WITN01090100 Andrew Winn - Witness Statement

Evidence on official site

WITNO1090100
WITNO1090100

Witness Name: Andrew Winn

Statement No.: WITN01090100

Dated: 26//2/22

POST OFFICE HORIZON IT INQUIRY

FIRST WITNESS STATEMENT OF ANDREW FRANK WINN

I, Andrew Frank Winn, will say as follows...

Background

1. I joined the Royal Mail Group in 1996 and started working for Post
Office Ltd (POL) around 2001 in the Network Improvement Team. I
then moved onto the Problem Management Team in around 2005.
After about 18 months I went onto a Data management team that I
cannot recall the name of. In approximately 2007 I moved onto the

Finance Reporting Team.
2. In 2008 I moved into a restructured Product & Branch Accounting team

(P&BA) in a new post in the Debt Recovery team as an Analyst. I

Page 1 of 24
Training

WITNO1090100
WITNO1090100

reported to one of 4 senior managers who reported to the head of

P&BA who in turn reported to the Finance Director.

. The role initially focused on process improvements and looking at

accounting queries from branches but over time concentrated almost
entirely on accounting problems in branches and as a single P&BA
input point for both branches and other relevant teams (primarily
Network, Helpline (NBSC), Product & Security) along with the National
Federation Of Subpostmasters (NFSP) where accounting problems

could not be resolved under business as usual (bau).

. The post was renamed Relationship Manager to better reflect the role.

. left POL in June 2016. Consequently, I have been unable to access

any documents that might assist me in better recollecting some of the
questions raised. I have tried to indicate where I am not certain in the

tesponses provided.

. I underwent the same classroom training course as a sub postmaster

would. I found the basic operation relatively straightforward apart from

the balancing process which felt quite daunting.

. Iroutinely worked in several Crown Offices during Christmas pressure

and during strike action. I was able to process most transactions
without difficulty and balanced each time with no or nominal
discrepancies. As far as I am aware no Transaction Corrections (TC)

where generated following my actions in branch. Of course, there

Page 2 of 24
WITNO1090100
WITNO1090100

were more experienced staff around who I did refer to when less

mainstream transaction requirements arose.

8. I did on occasions input towards training revisions as processes and
products evolved. I was not aware of any fundamental approach to
the type of training provided to subpostmasters.

9. It was clear from branch and colleague feedback that some
subpostmasters did not feel adequately trained to operate without
experienced supervision once their training was completed.

10. Specific training aids would be distributed to branches on a regular
basis. This might be around a new product being rolled out, a change
to existing processes or to cover specific scenarios where
confusion/errors have been identified within the Network. These would
typically be identified through calls to NBSC or TCs being required in
P&BA.

11.1.am confused as to what is being raised in relation to communications
with Gurnos PO. There are possibly 2 issues.

12. Firstly there appears to have been a problem with understanding in
branch due to the use of acronyms — although the only one usedin the
example provided was ATM. POL did seem to reduce everything
down to an acronym and could confuse staff as well as
subpostmasters when new ones were introduced. My e mail
suggested that a glossary that was available to all might be an idea if it

could be updated regularly. Whether not understanding what a POL

Page 3 of 24
WITNO1090100
WITNO1090100

acronym meant would be a significant driver of losses to a branch is
questionable.

13. Secondly my verbal communications with this branch were only (to the
best of my knowledge) with Mr Etheridge. He clearly had little
knowledge of how to run a Post Office — his wife being the
subpostmistress. He repeatedly stated that surplus cash was held in
the safe and was given clear instructions on how to introduce the cash
to the Horizon balance. Eventually a site visit was arranged to confirm
that no surplus cash was actually on site.

14. The branch had made several errors in recording ATM dispensed
cash. This was probably due to a lack of knowledge. However all
ATM discrepancies were compensated for by TCs. No other

transactional errors could be identified to recover the missing cash by.

Function of Horizon

15. TCs were issued to branches to correct errors made when processing
sales transactions along with cash and stock remittance errors. These
would typically be identified by P&BA teams when reviewing their
relevant accounts or clients who identified errors.

