Advice Only for Greater Anglia - Delay Repay Fraud

[35B]

Any notions that data retention or privacy policies are in any way relevant here are a complete red herring. They are totally irrelevant, even more so during a criminal investigation.

The matter really is very simple.

1) People on this forum have admitted to submitting invalid/fraudulent claims DELIBERATELY on X number of occasions.
2) Those same people have also submitted some legitimate claims which the TOC appears to accept are legitimate.
3) The TOC requests an OPTIONAL out of court settlement (a "civil disposal") to prevent the case being referred to the police/prosecution.

I think this particular thread has ran it's course in the absence of further information from the posters, (who largely, appear reasonably content with the outcome they have reached).

All true, but the order is wrong. The admissions follow item 3 in your list, which has led to responses from the individuals in question that are based on various permutations of items 1 and 2. The retention of this data by GA, seemingly contrary to the provisions of their Data Privacy Policy, raises questions in and of itself about their entitlement to access that data.

I suspect the court are not going to care less about data being retained over 12 months and will tell the Defence that it is a separate matter and not in their remit and to complain to the ICO
I don't see that this is illegal
It is probably nonsense for any company to claim 12 months anyway as you can be sure that that Data is backed up somewhere so maybe they don't retain them but thier outsourced IT company do

Have a look at their policy
The 12 months is for gateline records on their own

A smart prosecutor could draw the courts attention to the suggestion that gate line records are matched against delay repay claims and those delay repay records will have the gateline records appended to them and added to the customer relations database which are then kept for the duration of the franchise.

You may want to look more closely at the concept of "Data Controller" - it is absolutely on GA to ensure that their supplier abides by the terms of GA's data retention policy, and if the provider does retain those records contrary to instructions, then that is a criminal offence by that supplier. As for the "smart prosecutor", it would be necessary to prove that the gateline records were appended as part of teh processing of the Delay Repay claim. As I said earlier, the sort of tangle that will keep lawyers profitably employed for some time.

 

[b0b]

As for the "smart prosecutor", it would be necessary to prove that the gateline records were appended as part of teh processing of the Delay Repay claim. As I said earlier, the sort of tangle that will keep lawyers profitably employed for some time.

If the dateline data was appended as part of the processing of the DR claim, the system should have noticed the anomaly at the time, and not over a year later.

 

[MikeWh]

Yes, it would be farcical and I am sure that GA could defend writing a longer retention clause into their privacy policy but they have not so it would be an entirely self-inflicted defeat.

I think the point that people are missing is the difference between data being stored in a database which can be searched using any number of queries, and a report which contains specific data legally held at the time that the report was generated. As long as such reports are only held while conducting justifiable legal proceedings then their contents are not subject to the original data retention policies.

 

[jon0844]

During the discussions here, we've (myself included) talked about a range of scenarios as to when it might not be obvious that a claim WAS genuine, and theorised about someone not travelling because they knew of a delay (but by staying at home or at work, they were still delayed because they travelled later) and so on.

But have any of these scenarios applied to the people who were actually contacted and all seem to have opted to settle, and admit they perhaps made one or more fraudulent claim (either not travelled at all, or maybe exaggerated the level of delay).

I can't see what GA has done wrong as so far it seems they're not randomly fishing - they've picked out people that seem more than unusual. Maybe they did this with smartcard data, maybe they put out some fake data for the third-party sites to publish, or maybe something else we're not aware of.

They also seem to have made clear how long they can hold data (the whole franchise) for certain purposes, of which this seems to apply.

 

[glitterjo]

Just popping back on here as I'd been on a deadline and had rather ignored the whole thing till today.

I promise I am a real person - it's quite weird that all those who joined have first names starting with J, I agree. And, nothing fishy, I googled Delay Repay Fraud (as advised in the original letter by GA) and this forum - and indeed, this thread, came up.

Someone waaaay back also questioned my ability to walk 30 minutes to the station but then being unable to stand up on a train. I had a mild stroke - can walk fine but my balance was affected quite badly. Fortunately (!) the pandemic has dragged on long enough that this is now almost completely better so when I have to go back to squeezing myself on a delayed and overcrowded GA train to work again, I won't have to worry about that. Small mercies, eh?

