Knife offences fell 8% — yet 41% of adults experienced or witnessed antisocial behaviour.

Stop one scam route faster — purchase scams that start on online marketplaces and social media — then test whether losses fall

Decision

Approve a time-limited pilot covering one scam route — purchase scams that begin with a listing or advert on an online marketplace or social-media platform and are paid by Faster Payments from a personal account — with named participating banks and platforms, a written agreement on which signals each may share and the lawful basis for each, a response-time standard, customer safeguards, and a pre-registered phased evaluation; and decide whether to expand, change route or stop when it reports.

Accountable institution

Home Office (Fraud Strategy lead) and the Online Crime Centre (Home Office and National Crime Agency, with the City of London Police)

Geography

UK

Powers

Existing law; no new legislation for a pilot. Signals about suspected fraud can be shared under UK GDPR’s recognised legitimate interest for preventing crime (Data (Use and Access) Act 2025, in force 5 February 2026); bank-to-bank disclosures carry civil-liability protection under section 188 of the Economic Crime and Corporate Transparency Act 2023, which does not cover platforms; banks may hold a suspect payment for up to four business days under the Payment Services (Amendment) Regulations 2024; Payment Systems Regulator reimbursement rules already apply. Online Crime Centre funding comes from the Fraud Strategy budget.

Delivery partners

Participating sending banks and receiving payment firms (to be named in the pilot agreement); One or more online marketplaces or social-media platforms where the scams begin (to be named in the pilot agreement); City of London Police (Report Fraud) and the National Crime Agency, through the Online Crime Centre; UK Finance (industry case-sharing platform) and Stop Scams UK; Payment Systems Regulator and FCA; Ofcom (Online Safety Act duties); Information Commissioner’s Office (data-sharing guidance); An independent evaluator commissioned by the Home Office

Target population

Personal-account customers of participating banks who pay by Faster Payments for goods or services advertised on participating platforms, and the sellers and listings on those platforms.

Eligibility

Payments to new payees that match the agreed purchase-scam risk signals; business accounts, card payments and other scam types are outside the pilot but are monitored for displacement.

Scale

Set in the pilot agreement from the number of participating firms and the power calculation for the evaluation — not a national number. Purchase scams are chosen for the first pilot because they are the most common push payment scam, begin mostly online and are discovered quickly, so an effect on losses can be read within months; investment scams cause larger losses per case (£221.5m in 2025, 38% of push payment scam losses) but victims take longer to notice, so they are the natural second route once the method works.

Baseline

The Crime Survey estimated 4.5 million fraud incidents and 3.8 million adult victims in England and Wales in 2025–26, 47% of all incidents the survey measures; it does not split fraud by scam route. Bank data for the chosen route (UK, calendar 2025): 175,809 purchase-scam cases (71% of authorised push payment scam cases) and £118.1m lost, a record, of which 78% was returned to victims; 66% of all push payment scam cases began online; Faster Payments carried £460.9m of the £576.4m lost to these scams.

Source: Office for National Statistics, Crime in England and Wales: Appendix tables, year ending March 2026 (Table A1a (fraud incidents and victims)); UK Finance, Annual Fraud Report 2026 (Sections 08–10 and 12)

Change beyond existing law or funding

Already in place: mandatory reimbursement (88% of in-scope losses repaid in its first year), banks’ power to delay suspect payments, the industry case-sharing platform, telecoms scam signals to banks, and £31m to launch the Online Crime Centre from April 2026. Not in place: an agreed, lawful two-way flow of signals between banks and platforms for one scam route, a response-time standard, a customer appeal route designed for wrongly held payments and delisted sellers, and an evaluation that counts victims and losses against a comparison rather than blocks.

Do: Platforms pass banks the payee details and seller identifiers attached to listings they have flagged as fraudulent, and banks pass platforms confirmed scam reports linked to a listing or seller; each acts within an agreed time — the bank warns, holds or refuses the payment, the platform removes the listing or seller Changing: fewer payments reach scam accounts before the victim or bank notices, and scam listings stay up for less time

• Purchase-scam victims and losses per exposed customer in the pilot route

  Hypothesis to testEffect record E-A06-VICTIMS

  Untested: no evaluation shows cross-sector sharing reduces victimisation. Measured gross, before reimbursement, because reimbursement moves the loss to banks without preventing the scam. Blocked attempts are an activity count, not victims prevented.