16. Different teams might use different approaches depending on the
product. For example the Lottery team would review Horizon and
Camelot data overa month to issue one TC if required (to avoid simple
timing errors) whilst the Santander team would issue one TC per

transaction to aid branch understanding.

Page 4 of 24
WITNO1090100
WITNO1090100

17.Each TC would have two elements (a debit & credit) in line with
Horizon functionalities. TCs would be accompanied by evidence to
support their issue — this would be in different formats depending on
the product. They would arrive at the branch the next moming and
should resolve the specific error. The branch was required to accept
them prior to their next branch trading rollover.

18. The NBSC were the designated contact point for branches if the were
unable to resolve an error they had made or if they required assistance
in investigating a discrepancy.

19. Understandably in practice branches would try to resolve errors
themselves in the first instance but then lose the track of what they
had done. This would often turn simple fixes into complex untangling
exercises.

20. If a branch believed transaction anomalies had occurred the NBSC
would be their first point of contact. If they were unable to answer the
query the case would be passed onto Fujitsu — and possibly P&BA.

21. If a branch were unsure of the validity of a TC, believed they needed a
TC issuing or required investigation into a branch discrepancy they
would contact the relevant team in P&BA — possibly via the NBSC.

22. If they did not accept the answer provided, the branch would be invited
to submit an appeal in writing to The Relationship Manager laying out
their case. The requirement to write in was to ensure a proper record
was kept, to ensure all aspects of the case were covered and tried to

lead the branch into thinking through what the key problem was from

their perspective.

Page 5 of 24
WITNO1090100
WITNO1090100

23.On receipt of an appeal the case would be logged and any relevant TC
or settled centrally debt would be blocked until the case had been
investigated and responded to in writing.

24.My general approach in investigating the wide range of cases received
was to identify what the problem was (often not what the branch
thought it was), where the cash is (P&BA
accounte/Client/Customer/unknown) and whether it was possible to
recover the funds. This would form the basis of my responses.

25.Compensating TCs would be issued if appropriate and any relevant
blocked amounts would be un blocked.

26. There was no further designated escalation point in the process.
However if a branch was determined to continue to dispute they could
approach a senior manager.

27.An Operating Level Agreement (OLA) was drawn up with the Network
and Service Delivery teams. This was to define what P&BA’s principal
customers should expect from us covering aspects like time to issue
TCs, how disputes would be handled and how debt would be
recovered.

28. By the time the document was signed off I believe all parties were
aware of what was happening and this was reflected in the document.
As such its usefulness was, I believe, limited. That said it would be
useful for new staff or any future reorganisation.

29.The document was routinely updated during my employment but I

cannot comment on it now.

Page 6 of 24
WITNO1090100

WITNO1090100

30. In 2009 I made a presentation relating to TCs, debt recovery and
reporting. I am unsure who the audience was, but my best guess
would be the NFSP. The intention would have been to clarify the state
of play within P&BA and our aspirations for the future.

31.At the time many teams within P&BA were struggling to clear the
accounts they owned and this was widely regarded as the prime
target. I made the point that unless all TCs were accurate and
supported with appropriate evidence they wouldend up getting
disputed and simply get stuck on the debt teams’ accounts meaning
that overall P&BA funds owed would not reduce as quickly as might
appear from simply clearing product accounts.

32.1 cannot recall making a statement about investigating why large debts
might consistently emerge but that would seem a sensible thing to do.

33. The facility for branches to have a “Dispute” button on Horizon was
considered on more than one occasion. Branches, quite
understandably, were reluctant to accept TCs or unaccounted for debt
when carrying out their monthly trading period roll over.

34. From a POL perspective the management of such a facility appeared
problematic. In the first instance POL would need to contact the
branch to determine the basis of a dispute. Phoning a branch during
Office hours can be extremely inconvenient to busy branches whilst
writing to enquire and waiting for a response might take significant
time and make resolution of genuine issues harder than they need be.

