• travelling 7.3 miles between 09:11 and 09:33;
  
• travelling another 10.4 miles between 09:55 and 10:40;
  
• making several further movements around Burton;
  
• remaining at another location between 11:01 and 18:12; and
  
• travelling again between 18:12 and 18:28. 
  

• the vehicle registration or tracker/device allocation;
  
• full timestamps and GPS coordinates;
  
• a map of the journeys;
  
• the tracker provider’s name;
  
• any accompanying job sheets, delivery records, invoices or other contemporaneous business records.
  

Quote:Dear Horizon Parking,

I appeal as the registered keeper. No admission is made as to the identity of the driver.

This Parking Charge Notice is the result of an obvious ANPR double-dip error. Your system has paired an entry recorded at 09:03:13 with an exit recorded at 18:28:43 and has incorrectly treated those two isolated camera events as evidence of one continuous parking period.

Contemporaneous vehicle-tracking evidence proves that the vehicle left the location shortly after the first visit and travelled extensively throughout the day. Among other movements, it travelled 7.3 miles between 09:11 and 09:33, a further 10.4 miles between 09:55 and 10:40, made several subsequent journeys around Burton-on-Trent, and travelled again between 18:12 and 18:28.

It was therefore physically impossible for the vehicle to have remained parked at Tesco for the period alleged. Your ANPR system has failed to record, recognise or correctly match an intervening exit and re-entry.

The attached tracking evidence conclusively disproves the allegation. The charge must be cancelled.

Before issuing any rejection, you are required to undertake a proper manual examination of all ANPR records relating to this vehicle on 25 June 2026. This must include any unmatched, orphaned, discarded or otherwise unpaired entry and exit images. Please preserve all relevant images, system logs, camera records and audit data.

The Notice to Keeper also fails properly to specify a period of parking as required by paragraph 9(2)(a) of Schedule 4 to the Protection of Freedoms Act 2012. It merely provides two ANPR timestamps showing a vehicle passing cameras. Those timestamps are demonstrably not the beginning and end of one continuous period of parking.

This appeal is also formal notification that the ANPR-derived assertion that the vehicle remained at the site continuously is inaccurate. Horizon must correct that inaccurate record and cease processing the keeper’s data for the purpose of pursuing this false charge.

Please confirm cancellation. Should Horizon nevertheless reject the appeal, it must provide a POPLA verification code together with its complete explanation and evidence concerning the missing exit and re-entry records.

Yours faithfully,

Registered Keeper

Quote:Subject: Formal complaint—false double-dip PCN and unsupported assertion that ANPR errors are "mercifully rare"

Dear Sir or Madam,

I refer to PCN [reference], which Horizon cancelled after accepting evidence that its ANPR system had combined two separate visits into one purported continuous stay.

In its cancellation response, Horizon stated:

"As with any technology, errors can unfortunately occur. Whilst errors are mercifully rare…"

That statement does not adequately address what occurred.

The Private Parking Sector Single Code of Practice requires ANPR images to be subjected to a manual quality-control check. The accompanying guidance specifically states that such checks are particularly important for detecting "double dipping", where the camera system has failed to record every occasion on which a vehicle entered and left the land.

The false PCN demonstrates that Horizon's pre-issue control failed in this case. Cancellation after the recipient produced contrary evidence does not establish that Horizon's original process was adequate.

Please therefore answer the following:

   1. Was the ANPR evidence manually reviewed before this PCN was issued?
   2. What precisely did that review involve, and on what date was it undertaken?
   3. Did the reviewer examine all captures, partial reads, rejected reads and unmatched vehicle images from the relevant period, rather than merely viewing the first entry and final exit images?
   4. Why did the review fail to identify that the images represented two separate visits?
   5. What objective data supports Horizon's assertion that such errors are "mercifully rare"?
   6. During the last twelve months:
      a. how many proposed PCNs were prevented from being issued because manual review identified a possible double dip or missed vehicle movement;
      b. how many issued PCNs were subsequently cancelled because two visits had been combined;
      c. how many were cancelled because of an incorrect or partial registration read; and
      d. what proportion of Horizon's ANPR-generated PCNs do those figures represent?
   7. What corrective action has been taken in relation to the cameras and review procedures at this location?
   8. Has the landowner or principal been notified that the system generated an erroneous charge?
   9. Has Horizon examined other PCNs issued from the same location to determine whether comparable errors occurred?

