It is unfair to compare VFM with Digilog results, says Kerry Furber

On 8 and 15 April Insurance Times published results achieved by VFM/AXA, claiming that these have the upper hand over Digilog [the news item makes no such claim - ed] and the use of "lie detectors", the latter under the banner "Experienced interviewers are key to rooting out fraud." Digilog UK wishes to point out the following:

The 15 April article does not make it clear that the AXA results relate to household claims and yet draws comparison to a Digilog pilot with a totally different company in a totally different sector (motor) with figures based on a different success criteria.

The article encourages the reader to assume that because these processes are being compared, it is reasonable to make a comparison on the basis that the risks in each are in fact similar. Unfortunately, neither article represented such a comparison in an appropriately balanced context, as the following very important factors need to be taken into account:

  • Different underwriting philosophies and practices mean that individual insurers have immensely varying fraud risk levels
  • Different sectors and product lines also exhibit dramatically differing fraud risk levels
  • The net result of these two factors, when combined, is that any accurate screening technique will identify a much smaller incidence of fraud on a low risk book of business as opposed to a high. It is naive to expect that 45% per se of claims in any given situation will be defeated by any solution
  • The 15 April article states that the VFM solution is defeating 45% of claims, inferring by way of the headline, that this number relates to the number of fraudulent cases discovered - yet within the figures released by AXA and published on 8 April by Insurance Times in a pie chart, the fraud related defeat rate is clearly stated as 2.2%. This was not included in the 15 April piece and thus the associated commentary is misleading
  • The 8 April article shows that 33% of the AXA cases were withdrawn (and are clearly shown as outside the 2.2% fraud related success) - but offers no reason for this. Unless an insurer already knows that a customer is being disingenuous at the time of withdrawal, is there not a very real possibility that some of these withdrawals may have been genuine customers - possibly giving up their claims following lengthy and frustrating examination

  • The statement on 8 April that "18% of claims were queried at the first stage of the VSA process (Highway) and of that, around 60% of the claimants withdrew from the process" is inaccurate and grossly misleading [These figures were supplied to Insurance Times by Kerry Furber in a telephone conversation on 21 August 2003 - ed].
  • The Highway figures, cited by Insurance Times on 8 April, were released to the media, including Insurance Times, in 2003 and show that Highway is able to quickly and accurately identify over 70% of claimants as genuine and stream these claims to fast track settlement. The fact of the matter is that

    30% of claims are queried at the first stage of the VSA process and are hence high risk, 61.5 % of which are ultimately being defeated - this equates to 18% of all motor theft claims screened being defeated/repudiated on fraud related grounds.

    Hence the Highway success figures are drawn only from those cases that are fraud or suspected fraud and do not include any zero settlements from the 70% low risks cases (representing approximately 20% additional zero settlements). It is clear that the AXA figures include fraud related (2.2%) and all other zero settlements.

    Digilog AVS solutions has now processed many thousands of claims from the first point of notification and has recently been deployed by three major clients, each of which has been using VFM or a similar style of cognitive interviewing technique for considerable periods.

    None of the clients achieved anything like the 'successes' indicated by AXA and all suffered from an inability to apply the cognitive interviewing process effectively in a volume screening, front end deployment, where time and accurate decision making is of the essence. Each of these clients is reporting significant increases in handling efficiency when using our process, together with improved fraud detection over their previous cognitive interviewing practices.

    In the AXA pilot, of the 937 claims over the four months, there is no indication of the timescales involved in achieving individual case results, although it is interesting to note that over one third of them are still outstanding. This may suggest that the process takes a considerable time - possibly for both genuine and suspect customers alike.

    Digilog's solution is designed to address all claims or voice-based proposals from first notification, being deployable for call centres dealing with tens of thousands of notifications per annum, facilitating informed and empowered decisions in ten to 20 minutes, without any pre-filtering mechanism in place.

    This enables companies to positively screen out low risk customers very quickly and concentrate on delivering professional acceptance decisions or settlements to deserving and valued customers. The main advantage of the anti-fraud activity within our process is that validation/investigation work is then targeted solely on the high risk cases.

    We must also point out that Digilog's process is far more than a "hi-tech gadget". The majority of insurers have now seen our deployed process working and know that the process is a three-stage solution plus closure, with the technology featuring largely in stage one.

    The process includes cognitive interviewing and conversation management techniques, alongside narrative integrity analysis techniques that supersede and enhance more traditional cognitive interviewing delivery. We have plenty of feedback that supports this from customers. Additionally, claim handling times and customer complaints are far below previous levels.

    Further, we acknowledge that savvy claims handlers are critical in delivering success. The problem is that when screening volume claims at first point of notification, call centres are populated by call centre staff, not experienced specialists.

    In our processes, the call centre staff make highly empowered and efficient decisions from which the "savvy" handlers receive the high risk output, in which specific integrity related risks are now highlighted; hence they are able to concentrate on highly focused validation activity that ultimately evidences the fraud or manages the claim away.

    Ultimately, insurers want maximum savings while enhancing customer service. They want a cost-effective, volume handling process, effective at managing out fraud while being both customer and staff friendly. This is exactly what Digilog delivers.

    Kerry Furber
    Managing director