35. More importantly would be who would have the power to reject the
dispute? For example many TCs were issued in pairs to credit one

Page 7 of 24
WITNO1090100
WITNO1090100

product and debit the other with no balancing impact on the branch. If
a branch accepts the credit (and being fully entitled to pocket the cash)
but then disputes the debit and refuses to accept further evidence then
POL must have the option to enforce the debt. Conversely where
cash appears to have simply gone missing, ownership is not so clear
cut.

36.My preferred approach was to stick with the dispute process to ensure
there is a clear path for branches to challenge losses, instigatea
comprehensive investigation and receive evidence supporting why
funds cannot be recovered before any relevant debt recovery action
takes place

37. Within the process was the facility to record a branch being forced to
accept a TC issued too close the Branch Trading to allow for proper
investigation. I believe that facility was utilised once during the
remainder of my employment.

38.1 produced 2 documents clarifying the purpose of the settle centrally
process. One will have formed a part of the weekly branch
communications document sent to each branch. The second I cannot
recall the medium or specific purpose.

39.Such communications would be produced if there was evidence
coming from potentially a wide range of sources to indicate that some
branches did not have a clear understanding of the settle centrally and

dispute processes.

Page 8 of 24
WITNO1090100
WITNO1090100

40. The process outlined how branches might challenge shortfalls by
initially discussing their case with the relevant product team within
P&BA via either the contact details on the relevant TC or the NBSC.

41. If the product owner could not satisfactorily resolve the query or if the
cause of a discrepancy could not be identified the branch could then
write to the Relationship Manager.

42. The idea of the “settle centrally” process was to allow branches to
move from one trading period to the next in a balanced position
without having to resolve cash discrepancies at the point. Apart from
giving the branch the opportunity to resolve a discrepancy they did not
believe was proper to them, branches would often be aware that
compensating TCs would be due to arrive to net off the relevant
discrepancy created by the settled centrally TC.

43. If discrepancies were not placed into a suspense type account,
branches would potentially struggle to establish the actual balance
state they were in, whist there would be less visibility of potential
balancing problems to POL.

44.The settle centrally process offered the opportunity to balance at
trading period end without using the make good cash process — either
adjusting cash gains or losses they really thought was not proper to
the branch or to fail to physically actually remove or add the cash as
declared through the make good cash option.

45.1 cannot recall what Problem Updates were. The examples provided

all postdate my POL employment.

Page 9 of 24
WITNO1090100
WITNO1090100

46.1 was not aware of how Fujitsu and POL monitored problems apart
from during my time as a Problem Manager where effectively Fujitsu
simply alerted the team to problems like outages and would routinely
update on resolution progress.

47.1 cannot recall being provided with detailed description of what
resolution processes entailed.

48.As Problem Manager my role was essentially to identify who would
own the resolution of a problem that had arisen, monitor and report
into POL on impact and/or progress.

49.1 cannot recall a scenario where I would have withheld authority for
Fujitsu to proceed the fix a problem that was proper to them.

50.1 was aware that Fujitsu were able to access branch data remotely as
this would be necessary to amend branch profiled to deal with
changes to product sets and branch profiles. Equally any software
issues would need to be amended remotely.

51.1 was aware that Horizon records could be altered remotely by Fujitsu.
I cannot recall from whom I obtained this information from. My
understanding was that any such action was tightly regulatedand
would require signoff from POL — I would not know who would have
such authority but would assume this to be at director level.

52.1 would typically produce letters to be sent to branches when software
problems had caused discrepancies. These would reflect the issue
‘and impact arising and outline what method would be used to resolve.

I believe all such letters would be signed by the Head of P&BA. This

Page 100f 24
WITNO1090100
WITNO1090100

would follow a review of the problem involving a range of POL
personnel and their subsequent agreement on how to proceed.

53.1 became involved in a review of the option ‘Make Good Cheque”
where the process was not self-evident in terms of branches
despatching the cheque. I included a “throw away” comment to reflect
the fact that we had a few examples were cheques making good
losses did not arrive for processing as they should have.

54. My understanding of Horizon accounting developed over time and was
possibly as good as anyone’s in POL by the end of my employment.
However my technical/coding knowledge remained relatively low.
Consequently, I would lean on others with a greater understanding

where necessary.