Please preserve all ANPR records, rejected reads, audit logs, review notes and internal correspondence relating to this PCN pending the conclusion of this complaint.

If Horizon is unable or unwilling to provide the figures relied upon, it should withdraw the unsupported assertion that errors are "mercifully rare".

Please provide a final complaint response so that the matter may be referred to the British Parking Association, the DVLA, the landowner and my Member of Parliament.

Yours faithfully,

[Name]

Quote:Subject: "Mercifully rare"—request for action over false private parking ANPR charges

Dear [MP's name],

I am asking you to raise an important consumer-protection issue concerning the use of Automatic Number Plate Recognition by private parking companies.

Horizon Parking recently issued a parking charge based on ANPR images purporting to show that my vehicle had remained at a location continuously between [time] and [time]. The vehicle had in fact made two separate visits. Horizon's cameras had failed to record, or its system had failed to use, the intervening exit and re-entry.

I was able to provide evidence proving the error, and Horizon cancelled the charge. Its explanation was:

   "As with any technology, errors can unfortunately occur. Whilst errors are mercifully rare…"

That response raises rather than resolves the concern.

The private parking sector's own Code requires a manual quality-control check before ANPR evidence is used to issue a parking charge. The Code specifically identifies "double dipping"—the failure to record every entry and exit—as an error which manual checking should address.

Either Horizon did not perform the required check, or the check was ineffective. In either case, the motorist received a false financial demand and was required to produce evidence to disprove an allegation generated by Horizon's defective records.

Horizon's description of such errors as "mercifully rare" is unsupported unless it publishes:

• the number of vehicle movements processed;
  
• the number of potential PCNs rejected during manual review;
  
• the number of PCNs cancelled because of double dipping or incorrect reads;
  
• the number of motorists who paid before the error was identified; and
  
• independently audited error and capture rates.
  

There is presently no publicly available sector-wide information answering those questions.

The significance is amplified by the scale of private parking enforcement. Approximately 14.4 million private parking charge notices were issued in 2024/25. Official DVLA figures also show that manual review identified 45,427 number-plate misreads among 511,801 unique potential enforcement detections examined by staff. Those figures demonstrate the importance of robust human verification before financial or enforcement action is taken.

I ask that you:

1. Write to Horizon Parking requesting the factual basis for its "mercifully rare" assertion and details of the failed manual review in this case.
   
2. Ask the DVLA whether it records how many keeper-data requests result in PCNs subsequently cancelled because of ANPR errors, missed captures or double dipping.
   
3. Ask the Secretary of State responsible for the statutory Private Parking Code of Practice whether the Code will require:
   
4. independent testing and certification of private parking ANPR systems;
   
5. an auditable manual-review record for every ANPR-generated PCN;
   
6. mandatory reporting of false positives and post-issue cancellations;
   
7. publication of operator and site-specific error statistics; and
   
8. effective sanctions, including suspension of DVLA data access, where systemic failures are identified.
   
9. Consider tabling written parliamentary questions on these matters.
   

A private parking operator should not be permitted to dismiss ANPR errors as "mercifully rare" when it neither publishes nor independently verifies the figures necessary to support that assertion. Where automated evidence is used to issue financial demands to the public, accuracy must be measurable, auditable and transparent.

Yours sincerely,

[Name and address]

Quote:Dear Mr Middleton,

Thank you for your response.