They did send over smartcard data and all the claims I'd made over a period from November 2019 to January 2020 - 27. I went through them all carefully and found that the data showed claims that were clearly wrong on six of those occassions. I know there's no reason why you, people on an anonymous internet forum, should believe me, but these were genuine errors (I think possibly the wrong dates put in). Otherwise - all just... Greater Anglia delays. Occasionally something a bit odd in the data that I can see why they'd flag it up but just me going to a meeting/doing something else other than going straight home or to the office. I can see why they pursued it though and I've offered to repay the compensation and costs for those claimed in error... will see how it goes.

 

[Titfield]

Just popping back on here as I'd been on a deadline and had rather ignored the whole thing till today.

I promise I am a real person - it's quite weird that all those who joined have first names starting with J, I agree. And, nothing fishy, I googled Delay Repay Fraud (as advised in the original letter by GA) and this forum - and indeed, this thread, came up.

Someone waaaay back also questioned my ability to walk 30 minutes to the station but then being unable to stand up on a train. I had a mild stroke - can walk fine but my balance was affected quite badly. Fortunately (!) the pandemic has dragged on long enough that this is now almost completely better so when I have to go back to squeezing myself on a delayed and overcrowded GA train to work again, I won't have to worry about that. Small mercies, eh?

They did send over smartcard data and all the claims I'd made over a period from November 2019 to January 2020 - 27. I went through them all carefully and found that the data showed claims that were clearly wrong on six of those occassions. I know there's no reason why you, people on an anonymous internet forum, should believe me, but these were genuine errors (I think possibly the wrong dates put in). Otherwise - all just... Greater Anglia delays. Occasionally something a bit odd in the data that I can see why they'd flag it up but just me going to a meeting/doing something else other than going straight home or to the office. I can see why they pursued it though and I've offered to repay the compensation and costs for those claimed in error... will see how it goes.

Thank you for the update and I hope your recovery is 100%.

I sometimes think that on forums we forget that peoples lives are affected and whether innocent or guilty we should treat all with respect and courtesy.

 

[Wolfie]

Just popping back on here as I'd been on a deadline and had rather ignored the whole thing till today.

I promise I am a real person - it's quite weird that all those who joined have first names starting with J, I agree. And, nothing fishy, I googled Delay Repay Fraud (as advised in the original letter by GA) and this forum - and indeed, this thread, came up.

Someone waaaay back also questioned my ability to walk 30 minutes to the station but then being unable to stand up on a train. I had a mild stroke - can walk fine but my balance was affected quite badly. Fortunately (!) the pandemic has dragged on long enough that this is now almost completely better so when I have to go back to squeezing myself on a delayed and overcrowded GA train to work again, I won't have to worry about that. Small mercies, eh?

They did send over smartcard data and all the claims I'd made over a period from November 2019 to January 2020 - 27. I went through them all carefully and found that the data showed claims that were clearly wrong on six of those occassions. I know there's no reason why you, people on an anonymous internet forum, should believe me, but these were genuine errors (I think possibly the wrong dates put in). Otherwise - all just... Greater Anglia delays. Occasionally something a bit odd in the data that I can see why they'd flag it up but just me going to a meeting/doing something else other than going straight home or to the office. I can see why they pursued it though and I've offered to repay the compensation and costs for those claimed in error... will see how it goes.

Re the stroke thing l've been there and had similar issues albeit travelling on the tube not GA. Are adjusted hours as a reasonable adjustment an option? That worked well for me.

 

[sheff1]

Be careful what you wish for.

This sorry saga may encourage GA and other TOCS to rewrite the rules so that you only get delay repay when you have presented yourself for travel for the "delayed" service at a station and that can be proven. This may be perceived to be unfair / harsh etc but if Delay Repay has increased very significantly then it is inevitable either or both claims will be scrutinised very carefully and the benefits of the scheme reduced.

Alternatively, if Delay Repay has increased very significantly then perhaps TOCs/Network Rail could take action to reduce the number and length of delays and hence reduce the number of claims. If there were minimal delays there would be minimal opportunity to make fraudulent claims .....

 

[Wolfie]

Be careful what you wish for.

This sorry saga may encourage GA and other TOCS to rewrite the rules so that you only get delay repay when you have presented yourself for travel for the "delayed" service at a station and that can be proven. This may be perceived to be unfair / harsh etc but if Delay Repay has increased very significantly then it is inevitable either or both claims will be scrutinised very carefully and the benefits of the scheme reduced.

We do not know what data in aggregate GA have, and whilst some on this forum have questioned their actions, personally I find it difficult to believe that GA would embark on this course of action unless they had good grounds to do so. That is not to say that there are not individual cases caught up in this which should not have been brought in the first case.