Suspense Account Problem

55.1 believe the “Local Suspense Problem” refers to a scenario in 2012
where a small number of branches in a specific transactional state
proved to be impacted by a error resulting in discrepancies in their
Local Suspense accounts. P&BA were able to confirm the
discrepancies where not proper to the branches ~— a balancing figure
would have been seen in a different account — I cannot recall which.
The values were then seen to reappear exactly 12 months after they
arose which generated Fujitsu involvement. In hindsight the issue
‘should have been investigated by Fujitsu in 2012 instead of the
accounting discrepancy being resolved by P&BA. It is possible that

Page 11 of 24
WITNO1090100
WITNO1090100

Fujitsu did investigate at that time but did not pick up the subsequent
reappearance of the problem.

56.1 was asked to check POLFS records to establish that all impacted
branches had been identified. Some further cases were identified. I
also asked the question as to whether any other branches could have
been in a similar transactional state as the others impacted, but had no
discrepancy. Logically I could not envisage any balancing issues
going forwards but the question seemed worth asking.

57.1 believe the issue was first identified by the Current Agents Debt team
who identified the duplicated suspense entries. Once the problem was
identified the financial consequences were identified by reviewing all
branches with problem on a stock unit basis and confirming the
recuring values on the relevant POLFS account.

58. Fujitsu would be able to confirm my calculations by checking the
transactional movement of the funds into suspense by FAD code and
stock unit. I cannot say how they might have identified any branches I
may have missed. However, my analysis would have been confirmed
by the Current Agents Debt team leader.

59. POL's Litigation Lawyer would typically talk through scenarios like this
to obtain a non-technical view of the time line and accounting impact

on branches.

Receipts, Payments and Discrepancies Issue

Page 12 of 24
WITNO1090100
WITNO1090100

60. Receipts and Payments discrepancies occasionally arose when small
numbers of branches did not balance at branch trading with the
balancing value showing in one of POLFS’s suspense accounts.

61. These issues would be monitored and identified by Fujitsu. I do not
know how they monitored for such events but understood it to be a
bau process.

62. The issue could also be identified and reported by branches reviewing
their branch trading statements but I cannot recall being aware of this
ever happening. Issues could also be identified in P&BA but only after

a time lag when a historic discrepancy had not been cleared.

Identification of errors

63.1 am informed that a subpostmaster reported a phantom cheque had
appeared on his cheque line. He claimed to have evidence but did not
present any. He did not have any discrepancy and did not ask to be
contacted about this.

64.Such an event would probably occur in multiple post office branches
each day due mistakes when cutting off the cheque line or choosing
the incorrect method of payment. Such events are simply corrected in
branch. NBSC would talk a branch through the required correctional
process if required.

65. If a single sided phantom cheque did appear on a branch’s cheque
holdings a receipts and payments mismatch would occur with a

resultant discrepancy.

Page130f24
WITNO1090100
WITNO1090100

66. If a phantom 2-sided transaction appeared then errors would result in
both cheque and client areas requiring TCs to be issued.

67. The principal aspect of my role involved “resolving branch accounting
issues.” The great majority of these were caused by branch error,
fraud, client, or supply discrepancies.

68. More than one subpostmaster described how stressful the period was
waiting for my investigation to conclude whilst a significant potential
loss hung over them. Consequently, I always attempted to investigate
cases in full but in as timely manner as possible.

69. Given that no accounting issue occurred in the claim above I believe
my approach to focus on the multiple cases with actual discrepancies
arising was entirely appropriate and my approach in a similar situation
would be the same in the absence of any altemate directive from my

managers.
Investigations

70.1 could be asked to investigate shortfalls by either branches, other
areas of POL, NFSP or Fujitsu. I cannot recall an instance where I
would not take on an investigation apart from when in the later months
of my employment was required to pass on any shortfall investigations
where the branch blamed the Horizon system to a Network team.

71. The start point of any investigation would always ideally be to obtain
the branch’s view (ideally in writing) on what had gone wrong, when,
what the impact has been and to supply any evidence they had they
felt might help their case.