• Genuine payments held or refused and genuine sellers wrongly removed

  Hypothesis to testEffect record E-A06-FALSEPOS

  A harm to measure from the first day, with the share of held payments later released as genuine and the time they were held.

• Time from first signal to action

  Hypothesis to test

  The operational measure the response-time standard sets; quicker action is a means, not an outcome in itself.

• Money recovered and returned to victims

  Hypothesis to test

  Banks already returned 78% of purchase-scam losses in 2025, mostly by reimbursement; recovery from scam accounts is reported separately so it is not mistaken for prevention.

• Displacement to other scam types, payment methods, banks or platforms

  Hypothesis to test

  Scammers can switch to card payments, cryptoassets, non-participating platforms or investment scams; the evaluation must look for this rather than assume it away.

Outcomes are kept separate and never added into one impact figure.

• Annual Fraud Report 2026

  Author or institution:

  UK Finance (banking trade association)

  Published:

  2026 (uploaded June 2026)

  Evaluation period:

  Calendar years 2020–2025

  Design:

  Descriptive statistics of confirmed cases and losses reported by member banks; scam origins from a subset of closed cases

  Sample or population:

  248,070 push payment scam cases in 2025 (counted by account, not person)

  Effect and uncertainty:

  Not an effect estimate. Purchase scams: 175,809 cases, £118.1m lost, 78% returned. 66% of cases began online (32% of losses), 17% via telecoms (28%). A trial sharing intelligence with one platform raised detection of fraudulent adverts by 30% — an activity measure, not fewer victims.

  Limits:

  Industry body arguing for platform and telecoms liability; origin data depend on what victims report; no causal analysis.

  Where: Sections 08–10 (APP fraud, enablers, purchase and investment scams) and 12 (payment type); Industry Response. Affiliation is provenance, not proof of quality.

• Enabling Cross-Sector Data-Sharing to Better Prevent and Detect Scams (RUSI Conference Report)

  Author or institution:

  Royal United Services Institute with Stop Scams UK (Kathryn Westmore and colleagues)

  Published:

  September 2022 (web page 20 October 2022)

  Evaluation period:

  Workshops April–May 2022

  Design:

  Conference report from three workshops with industry specialists

  Sample or population:

  Technical specialists from Stop Scams UK members (banks, telecoms and technology firms)

  Effect and uncertainty:

  No effect measured. Found the law already permits much sharing for fraud prevention and that the barriers are mainly cultural and about risk appetite; recommended starting small with prioritised pilot use cases, such as banks sharing mule-account details with platforms.

  Limits:

  Industry consensus, not evidence of effect; predates mandatory reimbursement and the 2025 data law.

  Where: Full report sections II–IV, pp.7–13. Affiliation is provenance, not proof of quality.

• Fraud Strategy 2026 to 2029

  Author or institution:

  Home Office

  Published:

  9 March 2026 (updated 30 April 2026)

  Evaluation period:

  Strategy period 2026 to 2029

  Design:

  Government strategy

  Sample or population:

  UK (policing measures England and Wales)

  Effect and uncertainty:

  No effect estimate. Commits £31m to launch the Online Crime Centre; says the Data (Use and Access) Act 2025 establishes crime and fraud prevention as a lawful basis for sharing but that legal uncertainties remain, and opens a call for evidence on economic-crime information sharing; Ofcom’s fraudulent-advertising duty is expected in 2027; sets indicators but no reduction target.

  Limits:

  Promises indicators rather than an impact evaluation of the Online Crime Centre.

  Where: Paragraphs 3, 32–39, 54–57, 61 and 138–140. Affiliation is provenance, not proof of quality.

Closest feasible alternative

Make marketplaces verify sellers and offer protected payment for goods, and give platforms a share of the cost of reimbursement — UK Finance’s proposal — so firms that host scams have their own reason to stop them, without building a new data flow.

Strongest counter-evidence

The 2023 Fraud Strategy aimed to cut fraud to 10% below 2019 levels by the end of that Parliament, yet Crime Survey fraud rose. Telecoms firewalls have blocked over a billion scam texts since 2022 with no measured effect on victims, and Singapore’s reported falls come without a comparison group. Blocking counts can grow while victim numbers do not.