Likewise we do not know for certain the actual actions of those contacted by GA and whether their claims were always legitimate.

We can speculate till the cows come home but whilst this may encourage debate it is often pointless given that individual cases will turn on the "facts" which we may not be privy to.

It is one thing saying that you have to be at the station. It's a whole other thing saying that you have to stay there, particularly if there are little or no facilities.
If the TOCs really want to draw attention to how much compo they get when there are issues compared to the pittance that they pay out...

 

[Titfield]

The point is that the TOCS will be comparing the number of claims made to the number of delays occurring. This ratio, whatever it is called, should stay reasonably constant.

If however the ratio changes significantly then they will investigate because they will wish to identify the cause of the change in behaviour.

The TOCS will know how many delays they have, roughly how many passengers are affected and what % will claim. If the number of train delays stays the same but the number of claims goes up significantly then this needs to be investigated.

Statistical data is based on hard factual evidence. So for example the probability of a 10 journey per week passenger being delayed once a week will be x, delayed twice a week y and three times a week Z and so on. The chances of someones travel experience ie number of delays suffered can be calculated. If their experience changes significantly ie they go from a 1 times a week delay claim traveller to a 3 times a week delay claim traveller this would be flagged because of the statistical probability of this happening is quite low (but that isnt to say it doesnt happen). The mathematical technique they probably use is Standard Deviation from the mean but explaining this is a bit beyond me.

 

[FOH]

The point is that the TOCS will be comparing the number of claims made to the number of delays occurring. This ratio, whatever it is called, should stay reasonably constant.

If however the ratio changes significantly then they will investigate because they will wish to identify the cause of the change in behaviour.

The TOCS will know how many delays they have, roughly how many passengers are affected and what % will claim. If the number of train delays stays the same but the number of claims goes up significantly then this needs to be investigated.

Statistical data is based on hard factual evidence. So for example the probability of a 10 journey per week passenger being delayed once a week will be x, delayed twice a week y and three times a week Z and so on. The chances of someones travel experience ie number of delays suffered can be calculated. If their experience changes significantly ie they go from a 1 times a week delay claim traveller to a 3 times a week delay claim traveller this would be flagged because of the statistical probability of this happening is quite low (but that isnt to say it doesnt happen). The mathematical technique they probably use is Standard Deviation from the mean but explaining this is a bit beyond me.

Disagree on this. During the Southern horror story of a few years ago I racked up substantial Delay Repays for cancellations. My favoured train was generally always the fall guy whereas regulars travelling 20 mins earlier would be maybe less than half as affected. Operational decisions will therefore have quite an effect on distortions by and large as commuters really are creatures of habit.

 

[jumble]

All true, but the order is wrong. The admissions follow item 3 in your list, which has led to responses from the individuals in question that are based on various permutations of items 1 and 2. The retention of this data by GA, seemingly contrary to the provisions of their Data Privacy Policy, raises questions in and of itself about their entitlement to access that data.

You may want to look more closely at the concept of "Data Controller" - it is absolutely on GA to ensure that their supplier abides by the terms of GA's data retention policy, and if the provider does retain those records contrary to instructions, then that is a criminal offence by that supplier. As for the "smart prosecutor", it would be necessary to prove that the gateline records were appended as part of teh processing of the Delay Repay claim. As I said earlier, the sort of tangle that will keep lawyers profitably employed for some time.

In theory I am sure you are correct that the supplier can be prosecuted for retaining longer than instructed but in the real world ?
In many cases the backups are encrypted and the archive supplier has no interest or idea what is contained in them.
In reality I suspect the person who wants to create a fuss will to a solicitor who will say
You have a good case.
Could you transfer £2000 to my account for my fees and we can get started
Then what

 

[35B]

In theory I am sure you are correct that the supplier can be prosecuted for retaining longer than instructed but in the real world ?
In many cases the backups are encrypted and the archive supplier has no interest or idea what is contained in them.
In reality I suspect the person who wants to create a fuss will to a solicitor who will say
You have a good case.
Could you transfer £2000 to my account for my fees and we can get started
Then what

In the real world, any supplier who processes data in that way risks being put dramatically out of business when - and it's a when, not if, for a supplier that inept - it finds that data leaks out and it's not been running it's data in accordance with either the contracts it's signed or the law.

And if the storage provider just holds that data in encrypted form, then the data controller is firmly on the hook for not managing that retention period at all effectively.