Page 140f 24
WITNO1090100
WITNO1090100

72.1 would then review the case within POL utilising POLFS, Credence,
Horice and relevant teams within P&BA. I might also approach people
in other parts of POL to include Service Delivery, Network and NBSC.
I cannot recall exactly how I would approach the different case types
but I did produce a generic case type approach for use prior to leaving
the business.

73. Essentially the investigation was to initially find out what had
happened. In many instances there was problem but not what was
thought to be the issue in branch. For example if a branch had a loss
or circa £100 they would typically search their transaction logs for
transactions around this size. I do recall a fraud that hit branches after
POL introduced a product which I believe was like a gift card. The
product only allowed deposits to be made. Fraudsters wouldproduce
their card and ask to withdraw say £50. Branches would be unfamiliar
with the product, find a button that related to the card and process the
withdrawal. Most branches would quickly realise the transaction was
requiring the branch to take cash rather than issue and decline the
transaction. Some unfortunately did not notice and would see a £100
loss emerge and not be able to identify the cause.

74. Most branches balanced their branch routinely with little or no
discrepancies. This made it easier to identify when discrepancies
have occurred and allowed the opportunity to scrutinise transactional
history in close detail. Unfortunately, some branches reported wild
discrepancies over continuing weeks. Often, on close examination, it

was clear that some branches were reporting they were balancing the

Page150f24
WITNO1090100
WITNO1090100

branch by removing or adding cash at their trading period end. As
similar discrepancies would reappear straight away it was clear the
cash had not been added/removed. This made it very difficult to
identify the causes and times that actual problems had risen.

75. Once the issue has been identified the next stage would be to
determine where the cash was e.g. client, P&BA accounts, customer,
stock/cash centre.

76.The next stage would be to determine whether the funds could be
recovered. In most cases they could be and a compensating TC
would be issued. I would not be able to recover funds from a customer
for example.

77. Regrettably there were occasions when it was not possible for me to
identify the cause of discrepancies. Even where branches routinely
balanced well only to discover a loss within a limited period it might not
be possible to identify any transaction that was not accurately
recorded. The branch might then need to face the possibility in the
absence of a transaction not having been recorded at all (which never
arose to my knowledge) the discrepancy might well relate to cash
being simply removed from or added to the till. In some cases that
might relate to money moving to/from the retail side but some could
involve theft. Where it was possible to identify a specific period a loss
arose in I would routinely the cash movements on branch transaction
logs (Horice) to ensure the discrepancy accurately reflected the

difference between opening cash balance, cash transactions recorded

Page160f24
WITNO1090100
WITNO1090100

and final balance totals. No balancing anomaly ever arose in my
investigations.

78. Where value might be added I would involve branches with the
investigation. In most cases that would be to request specific
evidence the branch might hold or to ask them to review transactional
history within a specific timeframe. I do recall cases where
subpostmasters would knock on a customer's door if I was able to
identify a problem with a specific transaction.

79.1 routinely discussed my cases with branches over the phone and on
occasions this would lead to further lines of investigation.

80. TCs were issued with supporting evidence. My investigations would
detail the evidence available and lay out the reason(s) why I would not
be able to recover any missing funds. Often a satisfactory resolution
was found by obtaining a wider perspective of actions and events and
linking new data together. This was usually more productive than
trying to disprove evidence.

81.1 did have sympathy with branches in relation to the evidence available
to them apart from transaction logs and Trading Statements that would
be available in branch. This would have been even more difficult in
the period following the roll out of Horizon on line when P&BA teams
were typically in back log meaning many TCs were issued long after
the event, giving branches little chanceof recalling specific events.

82.Bar retuming to a paper-based accounting system I could not see an

alternate approach to suggest.

The dispute process

Page 17 of 24
WITNO1090100

WITNO1090100

83. My letter to Gurnos Post Office in August 2010 would represent a
typical response to a branch dispute.

84.1 don’t believe the letter suggests that a training gap caused a shortfall.
It may well have been the reason why the branch repeatedly input the
wrong figures from the ATM log into Horizon. Such cash
discrepancies would then be corrected by TCs. The short term ATM
cash discrepancies may well have reduced the visibility of cash
leakages that appear to have happened.