Source: Home Office, Fraud Strategy (May 2023) (Executive summary paragraph 3; paragraph 154); Home Office, Fraud Strategy 2026 to 2029 (Paragraph 46); Office for National Statistics, Crime in England and Wales: Appendix tables, year ending March 2026 (Table A1a)

Distributional effects

Gains go to buyers who would have been scammed; banks gain through lower reimbursement costs. Costs fall on genuine customers whose payments are held, on genuine sellers wrongly delisted, and on firms’ budgets, some passed on through prices. People without easy access to a branch or phone support are hit hardest by held payments.

Main downside

Privacy intrusion and false positives: wider sharing means more data about innocent customers and sellers moving between firms, and every held payment is a cost to a genuine buyer unless the appeal route is quick.

• law — A written data-sharing agreement and data protection impact assessment for each signal flow, with ICO guidance on the recognised-legitimate-interest basis; bank-to-platform disclosures lack the civil-liability protection that section 188 gives bank-to-bank sharing. (Requirement; owner: Participating firms’ data protection officers; Home Office; Information Commissioner’s Office)
• dependency — At least one large platform where purchase scams begin, and banks covering enough of the market to measure an effect, agree to take part and to act within the response-time standard. (Genuine veto: nothing proceeds without it; owner: Home Office and the Online Crime Centre with platform and bank chief executives)
• staff — Fraud analysts and data engineers in each firm and in the Online Crime Centre to triage signals and act within the agreed time. (Scarce resource shared with other priorities; owner: Participating firms; Online Crime Centre)
• infrastructure — A secure channel for near-real-time signals between banks and platforms, built on the industry case-sharing platform or the Online Crime Centre rather than a new system. (Requirement; owner: UK Finance; Online Crime Centre)
• public consent — Customer tolerance of held payments depends on a fast, staffed release route and clear notices; the law already requires banks to tell customers why a payment is held and to pay resulting charges and interest. (Requirement; owner: Participating banks; Financial Ombudsman Service for complaints)
• data — Comparable case, loss and false-positive data from participating and comparison firms, agreed before launch. (Requirement; owner: Independent evaluator with UK Finance and the Payment Systems Regulator)

How the cost is built up: Public cost = Online Crime Centre analyst and integration time + victim support + assurance + independent evaluation. Firms’ cost = each participant’s integration, analysts and customer contact + redress for wrongly held payments + seller appeals. Report both; attempted transfers stopped are not a spendable saving, and lower reimbursement is a saving to banks, not to government.

• Cross-sector pilot: data integration, analysts, customer support and evaluation

  One scam route across participating banks and platforms

  Not yet costed (ledger C-A06-PILOT)

  recurring; price year not stated; Pilot period; Not yet established

  Read from the ledger record, which has no amount: no source prices a pilot.

  Source: Royal United Services Institute / Stop Scams UK, Enabling Cross-Sector Data-Sharing to Better Prevent and Detect Scams (RUSI Conference Report) (Section IV (recommendations))

• Fraud Strategy 2026–2029 public funding (programme total, a possible route)

  All strands of the Fraud Strategy, including the Online Crime Centre launch

  at least £250m in total, 2026 to 2029, as stated in the strategy (“over £250 million”); fiscal years and price basis not stated, price basis not stated by the source (ledger C-FRAUD-STRATEGY-2026)

  recurring; price year not stated; 2026 to 2029; Not yet established

  Context, not the cost of this pilot: a stated lower bound for the whole strategy.

  Source: Home Office, Fraud Strategy 2026 to 2029 (Paragraphs 3 and 32)

• Firms’ own integration, analyst and customer-contact time

  Each participating bank and platform

  Not yet established per participating firm per year

  recurring; price year not stated; Pilot period; Not yet established

  Paid by the firms, not the Exchequer; must be reported so the total cost is visible.

• Redress for wrongly held payments (charges and interest) and seller appeals

  Customers and sellers wrongly affected

  Not yet established per wrongly held payment or delisting

  recurring; price year not stated; Pilot period; Not yet established

  Banks bear charges and interest from a delay by law; platform appeal costs fall on platforms.