 

[Titfield]

Disagree on this. During the Southern horror story of a few years ago I racked up substantial Delay Repays for cancellations. My favoured train was generally always the fall guy whereas regulars travelling 20 mins earlier would be maybe less than half as affected. Operational decisions will therefore have quite an effect on distortions by and large as commuters really are creatures of habit.

But Southern were fully aware of the increase in the number of trains cancelled / delayed and could factor this in accordingly. The fact you state that your favoured train was the fall guy suggests to me that they were looking at travel patterns / numbers of passengers etc and were allocating their resources to minimise the impact. Nearly all transport providers do this as a matter of course during disruption. This would not invalidate their ability to analyse the data to identify claims or series of claims to investigate.
The more I read through this thread the more I am convinced that GA are using big data and have got some very compelling evidence to encourage them to act in the way they are.

 

[Tazi Hupefi]

But Southern were fully aware of the increase in the number of trains cancelled / delayed and could factor this in accordingly. The fact you state that your favoured train was the fall guy suggests to me that they were looking at travel patterns / numbers of passengers etc and were allocating their resources to minimise the impact. Nearly all transport providers do this as a matter of course during disruption. This would not invalidate their ability to analyse the data to identify claims or series of claims to investigate.
The more I read through this thread the more I am convinced that GA are using big data and have got some very compelling evidence to encourage them to act in the way they are.

I suspect you are correct around "big data".

I did some digging and found that Abellio TOC (which seems to have several operators), use third party software to process the claims, looking at the footer, the page copyright is © Copyright 2021 Tracsis Travel Compensation Services Ltd.

Now whatever product Tracsis Travel Compensation Service offer seems to have a focus on fraud, for example:

Minimise your train operator’s risk of EXPOSURE to cross-TOC fraud with TTCS | Tracsis Travel Compensation

www.tracsistcs.com

"Tracsis" presumably the parent company seem to be involved in all sorts of large rail industry data and projects but that's just from a Google.

Maybe Greater Anglia just "upgraded" their software or bought a new module etc, or perhaps the system just got smarter, possibly with AI?

 

[Haywain]

"Tracsis" presumably the parent company seem to be involved in all sorts of large rail industry data and projects but that's just from a Google.

Tracsis are a significant supplier of specialised rail industry software in both commercial and operational areas.

 

[Horizon22]

The point is that the TOCS will be comparing the number of claims made to the number of delays occurring. This ratio, whatever it is called, should stay reasonably constant.

If however the ratio changes significantly then they will investigate because they will wish to identify the cause of the change in behaviour.

Not entirely sure on that - some people have varying shift patterns. Whilst some is true - i.e your average 9-5 commuter into Liverpool St will claim X times a year - flexible working patterns were growing far before Covid and there will be many legitimate "edge-cases". I doubt those that have been defrauding GA are a significant enough percentage to shift that ratio. In fact I suspect the number of people GA have contacted might well be lower than 1% of all DR claimants.

I think GA are right to go after these people as the vast majority of who have posted say they've either
a) misunderstood the system significantly or b) have made fraudulent claims. This suggests that their data is fairly solid.

 

[Titfield]

Flexible working could mean you are working from home more. Thus if the number of claims stays the same but the number of journeys undertaken is less then the hit ratio will be higher.

The TOC has access to a lot of the data and imho is analysing this data in great detail. It may well be that only 1% of claimants are being investigated but nevertheless that could be a substantial sum in aggregate.

TOCS arent going to investigate in Tracsis unless there is a financial benefit whether that is cheaper claims processing and / or identifying potential fraud.

Quite a few of the posters on this issue are explaining why travel patterns may be less regular than in previous years: wfh, flexible working etc and thus why claims on differing timed services may be legitimate.I am not disputing that.

What I am saying is that the TOCS have the power to analyse claims in far more detail and thus are more able to identify travel and claims behaviours worth investigating. They are not having to sit down and aggregate all the claims that Joe Bloggs has made. Its all there at the touch of the button and if it deviates significantly from the average or norm (whatever that might be) it is worth their while investigating. It is for them to decide what deviating significantly might be. The very fact that some posters on here have admitted to submitting shall we say a speculative claim would suggest that the TOCS have found lines of enquiry / behaviour worth pursuing / investigating.

