Whilst there was some surprise at the timing of Atriax’s closure (it was widely acknowledged that it was in merger talks with rival FXall) few were surprised that it became the first name removed from the list of multibank portals (albeit swiftly followed by SunGard’s STN Treasury). What prompted a company, which at face value appeared to have a lot going for it, to close its doors? Despite being backed by three of the FX market’s largest liquidity providers, efforts in STP beginning to bear fruit and volumes climbing, collapse still could not be averted.
Colin Lambert asks the parties concerned, ‘what went wrong?’ and looks at the lessons to be learned from this high profile casualty.
Born amid confident expectations of market dominance, what caused Atriax, with its very strong shareholder base of Citibank, Deutsche Bank, JP Morgan and Reuters, to close its doors for business less than a year after going live? Already followed through the exit doors by SunGard’s STN Treasury, the story of Atriax is a fascinating one, that not only will provide case studies for business schools, but also raises serious questions about the future development and roles of the remaining multibank portals.
David Allanson became COO of Atriax in November 2001. He believes that the writing was already on the wall by the time he joined, and that the decision to seek a merger with FXall was taken as early as December. “By that time, shareholders had become increasingly frustrated at what they saw as an excessive cost base built by the company and the lack of any significant progress in the development schedule,” he says.
Sources within Atriax also speak of friction between the shareholders themselves, and the feeling within Atriax that there was a lack of support from the shareholders for their venture. This is corroborated to an extent by sources within the shareholder banks, one of whom reveals serious doubts about strategy. “I don’t know why we were involved in Atriax,” the source says, “We already had a multibank model in EBS, we should have either thrown in with FXall at the start, or developed the EBS model further.” (Such a suggestion was first floated in Profit & Loss magazine’s May 2001 issue.)
These frictions led to what the sources within Atriax describe as a fateful decision that impacted upon everything that followed in the company’s brief life: the choice of technology. One source claims, “There were three choices: AVT, Reuters Tibco and CFOWeb.com [a system developed by Integral]. Choosing the AVT system [which was the route taken by FXall] was strenuously opposed by the Chase [now JPMorgan] executives on the Atriax board. Similarly, the three shareholder banks were loathe to purchase the Reuters system. This led to the management being pushed in the direction of Integral, despite CFOWeb.com having only ever managed two trades during its existence.”
Profit & Loss could not obtain corroboration of the statement surrounding the resistance to AVT, as Chase has since merged with JP Morgan. As for CFOWeb.com, Harpal Sandhu, president and CEO of Integral, which produced CFOWeb.com and sold it to Atriax, disagrees with such claims.
“I cannot comment on the specifics of the particular deals that were done, but it is definitely not true that there had been only two deals on CFOWeb.com,” he says. “Atriax being shut down had absolutely nothing to do with the technology, nothing to do with the personnel, those are red herrings.”
Issues surrounding technology highlighted other problems within Atriax, allege company sources, specifically within the technology and development group. Allanson is critical of the group’s role. “Being pushed in the wrong direction is one thing. Failing to spot the downsides of a system and bringing this to the attention of the shareholders is another. Even though the technology management group had been seconded from senior positions within the shareholder organisations, they failed to recognise that CFOWeb.com had limited functionality. It could handle requests for spot pricing, but not simple forward pricing requests.
“More ominously,” Allanson continues, “Management underestimated the work required to develop the sell side API to link the system to the auto-quoting machines of the shareholder and other banks.”
Originally scheduled for release in June 2001, auto-quoting capabilities were not available until the beginning of November. This, says one source, “eroded the confidence of not only the buy side clients themselves, but also the staff within the shareholder institutions who had geared themselves up to sell the system”.
Again, Sandhu disagrees. “Technology was the least of the issue,” he says. “If anything, Atriax technology was open, everbody could link to it, and everybody could subscribe to it. All three shareholder banks are investing heavily in their proprietary sites, two of them using Integral technology to do it. I think the question should be: did the business model make sense for the sell side? The simple answer is no. Atriax, Currenex and FXall are all using different technologies and none of them have yet been dramatically successful, so technology really has little to do with it.”