  Source: legislation.gov.uk, The Payment Services (Amendment) Regulations 2024 (S.I. 2024/1013) (New regulation 94A)

• Victim support and assurance (Report Fraud follow-up, independent audit of data use)

  Victims in the pilot route; the data flows

  Not yet established per pilot

  recurring; price year not stated; Pilot period; Not yet established

• Independent evaluation

  Phased comparison of losses, false positives, time to action, recovery and displacement

  Not yet established per evaluation

  one-off revenue; price year not stated; Pilot period plus follow-up; Not yet established

Programme envelope (a possible route, not an award)

Fraud Strategy 2026–2029 (over £250m across the strategy): £250m

Allocation committed to this change

None

Remaining gap

Unknown: neither the public nor the firms’ share has been costed, and no allocation or industry contribution has been agreed.

Proposed financing

Public share from the Fraud Strategy budget (over £250m across 2026 to 2029, including £31m to launch the Online Crime Centre); participating banks and platforms pay their own systems, staff and redress under the pilot agreement. Nothing is allocated to this pilot yet.

Responsible payer

central government

Budget displaced

Online Crime Centre analyst time not spent on other fraud types or high-volume cyber crime; firms’ fraud-team time not spent on other controls.

Publish the pilot specification: the scam route, named participating firms, each signal and its lawful basis, the response-time standard, the customer and seller appeal routes, and the evaluation protocol. (Owner: Home Office and the Online Crime Centre, with participating firms. Proposed by UK Facts, not an official deadline.)

Time to first useful result: Time-to-action and false-positive data within weeks of launch; purchase scams are discovered quickly, so an effect on losses per exposed customer could be read within about six months of a phased start.

What could delay it: Firms agree terms and liability for bank-to-platform sharing The Online Crime Centre is operating with analysts in post ICO guidance on the recognised-legitimate-interest basis is published Comparison firms or phases can be agreed so the evaluation has a counterfactual

Baseline

For participating and comparison firms over the 12 months before launch: purchase-scam cases and gross losses per 100,000 personal-account customers making Faster Payments to new payees, payments held and later released as genuine, time from first report to action, money recovered, and other scam types and payment methods for displacement.

Comparison

A phased (stepped) start across participating banks or platform listing categories, so later starters act as the comparison for earlier ones; failing that, matched non-participating firms using the same case definitions.

Leading indicator

Share of signals acted on within the response-time standard, and the share of held payments later released as genuine.

Follow-up

At six and twelve months: losses and cases per exposed customer, false positives and complaints (including to the Financial Ombudsman Service), recovery, and displacement to other scam types, card and cryptoasset payments, and non-participating platforms.

Continue if

Losses per exposed customer fall against the comparison without an unacceptable rise in genuine payments held, and without a matching rise in displaced scams.

Adapt if

Signals are acted on but losses do not fall, or false positives are high: change the signals, the response or the route before expanding.

Stop if

Blocks rise but losses per exposed customer do not fall against the comparison, or wrongly held payments and complaints outweigh the losses prevented.

Thresholds

Not yet set: Acceptable false-positive levels and the size of loss reduction worth continuing for must be set in the published pilot protocol from the baseline and a power calculation; none exists yet.

Pick hot spots by harm and test each patrol surge against comparison places — inside the hot-spot funding forces already have

Decision

Add conditions to the 2026–27 Knife Crime Concentrations Fund grants (and any successor): select places using a harm measure that includes NHS data on violent injury, assign patrol days or places at random where operationally possible, measure a buffer zone around each place for displacement, publish stop-and-search, complaints and a resident survey for the chosen places, and set out in advance when a force will continue, change or stop.

Accountable institution

Home Office (grant conditions) with chief constables and Police and Crime Commissioners in the funded forces

Geography

England and Wales

Powers

Existing powers: Home Office grant conditions; operational policing powers of chief constables; data sharing between police, councils and NHS bodies under the Serious Violence Duty (Police, Crime, Sentencing and Courts Act 2022, sections 8 and 16, in force 31 January 2023). No new law.

Delivery partners

The 27 police forces in the Knife Crime Concentrations Fund (the Home Office announcement does not list them); Violence Reduction Units and Serious Violence Duty partners (councils, NHS trusts and integrated care boards, youth justice); Force crime analysts and a university or Home Office evaluation partner; Community oversight or independent advisory groups in the chosen areas

Target population

Residents of and visitors to the highest-harm small places (around 0.1 square kilometres) in the funded forces, especially people under 25.

Eligibility

Places in the funded forces ranked by a harm measure combining recorded violence and robbery with violent-injury data from NHS partners; comparison places drawn from the same ranking.