Your explanation confirms that the Parking Charge was issued because Horizon's ANPR record showed only two events: an entry at 09:03 and an exit at 18:28. You now accept, on the evidence provided, that the vehicle actually left the site after the morning visit and returned again later that afternoon.

However, your explanation raises a more important technical issue which has not been addressed.

You state:

"it appears that the cameras have missed your exit from the car park in the morning and subsequent entry in the afternoon."

There is a critical distinction between:

  1. the cameras producing no images whatsoever of those two intervening movements; and

  2. images having been captured, but the ANPR/OCR system failing to read, correctly identify, match or associate those images with the vehicle's correct registration.

Those are entirely different failure modes.

An OCR or recognition failure means that an image may still exist, even though it was not associated with the correct VRM. A complete capture failure means that the vehicle passed through the camera-controlled entry or exit point and no usable image of that movement was generated at all.

Your response appears to assert the latter: that the vehicle was successfully captured entering at 09:03; was then completely missed when it left; was completely missed again when it returned later that day; and was then successfully captured leaving at 18:28.

That is a striking sequence and requires considerably more explanation than the statement that errors can occasionally occur.

The probability of two complete photographic capture failures for the same vehicle depends upon Horizon's actual physical capture rate, which you have not disclosed.

By illustration only, if the cameras successfully captured an image on 99.5% of vehicle movements, the probability of completely failing to capture both of two specified movements, assuming independent failures, would be:

0.5% × 0.5% = 0.0025%

or approximately 1 in 40,000.

If the physical capture rate were 99.9%, the probability would be:

0.1% × 0.1% = 0.0001%

or approximately 1 in 1,000,000.

I do not suggest that either figure represents Horizon's actual performance. That is precisely the point. Horizon has described these incidents as occurring in a "very small number of instances" while declining to provide the data necessary to establish what "very small" actually means.

Your further statement that there were no "partial reads, rejected reads and unmatched vehicle images for your vehicle" does not resolve the issue.

If an image was captured but the OCR system read the registration incorrectly, assigned a different VRM, failed to generate a VRM, or placed the event into an unmatched, rejected or exception dataset, that image would not necessarily appear in a search conducted against the correct registration number.

Accordingly, please answer the following precise question:

Has Horizon's technical investigation positively established that no image whatsoever was captured on either of the two intervening movements, or has Horizon merely established that no correctly recognised and matched ANPR record for those movements exists against the vehicle's correct registration?

Those propositions are fundamentally different.

If Horizon has established that no image whatsoever was captured on either occasion, please confirm:

   1. what records or underlying camera data were examined by the technical team to reach that conclusion;

   2. whether raw camera records were examined rather than simply conducting a search against the correct VRM;

   3. whether any camera fault, obstruction, lane issue, lighting condition, tailgating event or other technical or environmental cause was identified;

   4. whether both failures occurred on the same camera or lane; and

   5. what corrective action was taken following the technical investigation.

If, however, Horizon merely found no correctly matched ANPR event against the correct VRM, that does not establish that the underlying images do not exist.

You also state:

"For every Parking Charge, we review the events of the vehicle to ensure we have each entry and exit."

In this case Horizon demonstrably did not have each entry and exit.

If no intervening images existed at all, I accept that the individual member of staff reviewing the available images could not reasonably have been expected to identify images which did not exist. That would, however, expose a different weakness in Horizon's quality-control process.

Two isolated ANPR captures establish only that the vehicle was present at two particular moments. They do not, without further evidence, establish continuous presence throughout the entire period between those captures.

A manual review which merely confirms that there is a first entry image and a later exit image cannot establish that the ANPR record is complete.

This is particularly significant because the Private Parking Sector Single Code of Practice expressly recognises the problem of "double dipping", where cameras fail to record every entry and exit, and requires manual quality control to guard against such errors.

The issue is therefore no longer simply whether somebody looked at the photographs before issuing the Parking Charge. It is whether Horizon's review process is actually capable of identifying an incomplete ANPR sequence before a financial demand is sent to a motorist.