85. Despite a site visit and comprehensive reviews of the branch’s
transactional records, no outstanding product errors, or cash
unaccounted for were uncovered. My view was that there were no
further areas of investigation open to me. The problem appeared to be
one of cash disappearing from the branch. The relevant sentence, on
reflection, was not well worded as it is possible that the funds were
removed from the branch through some type of fraud or theft.

86. The subpostmistress would have had the opportunity to review ATM
and Horizon records and confirm that the TCs issued had netted off
the resultant discrepancy. This information would be available in
branch. She would also be able to check transaction logs and the data
entered into her branch trading statements.

87.My e mail to my colleague Robert Atkinson in 2012 suggests I had
received some feedback around the dispute process and wanted to
document roles and responsibilities. The communication does not
suggest any change of approach and appears to be a routine business

as usual piece of work.

Page 18 of 24
WITNO1090100
WITNO1090100

88.In 2014 all areas of P&BA were required to produce process maps of
their duties in a set format. Therefore, this was not a review as such
but it did give managers an opportunity to ensure their processes were
up to date and effective.

89.1 did not carry out an audit at Kevin Palmer’s branch. There was a
designated Audit team who would carry out branch audits.

90.1 may well have reviewed the POL, Lottery and branch records.
Essentially I would look at the scratch cards sent to the branch, the
sales input to Horizon and the balancing stock total at the time of the
audit. I cannot comment on the documentation provided by the office
as I have not seen it since 2013. If the documentation held in branch
indicated a problem with Horizon or Camelot, I would question why the
branch did not raise the issue back in 2008 but instead continued to
report scratch card holdings on Horizon that did not reflect actual
stocks held in branch.

9

=

.It appears as though the dispute from Lake Post Office related to a
cheque TC. My response outlines why the TC was required to have
been issued.

92.1 also explained why a cash/cheque adjustment in relation to the
cheque in question would have generated a cash gain that would
offset the disputed TC.

93. As the TC was issued some time after the problem transaction, the

gain should have shown clearly in the relevant trading period.

Unfortunately, after reviewing the branches transactional and

accounting records it became apparent that the branch routinely

Page 19 of 24
WITNO1090100
WITNO1090100

balanced with significant discrepancies at most trading period ends.
This suggested a lack of control and made it impossible to associate
single errors/actions to trading period discrepancies. I was able to
identify that over the previous months that the branch had made good
cash gains (i.e. removed cash from the till) of £1,222.98. The cheque

ertor under dispute would have been a part of these gains.

Prosecutions

94. If I was asked to provide a witness statement to aid a prosecution I
would have done so. Such a statement would outline any accounting
discrepancies that had been deemed relevant to the case. A report
would be effectively the same as would be produced as per a dispute.

95.1 cannot recall the basis upon which Mr Ismay decided not to allow

defence solicitors access to P&BA operations.
Civil Litigation

96. If I was asked to provide a witness statement to aid a civil litigation I
would have done so on the same basis as outlined above. In the case
of Castleton I was asked by the POL solicitor to help obtain pay
information. This did not fall under my remit but would be an example
of trying to assist through the wide range of contacts I had.

97.Whilst I never pursued debt recovery, my role clearly supported the
case for losses to be repaid. There would have been examples of
branches suffering losses where I did not consider recovery action to

be appropriate. Cases would generally fall into 2 categories.

Page 20 of 24
WITNO1090100
WITNO1090100

98. The first type would be where POL’s processes were not, in my
opinion, fair to branches. For example, branches might incorrectly rem
out stock to retum to the Stock Centre causing losses when stock was
not declared in the branch. The Stock Centre did not fully check all
retums making it difficult to find evidence to support any subsequent
claim from the branch.