 

[eoff]

What I am saying is that the TOCS have the power to analyse claims in far more detail and thus are more able to identify travel and claims behaviours worth investigating. They are not having to sit down and aggregate all the claims that Joe Bloggs has made. Its all there at the touch of the button and if it deviates significantly from the average or norm (whatever that might be) it is worth their while investigating. It is for them to decide what deviating significantly might be. The very fact that some posters on here have admitted to submitting shall we say a speculative claim would suggest that the TOCS have found lines of enquiry / behaviour worth pursuing / investigating.

If they are using Machine Learning software and have enough cases that get to a profitable resolution then they can train for maximum profit based on both the journey and claim data and details about the individual.

 

[Ediswan]

I suspect you are correct around "big data".

I did some digging and found that Abellio TOC (which seems to have several operators), use third party software to process the claims, looking at the footer, the page copyright is © Copyright 2021 Tracsis Travel Compensation Services Ltd.

Now whatever product Tracsis Travel Compensation Service offer seems to have a focus on fraud, for example:

Speculation, no evidence whatsoever. The letter 'J' has come up as a common factor. What if Abellio have signed up for a cheap/free limited trial service and have chosen J*.

 

[Tazi Hupefi]

Speculation, no evidence whatsoever. The letter 'J' has come up as a common factor. What if Abellio have signed up for a cheap/free limited trial service and have chosen J*.

It is absolutely clear that this Tracsis company are providing the technology. I don't see your point? So what if it's just a random letter of a surname? Clearly the process must work for any name.

I claim from East Midlands in normal times, and the claim form hasn't changed much really, so I'm assuming that Tracsis have been providing that for a while as well.

I'm sure some industry people on this forum will know more.

 

[Ediswan]

It is absolutely clear that this Tracsis company are providing the technology. I don't see your point? So what if it's just a random letter of a surname? Clearly the process must work for any name.

I claim from East Midlands in normal times, and the claim form hasn't changed much really, so I'm assuming that Tracsis have been providing that for a while as well.

I'm sure some industry people on this forum will know more.

My point is that Abellio may have taken up a trial offer that they can only use on a subset of their data. If they like the outcome, names A-I and K-Z may soon appear. As I said, speculation.

 

[Tazi Hupefi]

My point is that Abellio may have taken up a trial offer that they can only use on a subset of their data. If they like the outcome, names A-I and K-Z may soon appear. As I said, speculation.

Ah OK, but surely if that is the case, Tracsis must be using big data / AI to do this, because firstly they seem to provide this delay compensation service to many operators, so presumably have a massive amount of data simply from processing claims. Surely there must be millions of claims per year in normal times? That scale of data must be ripe for clever trend analytics and pattern matching. I don't know what data points they have, but surely must have quite a lot. I doubt they would process or investigate claims manually (at least initially) so some sort of computer algorithm must be in use to flag suspicious claims and then produce some sort of list for the operator to manually check (and send letters out where needed).

 

[35B]

Ah OK, but surely if that is the case, Tracsis must be using big data / AI to do this, because firstly they seem to provide this delay compensation service to many operators, so presumably have a massive amount of data simply from processing claims. Surely there must be millions of claims per year in normal times? That scale of data must be ripe for clever trend analytics and pattern matching. I don't know what data points they have, but surely must have quite a lot. I doubt they would process or investigate claims manually (at least initially) so some sort of computer algorithm must be in use to flag suspicious claims and then produce some sort of list for the operator to manually check (and send letters out where needed).

Tracsis may or may not have a lot of data, but the AGA privacy policy doesn’t obviously suggest that data collected by AGA is available to pool in Tracsis’ database for the use of other, non-Abellio, companies. More importantly, the point of AI and machine learning is not that the applications using it have a large database behind them, but that the technologies in that database are capable of learning from data as it arrives in a database, and detecting patterns in that dataset. I would be surprised to find that those patterns are especially similar between AGA and (say) EMR given the different service patterns and usage groups on the two franchises.

 

[Tazi Hupefi]

Tracsis may or may not have a lot of data, but the AGA privacy policy doesn’t obviously suggest that data collected by AGA is available to pool in Tracsis’ database for the use of other, non-Abellio, companies. More importantly, the point of AI and machine learning is not that the applications using it have a large database behind them, but that the technologies in that database are capable of learning from data as it arrives in a database, and detecting patterns in that dataset. I would be surprised to find that those patterns are especially similar between AGA and (say) EMR given the different service patterns and usage groups on the two franchises.