Scale

The existing funded scale: 27 forces covering 90% of knife crime in England and Wales, with places chosen by each force’s analysts. The change is to how places are chosen and tested, not a new national number.

Baseline

48,774 police-recorded knife or sharp-instrument offences in England and Wales in 2025–26, against 53,096 a year earlier. Funded hot-spot patrols cut violence and robbery by 7.0% in 2021–22 (95% CI 2.0% to 10.7%) but had no significant effect in 2022–23 (+0.9%, CI −1.6% to +3.5%). No published series records harm-weighted violence by hot spot, so the local baseline must be built first.

Source: Office for National Statistics, Crime in England and Wales: Appendix tables, year ending March 2026 (Worksheet B6); Home Office, Hot spot policing in England and Wales, year ending March 2022: evaluation report on Grip and bespoke-funded hot spot policing (Key findings); Home Office, Hot spot policing in England and Wales, year ending March 2023 (evaluation of Grip and bespoke-funded hot spot policing) (Key findings; section 2.2)

Change beyond existing law or funding

Already funded: the Knife Crime Concentrations Fund (over £26m, 27 forces, 2026–27) for patrol surges, CCTV, live facial recognition and knife arches in places identified by new Home Office mapping; £66.6m for Violence Reduction Units, the Serious Violence Duty and Young Futures Panels in 2026–27; and, since 2024–25, hot-spot funding extended to all 43 forces. Added here: harm-based selection using NHS injury data, random or staggered assignment so each force’s effect can be measured (11 of the 22 force evaluations in 2022–23 already randomised patrols in some way), displacement checks, published trust and stop-and-search measures, and a continue-or-stop rule.

Do: Put short, visible patrols and problem-solving into the places where serious injury concentrates, chosen by harm and assigned so they can be compared with similar places or days without patrols Changing: more deterrence where serious harm actually occurs, and evidence each year on whether it is working in that force

• Violence and robbery in patrolled hot spots, 2021–22 (existing funded patrols)

  Measured for this interventionEffect record E-A07-PATROL-2122

  A 7.0% fall (95% CI 2.0% to 10.7%) on patrolled days in funded forces. One programme year only.

• Violence and robbery in patrolled hot spots, 2022–23 (existing funded patrols)

  Measured for this interventionEffect record E-A07-PATROL-2223

  No significant effect (+0.9%, CI −1.6% to +3.5%). Shown so the earlier result is not read alone; the Home Office thinks waning or uneven implementation is the likeliest reason.

• Hospital admissions for sharp-object assault among under-25s in funded areas

  Measured for this interventionEffect record E-A07-SHARP

  About 550 fewer (−8.5%) from April 2019 to December 2024 against a synthetic comparison — the joint effect of Violence Reduction Units and hot-spot policing, which cannot be separated.

• Extra reduction in serious harm from choosing places by harm and testing each surge

  Hypothesis to test

  The marginal change this brief proposes has not been evaluated; harm-based selection was used in the Essex pilot, but no study compares it with count-based selection.

• Community trust and disproportionate stop and search in targeted places

  Hypothesis to testEffect record E-A07-TRUST

  A risk to measure, not a measured harm: the main review did not assess community effects.

Outcomes are kept separate and never added into one impact figure.

• Hot spots policing (Crime Reduction Toolkit)

  Author or institution:

  College of Policing, summarising Braga, Turchan, Papachristos and Hureau’s Campbell systematic review (2019)

  Published:

  Review 2019; toolkit entry updated 11 August 2021

  Evaluation period:

  Studies published up to the review’s search date

  Design:

  Systematic review and meta-analysis

  Sample or population:

  65 studies (62 in the meta-analysis): 51 from the USA, 4 from the UK

  Effect and uncertainty:

  Small, statistically significant reduction in crime; small diffusion of benefits to nearby areas; displacement rare; problem-oriented approaches did better than patrols alone. Mean effect size for violent crime 0.102 (as cited in the Home Office 2022–23 evaluation).

  Limits:

  Mostly US evidence; community effects not assessed; most studies run for under a year.

  Where: Toolkit entry “Effect” and “Moderators” (archived copy). Affiliation is provenance, not proof of quality.