Of more concern than these initial problems were the difficulties Atriax had in solving them in its early months. The sources suggest that time differences between California (where Integral was based) and London (where Atriax was based) caused delays in remedying the said auto-quoting issues as well as other developmental work. As Atriax fell further behind its competitors in developing functionality, these sources say, the need to develop products such as the STP, funds breakout, and content initiatives proceeded simultaneously in-house, and parallel with the Integral work. This produced a larger workforce than was expected (or budgeted), inflating what was already a large cost base. This problem, suggests one source, was inflated even further by the negotiation of an “onerous” hosting contract for the system.
All told, sources within the company tell Profit & Loss that Atriax spent in the region of $100 million by the time it closed its doors (see sidebar on page 17). This is far in excess of that rumoured to have been spent by its rivals, and had a lot to do with an estimated monthly cost base of around $3.25 million, again believed to be far in excess of Atriax’s rivals.
When Allanson arrived, he found other fundamental issues in need of remedy. These ranged from what he terms a lack of basic knowledge within the technology group as to how a dealing room operates and how markets work, to a dysfunctional organisational set-up. The CTO had agreed a deal to transfer to the US in June 2002, accordingly most of the development group were based in New York. This meant, however, that, taking the funds breakout vehicle as an example, a product could be sponsored by a salesperson in New York, the technical specification written in London, the development work handled in New York with the product testing taking place in London. “As costs escalated, and travel was pared to the bone, difficulties in communication created further delays,” says Allanson.
Not surprisingly, Allanson found the morale of the sales force within Atriax very low, and given the lack of functionality at that time, he remains amazed at how many customers the company managed to attract. “This is a tribute to the quality of the sales force, the STP vision and the reputation of the shareholder banks,” he suggests.
Even when Atriax introduced auto-quoting, it was not universal. Allanson believes this highlights fundamental differences between FXall and Atriax – specifically, that FXall only accepted banks willing to auto-quote on their portal and insisted they adopt the AVT system. “This created a consistency of service, he says, “especially in the speed with which prices were relayed to the customers”.
There were also philosophical differences between the two portals. Allanson believes that prices from auto-quoting banks on FXall appeared almost immediately upon request. Even though customers of FXall testify to receiving choice or even on occasions, inverted quotes, anecdotal evidence suggests that quoting banks very rarely change their prices. “This confirms the excellent support given to FXall by their shareholder banks,” says Allanson.
Mark Warms, chief marketing officer at FXall, confirms that FXall has a 95% acceptance rate on average, which is even higher for the top 10 banks. “Prices do go back to the bank on FXall, because we believe that is the most reliable way to approach this issue,” he says.
“In contrast to this,” he continues, “Atriax only had three auto-quoting banks in December 2001, and the speed with which they provided quotes ranged from two to eight seconds. At the same time, buy side customers might include a number of banks in their request for quote who quoted manually. This further slowed the process.”
Support for this comes from a corporate trader whose company used the Atriax system for a period of time. “We found quote times too varied,” he says. “We asked three banks for a two-way price, the auto-quoter was almost instant, the second bank quoted in 6-10 seconds, but a manual quoter could be anything up to 20 seconds – by which time, of course, the first and second banks had removed their prices. This made it almost impossible for us to fulfil our fiduciary requirements.”
In spite of the aforementioned disadvantages, Atriax remained successful in building customer numbers and volumes. A market source with knowledge of both companies suggests that Atriax’s volumes were similar to FXall and Currenex at the time it closed, but that they all lagged behind Global Link’s FX Connect, although as the portals do not release volume figures, this is impossible to confirm.