They actually say:

We are part of a Group of Companies and share administrative services and support. Your data may therefore be shared with other Group companies where appropriate. We are also required to pass certain customer data to successor franchisees, Secretary of State or Department for Transport.

I also suspect that it doesn't matter what the differences between TOCS are. If they are using big data, they'd surely profile each claim against TOCS specific criteria to determine validity/fraud. There's probably helpful trends looking at the data from the entirety of their operations, for example to stop people claiming to be on a East Midlands train whilst also somehow on a Anglia train at the same time in a different city.

 

[35B]

They actually say:



I also suspect that it doesn't matter what the differences between TOCS are. If they are using big data, they'd surely profile each claim against TOCS specific criteria to determine validity/fraud. There's probably helpful trends looking at the data from the entirety of their operations, for example to stop people claiming to be on a East Midlands train whilst also somehow on a Anglia train at the same time in a different city.

Precisely my point, and why I mentioned EMR - data is shared within the Abellio group. What that doesn't mean is that they can share the data with all franchises, or that the pool of data so shared is of particular interest to either - it is the circumstances of AGA that are interesting, not the data from elsewhere. The key word in "machine learning" is "learning" - and that requires the data to be relevant to what the system is actually dealing with, hence the concerns about bias in biometric recognition systems "taught" using skewed datasets.

 

[Tazi Hupefi]

Precisely my point, and why I mentioned EMR - data is shared within the Abellio group. What that doesn't mean is that they can share the data with all franchises, or that the pool of data so shared is of particular interest to either - it is the circumstances of AGA that are interesting, not the data from elsewhere. The key word in "machine learning" is "learning" - and that requires the data to be relevant to what the system is actually dealing with, hence the concerns about bias in biometric recognition systems "taught" using skewed datasets.

Tracsis seem to provide (using a subset of brands "Iblocks, exposure, TTCS") Delay Repay to operators I've not even heard of:

c2c
Thameslink
Southern railway
Transpennine Express
Great Western Railway
Avanti West Coast
Scot rail
Greater Anglia
East Midlands Trains
West Midlands Railway/London Northern Trains
Hull Trains
Southwestern Railway
Transport Wales

Maybe more.

Given a link I shared earlier talked about preventing "cross TOC fraud" - Tracsis seem to have a system that can share data even with other operators who aren't in the list above. It does caveat with "where data sharing agreements exist", which implies (to some extent) that some data sharing agreements already exist. From a legal perspective you probably don't need a data sharing agreement to share within the same "group" - as long as it is in the privacy policy like Abellio.

I did a "google" and found an actual Data Sharing Agreement for Northern and "First Group" here - and some of the companies above look like First Group companies on the face of it, although I defer to you guys on that point. It therefore seems probable that all these claims are being shared between all of the operators on a large national scale. The rationale is:

During the COVID-19 outbreak the industry suffered a significant reduction in rail travel and to facilitate
social distancing, relaxed ticket validation procedures. We and our industry partners have identified a
number of customers whom we believe may be falsely reclaiming a refund of fares and/or compensation
for delay from multiple operators.

Which would tie in with people getting letters around now - as COVID seems to be the factor. Not sure whether that's people claiming now when they have stopped travelling (and now dodgy claims are standing out like a sore thumb), or it's given them a chance to slow down a bit and take some time to implement technology during the quieter period.

So TTCS got to work to build a solution. The result? EXPOSURE is the new and essential industry tool for detecting, investigating and preventing cross-TOC fraud in relation to Delay Repay claims. Where data sharing agreements exist, it compares multiple components of Delay Repay claims from different Train Operating Companies (TOCs) and sends real-time and retrospective alerts concerning potentially fraudulent activity for further investigation.

 

[35B]

Tracsis seem to provide (using a subset of brands "Iblocks, exposure, TTCS") Delay Repay to operators I've not even heard of:

c2c
Thameslink
Southern railway
Transpennine Express
Great Western Railway
Avanti West Coast
Scot rail
Greater Anglia
East Midlands Trains
West Midlands Railway/London Northern Trains
Hull Trains
Southwestern Railway
Transport Wales

Maybe more.

Given a link I shared earlier talked about preventing "cross TOC fraud" - Tracsis seem to have a system that can share data even with other operators who aren't in the list above. It does caveat with "where data sharing agreements exist", which implies (to some extent) that some data sharing agreements already exist. From a legal perspective you probably don't need a data sharing agreement to share within the same "group" - as long as it is in the privacy policy like Abellio.