99. Subsequently the process was amended to assume the branch was
correct unless the Stock Centre could confirmthe stock received had
been fully checked and aligned to the rem out recorded by the branch.
Another example would be around currency sales. A typical branch
error would be for a customer to ask to buy some euros. The branch
might then press “Buy” on Horizon instead of “Sell”. This would result
in aloss and currency holding discrepancy. It was simple enough to
correct the error in branch by processing 2 sell transactions. However,
this would result in a 2" “Margin” value being paid to POL’s currency
provider leaving the branch with a larger loss than was proper to the
error. A process change was negotiated to correct this anomaly.

100. The second type would be regarding losses that were clearly
due to fraud. While in many cases branches could be seen to not
follow basic processes, some frauds were new and innovative. Whilst
several examples were considered sympathetically within POL leading
to branch losses being written off there were some examples when I
did feel branches were hard done by. I cannot recall a specific

example of this.

Relationship with Fujitsu

Page 21 of 24
WITNO1090100

WITNO1090100

101. Whilst working in Problem Management I had regular contact
with Fujitsu. This was basically a one way communication whereby
Fujitsu would alert POL to a Horizon problem, the impact, estimated fix
time and subsequently provide updates to resolution. I did not have
the technical knowledge or awareness of Horizon accounting
functionality to identify issues that might not have been made clear to
me at that time.

102. Working within P&BA I found ad hoc contact with Fujitsu very
difficult. Access did become slightly easier towards the end of my
employment. When we did work with the relevant personnel the
outcomes were generally positive and efficient. There did remain
protocols to follow which I did get pulled up upon when I tried to
circumnavigate them.

103. Gareth Jenkins suggestions in May 2013 did not create a
significant change in information access. The change proposed by
Gareth provided me with access to a very useful piece of Horizon
transactional data to support my investigations into one type of error
dispute — card payment authorisations. It did open a door to obtain
other potentially useful data.

104. The problem would be knowing what else might be there and
what it could offer. I would estimate I only used this facility 5-6 times
and only for that purpose.

105. The real change in obtaining data from Fujitsu came with the
introduction of “Horice” which gavea limited number of POL personnel

almost real time access to a branch’s Horizon activity. The level of

Page 22 of 24
Reflections

WITNO1090100

WITNO1090100

detail was significantly greater than that obtained from Credence. As a
result, I believe the standard of my dispute investigations and
subsequent responses improved dramatically in relation to certain

cases.

106. in hindsight there were several things I would have done

differently. For example the Stock Centre and Currency issues could
have been resolved quicker if I had pushed harder. I can't think what
actions I might have taken that would have positively impacted this
whole case. Whilst I had faith in the integrity of Horizon I was always
comfortable with directing branches who claimed to have Horizon
faults to the appropriate point. I was not aware of any branch that
raised a dispute with me subsequently finding out their losses were

caused due to Horizon errors.

107. I believe the 2 main causes of the problems lie within the POL

culture (that existed widely within the business in not supporting sub
postmasters well enough) and Fujitsu's reluctance to allow POL

access to Horizon data (either in enough detail or earlier)

108. I was shocked on joining POL to discover that losses

established within branches were widely assumed, and prosecuted,
based on sub postmaster theft. Whilst I supported the view that

‘branches were responsible for their losses (where it was not possible

to identify and recover) it seemed clear to me that in many cases

losses were due to theft was carried out by others with access to the

Page 23 of 24
WITNO1090100
WITNO1090100

POL cash, customer fraud or, rarely, branch incompetence. Branches
should have been given assurance that losses reported in a timely
manner would be investigated and operational procedures amended to
help identify where a problem was.

109. I cannot fairly say whether the roll out of Horizon on Line should
have been delayed but clearly many branches struggled in the early
stages and support from NBSC and P&BA was largely inadequate
due, I believe, to inadequate training and project accuracy.

110. While not being directly involved with the Second Sight
investigation it seemed odd to stop their work when it became evident
their findings were not in line with POL expectations.

111. POL’s contract with Fujitsu should have been drawn up
differently. I do believe this was due in part to both parties’

apprehension with POL’s limited technical experience.

Statement of Truth

I believe the content of this statement to be true.

Dated: = 26 //2./ 22

Page 24 of 24