• Hot spot policing in England and Wales, year ending March 2023 (evaluation of Grip and bespoke-funded hot spot policing)

  Author or institution:

  Home Office analysts, with technical support from the London School of Economics

  Published:

  27 March 2025

  Evaluation period:

  Year ending March 2023

  Design:

  Meta-analysis of 18 force-level repeated crossover evaluations (patrolled against unpatrolled days), 10 randomised by day

  Sample or population:

  22 evaluations from 19 funded forces; at least 90,000 patrols

  Effect and uncertainty:

  No significant programme-level effect on violence and robbery (+0.9%, 95% CI −1.6% to +3.5%); positive crime outcomes (crimes solved) rose 8.1% on patrol days; no displacement found in the 2021–22 buffer test; social cost £17.5m with no monetisable benefit that year.

  Limits:

  Day-on-day-off design can understate effects if deterrence carries over; patrol data quality varied; hot spots were chosen by violence counts.

  Where: Key findings; sections 1.4, 2.2 and 2.5. Affiliation is provenance, not proof of quality.

• Hot spot policing in England and Wales, year ending March 2022: evaluation report on Grip and bespoke-funded hot spot policing

  Author or institution:

  Home Office

  Published:

  15 February 2024

  Evaluation period:

  Year ending March 2022

  Design:

  Repeated crossover evaluations across funded forces, pooled

  Sample or population:

  Forces receiving Grip and bespoke hot-spot funding

  Effect and uncertainty:

  Violence and robbery −7.0% (95% CI −2.0% to −10.7%); the 2022–23 report later put that year’s crime-reduction benefit at £38.4m against £16.8m of cost (2024–25 prices).

  Limits:

  One programme year; not repeated the following year.

  Where: Key findings; section 2.3. Affiliation is provenance, not proof of quality.

• Violence Reduction Units, year ending March 2025: evaluation report

  Author or institution:

  Home Office

  Published:

  2 April 2026

  Evaluation period:

  April 2019 to December 2024

  Design:

  Generalised synthetic control (non-VRU areas weighted on pre-funding trends)

  Sample or population:

  20 Violence Reduction Unit areas, England and Wales

  Effect and uncertainty:

  About 550 fewer sharp-object assault admissions among under-25s (−8.5%; 95% CI about 164 to 902 fewer); homicide and recorded-crime results not conclusive.

  Limits:

  Cannot separate VRU work from hot-spot policing in the same areas.

  Where: Key findings; section 2.1; Table 2.1. Affiliation is provenance, not proof of quality.

Closest feasible alternative

Spend the marginal effort on problem-oriented work at the same places — changing the conditions that generate violence (lighting, licensing, the few people involved) — which the review found more effective and longer-lasting than patrols alone.

Strongest counter-evidence

The 2022–23 national evaluation found no significant effect of funded patrols, after a 7% fall the year before; effects may wane or depend on delivery quality, and the review cannot say what concentrated patrols do to trust where policing is already heavy.

Source: Home Office, Hot spot policing in England and Wales, year ending March 2023 (evaluation of Grip and bespoke-funded hot spot policing) (Key findings; section 2.2); College of Policing, Hot spots policing (Crime Reduction Toolkit) (Toolkit entry “Effect” (community effects not assessed; archived copy))

Distributional effects

Gains go to residents of high-violence places, who are often poorer and younger; the risks of over-policing fall on the same communities, particularly young men and some ethnic minority groups. Forces outside the fund gain nothing directly.

Main downside

Patrol hours and analyst time are diverted from response policing and investigation, and a randomised design means some high-harm places or days go without extra patrols during the test.

• data — NHS violent-injury data (emergency department and admissions) shared with force analysts at small-area level under the Serious Violence Duty. (Requirement; owner: NHS trusts and integrated care boards with Violence Reduction Units and force analysts)
• staff — Crime analysts able to build harm rankings and run randomised or staggered designs; officers for patrol dosage. (Scarce resource shared with other priorities; owner: Chief constables)
• dependency — Hot-spot funding confirmed for more than one year, so forces can plan successive evaluated cycles. (Scarce resource shared with other priorities; owner: Home Office)
• public consent — Community oversight agreed before deployment, with stop-and-search and complaints data published for the chosen places. (Requirement; owner: Police and Crime Commissioners)
• infrastructure — Patrol tracking (GPS or reliable logs) so delivered dosage can be checked against plan; quality varied in 2022–23. (Requirement; owner: Forces)

How the cost is built up: Added cost per force = analyst time for harm mapping and randomisation + NHS data-sharing set-up + buffer-zone analysis + resident survey + evaluation support; paid within or alongside the existing grant. Patrol hours are an opportunity cost, not new money.

