon to list in New York at $4.2bn valuation, backed by Palantir. Available at: https://www.ft.com/content/d41a17ca-8b06-4cb4-91f0-a410e52bf7a8 (Accessed 4 March 2024)
[1] The Times (28 October 2023): The app that promised an NHS ‘revolution’ then went down in flames. Available at: https://www.thetimes.co.uk/article/rise-and-fall-of-babylon-healthcare-the-doctor-in-your-pocket-3p6q6jjfx (Accessed 6 March 2024)
[1] The Financial Times (24 April 2017). Available at: https://www.ft.com/content/1f56997a-290f-11e7-bc4b-5528796fe35c (Accessed 28 March 2024)
[1] Pulse (30 May 2022): Babylon ‘cautious’ to expand UK GP services amid cash losses. Available at: https://www.pulsetoday.co.uk/news/technology/babylon-cautious-to-expand-uk-gp-services-amid-cash-losses/ (Accessed 7 March 2024)
[1] DigitalHealth (11 February 2019): New GP contract ‘penalises’ income for digital providers, Babylon claims. Available at: https://www.digitalhealth.net/2019/02/gp-contract-penalises-income-digital-providers-babylon/ (Accessed 6 March 2024)
[1] Healthcare Finance (April 2019). Available at: https://www.hfma.org.uk/system/files?file=hcf—april-2019.pdf (Accessed 7 March 2024)
[1] Winward S, Patel T, Al-Saffar M, Noble M. The Effect of 24/7, Digital-First, NHS Primary Care on Acute Hospital Spending: Retrospective Observational Analysis. J Med Internet Res. 2021 Jul 22;23(7):e24917. doi: 10.2196/24917. PMID: 34292160; PMCID: PMC8367118.
[1] Iacobucci G. Swedish company launches NHS partnership in bid to enter GP app market BMJ 2018; 363 :k4596 doi:10.1136/bmj.k4596
[1] Manchester Locality – Primary Care Commissioning Committee (25 May 2023). Available at: https://gmintegratedcare.org.uk/wp-content/uploads/2023/02/combined-public-manchester-primary-care-commissioning-committe-2552023.pdf (Accessed 7 March 2024)
[1] Wired (19 September 2023): The Fall of Babylon is a Warning for All AI Unicorns. Available at: https://www.wired.co.uk/article/babylon-health-warning-ai-unicorns (Accessed 6 March 2024)
[1] Sifted (18 October 2023): The rise — and fall — of Babylon. Available at: https://sifted.eu/articles/the-rise-and-fall-of-babylon (Accessed 6 March 2024)
[1] Sifted (27 October 2023): As digital health giant Babylon goes public, we look at what’s next for European telehealths. Available at: https://sifted.eu/articles/european-telehealth (Accessed 5 March 2024)
[1] Wired (19 September 2023): The Fall of Babylon is a Warning for All AI Unicorns. Available at: https://www.wired.co.uk/article/babylon-health-warning-ai-unicorns (Accessed 6 March 2024)
[1] Sifted (18 October 2023): The rise — and fall — of Babylon. Available at: https://sifted.eu/articles/the-rise-and-fall-of-babylon (Accessed 6 March 2024)
[1] Sifted (18 October 2023): The rise — and fall — of Babylon. Available at: https://sifted.eu/articles/the-rise-and-fall-of-babylon (Accessed 6 March 2024)
[1] Nordic9 (2 April 2019): Kry acquired Hello Shim, a Swedish chatbot provider for improving mental well-being. Available at: https://nordic9.com/news/kry-acquired-hello-shim-a-swedish-chatbot-provider-for-improving-mental-well-being-news0874638850/ (Accessed 13 March 2024)
[1] DigitalHealth (23 October 2020): Excl: Livi acquires practice communication service MJog. Available at: https://www.digitalhealth.net/2020/10/excl-livi-acquires-practice-communication-service-mjog/ (Accessed 15 March 2024)
[1] HealthTechWorld (13 September 2021): Livi acquires VIX Digital. Available at: https://www.htworld.co.uk/news/livi-acquires-vix-digital/ (Accessed 13 March 2024)
[1] Sifted (31 October 2022): Kry is laying off another 10% of its workforce and pulls out of Germany. Available at: https://sifted.eu/articles/kry-layoffs-digital-health-news (Accessed 13 March 2024)
[1] Sifted (22 February 2023): After 8 years of losses, digital health scaleup Kry is heading for profitability in 2023. Available at: https://sifted.eu/articles/kry-path-to-profitable-2023 (Accessed 13 March 2024)
[1] Sifted (22 February 2023): After 8 years of losses, digital health scaleup Kry is heading for profitability in 2023. Available at: https://sifted.eu/articles/kry-path-to-profitable-2023 (Accessed 13 March 2024)
[1] Forbes (10 January 2020): KRY Expands Further With $155 Million Investment Into Digital Health Platform. Available at: https://www.forbes.com/sites/jamessomauroo/2020/01/10/kry-expands-further-with-155-million-investment-into-digital-health-platform/ (Accessed 13 March 2024)
[1] The Lancet – Correspondence (6 November 2018): Safety of patient-facing digital symptom checkers. Available at: https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(18)32819-8/fulltext (Accessed 4 March 2024)
[1] Babylon Press release (2 August 2019): BABYLON ANNOUNCES A RECORD FUNDRAISE TO EXPAND TO THE US AND ASIA. Available at: https://assets.babylonhealth.com/pdfs/190802-Babylon-Funding-Announcement.pdf (Accessed 5 March 2024)
[1] Dr Murphy (aka David Watkins). Available at: https://twitter.com/DrMurphy11 (Accessed 13 March 2024)
[1] GPOnline (14 March 2019):GPs raise alarm over CQC failure to report on fast-growing GP at Hand. Available at: https://www.gponline.com/gps-raise-alarm-cqc-failure-report-fast-growing-gp-hand/article/1578691 (Accessed 15 March 2024)
[1] The Times (28 October 2023): The app that promised an NHS ‘revolution’ then went down in flames. Available at: https://www.thetimes.co.uk/article/rise-and-fall-of-babylon-healthcare-the-doctor-in-your-pocket-3p6q6jjfx (Accessed 6 March 2024)
[1] The Times (28 October 2023): The app that promised an NHS ‘revolution’ then went down in flames. Available at: https://www.thetimes.co.uk/article/rise-and-fall-of-babylon-healthcare-the-doctor-in-your-pocket-3p6q6jjfx (Accessed 6 March 2024)