I did a "google" and found an actual Data Sharing Agreement for Northern and "First Group" here - and some of the companies above look like First Group companies on the face of it, although I defer to you guys on that point. It therefore seems probable that all these claims are being shared between all of the operators on a large national scale. The rationale is:



Which would tie in with people getting letters around now - as COVID seems to be the factor. Not sure whether that's people claiming now when they have stopped travelling (and now dodgy claims are standing out like a sore thumb), or it's given them a chance to slow down a bit and take some time to implement technology during the quieter period.

Let's go back to there being two concepts here - machine learning/AI, and the datasets themselves. Machine learning/AI does not depend on bringing an existing database to a new client; the technology by its nature should be able to learn from a dataset provided by that client, and draw out the patterns from what is specific to that client. The subtleties of abuse of DR on the GEML will rely on specific local circumstances which may or may not align to those on AN Other TOC; what matters is not the general patterns (where I'm sure revenue protection managers across the industry have war stories to tell), but the specifics. For example, I'm sure that the use of Oyster/Contactless and the role of Stratford as an interchange for routes shared with Crossrail must have an impact, in a way that other Abellio operators may well not face. At a much larger scale, look at how HMRC have implemented Connect to clamp down on tax evasion.

That leads on to the legal aspects. Though I am not a lawyer, I do have some experience with data sharing agreements in the technology business. Data sharing agreements are required between any two companies, including within the same group. Some groups have "Binding Company Rules" in place, but these are expensive and onerous to put in place, so are not widespread. Meanwhile, the uses to which a company may put data are dictated by the basis on which they acquire it - which is why the Privacy Policy is so important. That Tracsis provide the same services to multiple clients does NOT mean that they are automatically allowed to share the data provided to those customers by end customers - if you pause to consider how Cloud or Software as a Service operators work, you will quickly realise that they could not work without maintaining robust separation of data received from different clients even (actually, especially) different clients in the same industry. The agreement you point to demonstrates how precise this is - the "purpose" is specifically associated with fraud analysis as Covid refunds, and is very specifically between First Customer Contact (so First Group's call centre business who work for the FG TOCs) and Northern (whose routes overlap heavily with some of those TOCs, especially TPE and Avanti). It involves a limited set of data being shared in limited ways and is absolutely not a general agreement to pool data.

The confusion of these concepts with the power of effective data interrogation using modern tools is a shame, as I think you are absolutely right that the opportunity to use new, more intelligent, data analytics will be helping AGA perform these kind of checks, regardless of whether they use personal data from elsewhere.

 

[Wolfie]

Tracsis seem to provide (using a subset of brands "Iblocks, exposure, TTCS") Delay Repay to operators I've not even heard of:

c2c
Thameslink
Southern railway
Transpennine Express
Great Western Railway
Avanti West Coast
Scot rail
Greater Anglia
East Midlands Trains
West Midlands Railway/London Northern Trains
Hull Trains
Southwestern Railway
Transport Wales

Maybe more.

Given a link I shared earlier talked about preventing "cross TOC fraud" - Tracsis seem to have a system that can share data even with other operators who aren't in the list above. It does caveat with "where data sharing agreements exist", which implies (to some extent) that some data sharing agreements already exist. From a legal perspective you probably don't need a data sharing agreement to share within the same "group" - as long as it is in the privacy policy like Abellio.

I did a "google" and found an actual Data Sharing Agreement for Northern and "First Group" here - and some of the companies above look like First Group companies on the face of it, although I defer to you guys on that point. It therefore seems probable that all these claims are being shared between all of the operators on a large national scale. The rationale is:



Which would tie in with people getting letters around now - as COVID seems to be the factor. Not sure whether that's people claiming now when they have stopped travelling (and now dodgy claims are standing out like a sore thumb), or it's given them a chance to slow down a bit and take some time to implement technology during the quieter period.

Interesting. If folk have committed fraud l have precious little sympathy. There have been occasions where l didn't claim Delay Repay because l wasn't 100% of the information.

 

[Kite159]

Interesting. If folk have committed fraud l have precious little sympathy. There have been occasions where l didn't claim Delay Repay because l wasn't 100% of the information.

No doubt there will be some dodgy chancers who will keep hold of an off-peak return which hasn't been marked and keep an eye out for any dates within the period of validity where there was delays on the network in order to put a claim in saying they were caught up in those delays/cancelled train etc to get some money back from the ticket.