Allanson is convinced Atriax’s success was because it better understood the broad environment in which the portals operated, as well as the STP requirements of the buy side. A collaboration with GE and their treasury system provider FXpress had created the first fully automated environment. “The GE model was extraordinarily good,” he says, “In effect it was a one-click environment from trade inception to settlement – a fully automated lifecycle.
“I believe that one of the main reasons we continued to attract customers was that we were simply better at STP,” Allanson continues. “We had already proven and provided excellent products in this area. I think this proves that customers will go where the functionality and efficiencies are, and with all of the major FX banks on all of the main systems, the choice of portal is now coming down to who gives the buy side the functionality they need.”
Atriax built upon the success of its link-up with GE and FXpress by releasing AtriaxAdapt, a product that enabled it to connect to other treasury management systems (TMS) using a generic API. Additionally, the company released AtriaxInsight, which delivered Reuters data over the Internet and, says Allanson, could have been used to deliver bank prices from their auto-quoters, functionality long requested by the buy side.
Atriax also saw volumes climb during this period. Integral’s Sandhu suggests that in its last month of operation, volume over the portal rose by 40% on a month-on-month basis. He adds that even in the last two weeks of operation, “three very large European buys side institutions” were still deciding between FXall and Atriax. The jump in volumes can probably be attributed to the launch of forward FX trading capabilities on Atriax, which came at that time. Allanson agrees that volumes “jumped quite sharply” once clients commenced trading short date forwards on the system.
“In spite of these successes, Atriax still only had five banks autoquoting when it closed,” Allanson continues, This was nowhere near good enough. Hooking up individual banks’ autoquoters to our portal proved to be a complex and time consuming process not only for Atriax, but also for the banks. Unlike FXall, which made connectivity to AVT a pre-condition of quoting on its system, many of the banks on Atriax preferred to continue to quote manually rather than invest their systems professionals’ time connecting to a platform which many expected to disappear in a short time.”
The relative uniformity of ticket volumes, notwithstanding this problem, highlights the crux of the matter in Allanson’s opinion. “Different customers require different things from a dealing system,” he says. “The only consensus amongst customers is that they want all the banks operating on the system they use.”
Taking the fund management community as an example, Allanson notes that there is a certain consensus in this community about what it requires from a system, as in aggregation of FX requirements prior to dealing and splitting into constituent accounts and value dates post-trade. Other buy side customers have different requirements however, although the majority require post-trade download into their TMS. Some, including hedge funds, want the ability to input bids and offers, something they can currently do on Hotspot FXi, a system Allanson admires. Others want the capability to trade inter-company, a market currently served by 360T and system vendor Exidio. Still others want money markets, prime brokerage, orders and other asset classes. Allanson believes it unrealistic to expect all the portals to develop these products independently as the cost in a fragmented market would be too high and the rewards extremely uncertain.
Atriax’s problems were evident from the beginning to more than just its shareholder banks. Its clients also recognised there were issues likely to undermine the company’s efforts – specifically, the delay in releasing products. One client suggests, “From the start, while Atriax was failing to deliver on technological enhancements, FXall was knocking on doors and demonstrating superior technology and crucially, it was live on their system. This may not have swayed Atriax’s existing customers, but those looking to select a portal would undoubtedly have been influenced.”
Within the shareholder banks, pressure was already starting to be felt from the buy side. It was inevitable in the current client-focused environment, that the banks would have to listen more and more to their clients, who were giving them an unpalatable message. Sources within two of the shareholder banks have told Profit & Loss that the pressure became intense as 2002 dawned, and clients began requesting the shareholder banks to get involved in some of the other portals.
One banker from within the shareholder organisations explains the extent of the pressure. “Two of our biggest fund clients, who together made up a chunk of our business, were apparently pressing for us to deal with them on FX Connect from the moment it was revealed we would be providing liquidity exclusively on Atriax. Early in 2002 we were told – apparently in no uncertain terms – that if we were unwilling to do so, we would no longer be seeing their business. That made the decision a ‘no-brainer’, it was really a question of when we would do it.”