In this case it plainly did not do so.

SUBJECT ACCESS REQUEST

For the avoidance of doubt, this section constitutes a formal Subject Access Request pursuant to Article 15 UK GDPR.

I require a copy of all personal data held by Horizon Parking relating to me and/or vehicle registration [VRM] arising from the vehicle's presence at [site] on [date].

This request includes, but is not limited to:

   1. Every ANPR image depicting the vehicle on that date, including overview images and number-plate or plate-patch images.

   2. Any image depicting the vehicle regardless of what VRM, if any, Horizon's ANPR/OCR system assigned to that image.

   3. Any image of the vehicle where the registration was unread, partially read, incorrectly interpreted, rejected, manually amended or associated with another registration.

   4. Any unmatched, unpaired, orphaned, rejected or exception ANPR event which depicts the vehicle.

   5. All ANPR event records associated with the vehicle, including timestamps, camera identifiers, lane identifiers, entry/exit classifications and machine-generated VRM data.

   6. Any OCR confidence score, read-confidence data, exception flag, rejection code or other machine-generated metadata associated with any image or event depicting the vehicle.

   7. The complete audit trail relating to the generation and manual review of this Parking Charge, including the date and time of review and any record showing what information and images were available to the reviewer.

   8. All records, notes, findings and communications produced or considered during the technical investigation referred to in your response, insofar as they contain my personal data or data relating to the vehicle and its ANPR events.

   9. All internal notes, appeal records, complaint records and cancellation records containing my personal data or relating to the processing of the vehicle's ANPR data.

The search must not be restricted merely to records indexed under the correct VRM.

The very issue under investigation is whether an image of the vehicle may have been captured but incorrectly recognised, rejected, unmatched or associated with another registration.

The vehicle is:

   [make/model/colour]

The two known images already held by Horizon at 09:03 and 18:28 provide an unambiguous visual reference from which the vehicle can be identified.

The approximate intervening movements were:

   Morning exit: [approximate time]

   Afternoon re-entry: [approximate time]

Those time windows should permit a focused and proportionate search of the relevant camera and ANPR records.

Please also preserve all relevant raw images, event records, associated metadata, audit information and technical-investigation records pending completion of this Subject Access Request and complaint.

Finally, you have stated that this type of occurrence happens only in "a very small number of instances".

You are of course entitled to decline voluntary disclosure of commercially held aggregate statistics. However, that does not answer the separate question of what objective evidence supports Horizon's assertion about the frequency of such failures.

Please therefore confirm:

   1. whether Horizon records the number of ANPR-generated Parking Charges subsequently cancelled because one or more entry or exit events were missed;

   2. whether Horizon records cancellations caused by incorrect, rejected or unmatched ANPR reads;

   3. whether Horizon has calculated the proportion of its ANPR-generated Parking Charges affected by such failures;

   4. whether those records are examined as part of BPA compliance auditing; and

   5. whether your statement that these incidents occur only in "a very small number of instances" is based upon measured data rather than assumption.

Cancellation of the individual Parking Charge does not resolve this complaint.

The complaint concerns the reliability of Horizon's ANPR evidence, the effectiveness of its pre-issue checking process, the precise nature of the two unexplained missing movements, and the evidential basis upon which Horizon describes these failures as exceptionally uncommon.

Please treat this as a continuation of the formal complaint and separately as a Subject Access Request.

If Horizon considers its internal complaints process exhausted, please confirm that this is your final complaint response so that the matter can be referred to the British Parking Association and raised with my Member of Parliament.

Yours sincerely,

[Name]

• a BPA complaint, focusing on the admitted double-dip failure, the ineffective pre-issue review, the unexplained change from two missing movements to "an ANPR capture" being missed, and Horizon's unsupported claims about how rare these failures supposedly are; and
  
• a separate ICO complaint, limited to whether Horizon actually carried out a reasonable search for images of the vehicle itself, rather than merely searching records associated with the correct VRM.