• Harm mapping, patrol hours, problem-solving, partner time and evaluation in selected places

  One force area’s selected hot spots and comparison places

  Not yet costed (ledger C-A07-LOCAL)

  recurring; price year not stated; Each funding year; Not yet established

  Read from the ledger record, which has no amount.

  Source: Home Office, Hot spot policing in England and Wales, year ending March 2023 (evaluation of Grip and bespoke-funded hot spot policing) (Section 2.5 (costs))

• Serious Violence Reduction Programme (programme total, context)

  20 VRUs, the Serious Violence Duty and Young Futures Panels

  £66.6m in total, 2026–27, price basis not stated by the source (ledger C-HO-SVRP-2627)

  recurring; price year not stated; 2026–27; Historic cost

  A related programme envelope, not hot-spot patrols and not this increment.

  Source: Home Office, Knife crime hotspots to see surge in police operations (List of measures)

• Reference: social cost of the funded hot-spot programme evaluated in 2022–23

  Elements of Grip and bespoke-funded hot-spot policing evaluated, 19 forces

  £17,500,000 per programme-year

  recurring; 2024–25 prices prices; Year ending March 2023; Historic cost

  A historic programme cost for scale, not the price of the added conditions; no benefit could be monetised that year (2021–22: £38.4m benefit against £16.8m cost, same prices).

  Source: Home Office, Hot spot policing in England and Wales, year ending March 2023 (evaluation of Grip and bespoke-funded hot spot policing) (Key findings (cost-benefit))

• Added cost of harm mapping, randomisation, displacement checks and trust survey

  Each funded force

  Not yet established per force per year

  recurring; price year not stated; Each funding year; Not yet established

  Mainly analyst time and a resident survey; may displace some patrol spending within the existing grant.

Programme envelope (a possible route, not an award)

Serious Violence Reduction Programme, 2026–27: £66.6m, 2026–27

Allocation committed to this change

None

Remaining gap

Unknown: the added analysis and survey costs are not estimated, and funding beyond 2026–27 is not confirmed.

Proposed financing

Within existing 2026–27 allocations: the Knife Crime Concentrations Fund (over £26m to 27 forces) for patrol surges and the Serious Violence Reduction Programme (£66.6m) for Violence Reduction Units and data sharing, if grant conditions require the added analysis. No allocation to harm-based selection or evaluation is identified.

Responsible payer

central government, councils, NHS

Budget displaced

Some patrol or equipment spending within the grant moves to analysis and surveys; officer hours are not available for response policing or investigation.

Each funded force publishes its harm-ranked list of places, its comparison design (random or staggered), its displacement buffer and its community-oversight arrangements before the next surge cycle. (Owner: Chief constables and Police and Crime Commissioners, with Violence Reduction Units. Proposed by UK Facts, not an official deadline.)

Time to first useful result: Crossover or staggered results within one patrol cycle of three to six months; harm and trust measures after a year.

What could delay it: NHS injury data can be shared at small-area level quickly enough Forces accept random or staggered assignment of some patrols Funding continues beyond 2026–27 Patrol tracking data are complete enough to measure dosage

Baseline

For each force’s chosen and comparison places: violence and robbery, harm-weighted violence, NHS violent-injury attendances, stop and search, complaints and a resident survey, over the 12 months before the next cycle.

Comparison

Randomised patrol days or places (the crossover design used in most force evaluations in 2022–23, 11 of 22 with randomisation), or a staggered start across matched places; plus a buffer zone around each place for displacement.

Leading indicator

Patrol dosage delivered as planned in the chosen places, from tracking data.

Follow-up

Each cycle and annually: harm in target, buffer and comparison places; under-25 violent injuries; stop and search by ethnicity and age; complaints; resident trust; crimes solved.

Continue if

Serious harm falls in target places against the comparison without a rise in buffer zones, complaints or disproportionate stop and search.

Adapt if

Harm falls but trust measures worsen, or dosage is not delivered: change tactics (for example to problem-solving or procedural-justice-trained patrols) before the next cycle.

Stop if

Harm does not fall against comparison places over successive cycles, or harm rises in neighbouring places.

Thresholds

Not yet set: Each force needs its own baseline and power calculation; the Home Office grant conditions would have to set and publish them.