Having taken the decision in principle to quote on FX Connect, the banks pushed ahead with their negotiations to merge with FXall, thus increasing the positive signals they could send to their clients. Why this merger failed to happen is the subject of many different theories. Some sources within the company say that there were hidden costs for the Atriax shareholders connected to the hosting contract with Loudcloud and that one bank in particular baulked at paying these costs.
Other sources have suggested that the FXall shareholders were determined to extract the most favourable terms possible in response to Atriax’s initial decision not to offer them a shareholing in the portal. Still others are convinced that FXall’s management were aware that a long, drawn out negotiation period, during which details of the transaction (especially that FXall technology would be used by the merged entity) would damage Atriax’s marketing drive and further the shareholders’ resolve.
Allanson was unwilling to share his thoughts on what caused the merger to fail, although he says that the shareholders showed little enthusiasm for the half-hearted negotiations with FX Connect and Currenex that followed. “Suffice it to say,” he concludes, “That the closure of Atriax created fundamentally the same outcome as a merger – consolidation of the industry and all the banks quoting on the remaining portals.”
If this pressure was not enough, there were financial issues at stake. According to a board level source within Atriax, the company’s contract with system host Loudcloud involved a clause requiring full payment in the event of a merger. Such a payment would obviously have been the shared responsibility of the shareholders. In general, the source says, the merger terms from FXall were agreeable, but a few days before Atriax was due to reply to them, one shareholder came back to the board and stated that the costs were too high, and that therefore they could not support the merger in its current format.
The other shareholders, who were willing to agree to the terms were unhappy at this, the source claims, but to compound the problem, Atriax then received another document from FXall regarding the merger which was allegedly, “substantially different to what we expected”. At this, the remaining shareholder baulked at the deal, which in effect meant no merger.
Allanson says that the company looked into carrying on with a much scaled down organisation with a lighter format, but it was soon realised that without exclusive access to the three banks’ liquidity (although Deutsche Bank had been on FX Connect from the start) it was impossible to carry on.
Sources also claim that Atriax made hurried moves to investigate a merger with other portals. Preliminary talks are said to have taken place with FX Connect, but failed as it apparently became clear that Global Link’s owner State Street would want a shareholding commensurate to the amount of business it was putting through the portal. This would have been the vast majority, and was therefore unacceptable to Atriax’s shareholders, the sources say. The sources also claim that tentative approaches were made to Currenex, but that this was much too late in the day and the die was already cast.
So where does the industry go from here? Should the approach be that apparently taken by Deutsche Bank in the wake of Atriax, to have no equity/management stakes in any of the portals but provide liquidity to them all? In the current environment there are many who believe that this is the most sensible approach – no development costs at the expense of uncertain returns a long time in the future, but with the opportunity to make money immediately from providing liquidity. Or is there another route which will provide adequate long term rewards for the upfront investment costs?
Allanson firmly believes that the only alternative solution is to provide something that the buy side will pay to use, but that it is definitely not being charged commissions on for FX trading. He is not sure the buy side will feel obliged to pay for STP solutions as long as there are competing portals in the marketplace. While he feels that there will be a move to prime brokerage and outsourcing of the back office function, creating a messaging system into the banks’ prime brokerage systems will be all that is required of the portals. Again, it will be difficult to charge for this in a competitive environment.
If there is an answer, Allanson believes it may ultimately lie in the consolidation of the banking industry and the blurring of distinctions between commercial and investment banks. More and more, the liquidity in the major asset classes is being provided by fewer organisations that are substantially the equity providers for the existing portals. “With very few exceptions,” he says, “The shareholders of FXall are the shareholders of EBS, for example. In an environment in which there is only one portal for buy side FX, these shareholders will have total ownership of FX data, ranging from dealable prices through historical data to – and this is the holy grail – flow information.
“This is all saleable information, easily deliverable over the Internet,” he continues. “Extend that to other asset classes and this data could easily replace that provided by organisations such as Reuters and Bloomberg. A link with a news agency would be the final piece of the jigsaw, and would provide a solution that will save banks on their market information bills, and they will all have a stake in the new information provider.” He adds the rider that to achieve this will require the total cooperation of the shareholder banks, something that has not always been in evidence in any joint venture to date.
Of course, one of the portals has gone some way to providing this. FX Connect has a very sophisticated and detailed information silo, however it is only covering a portion of the market.
Allanson believes that FX Connect is in a strong position going forward. He notes that all of the remaining systems are strong in certain areas, but he expects FX Connect to continue to dominate in its chosen space for some time, because it operates primarily in the fund management area and provides high quality operational support which has in turn created tremendous customer loyalty. FX Connect does have issues facing it, he notes for example that the system is suitable for very few corporates; however, it is in a good position to handle further shakedowns in the market.
He is not alone in this belief. As has been noted in these pages before (Profit & Loss, May 2002) many in the market see a link-up between FX Connect and a company with the ability to build on its technological infrastructure as creating an irresistible force. The head of e-commerce within a European bank goes as far as to predict market dominance if FX Connect and Currenex got together. “Their volume levels would be extremely high, they would have the technological expertise and innovation to develop further, and a client base eager to exploit the results.”
Allanson also sees the need for rationalisation and agrees it is likely to take place in a piecemeal fashion. “There seemed to be a general consensus after the closure of Atriax, that FXall would be the main beneficiary,” he says. “The move by the former Atriax shareholder banks to provide liquidity on other platforms challenges that consensus, as does the different strengths of the different platforms. Buy side customers continue to delay the decision over which portal to use except where one has clear cut benefits over the others – such as FX Connect for fund managers. Only a voluntary rationalisation of the industry will create a breakeven environment, and that requires the shareholders of the different portals to put aside their entrenched interests. This is much easier said than done and may ultimately only be driven by cash flow considerations.”
Allanson’s view is supported by several buy side clients spoken to by Profit & Loss who admit that their decision over a portal (in some cases as a replacement for Atriax) has been delayed by indecision from senior management over whether to commit to a particular model. It does not make economic sense to sign up to more than one, goes the argument, the problem is – in this era of the level playing field – which one? Many suggest they are unwilling to make that decision at this time and are content to trade with their banks on the latter’s proprietary systems.
What will convince them to select a portal? One fund manager has an interesting view on the conundrum, saying, “If we could be convinced that the future consolidation of the portal market would be via merger and acquisitions, rather than closures, it would make it easier to convince our management and trustees to sign up to one company or another. As long as the fear is there that we could be left in the lurch as some of Atriax’s clients were, we will continue to trade with our banks on their own systems.”
“Ultimately,” says Allanson, “The platform with cash flow and a spread of functionality will prevail. That company will need to make judicious economic decisions and build products that the buy side wants to use. High quality STP is a neccesity, as is prime brokerage, intra-company dealing, orders and widening the product mix to include other asset classes, especially money markets.
“Overriding all of this,” he continues, “Is the need for the executive management and the shareholders to work in tandem, with clear lines of communication and a commitment to building the business. Atriax had problems from the start, but as we reorganised the company along more professional lines, we realised how much we could achieve. We started to get results in the form of better functionality and customers who wanted to be on our system.”
What was right about Atriax is as much the lesson, as what was wrong. It is a lesson that those looking to develop their products further would be well advised to heed.
Sources from within Atriax have provided Profit & Loss with estimates of the cost of establishing and running the platform. It has not been possible to get official confirmation of these numbers, however, they are believed to be an accurate guide.
The total spend of the operation is said to stand close to the $100 million mark, and was predominantly made up of the following:
Purchase of CFOWeb.com $20 million
Further development costs $10 million
Hosting contract $8 million per annum for three years
Monthly running cost $3.25 million