SBLN Traffic Data Analysis, Independent Report

Three Council datasets, three different answers, none supporting "high numbers".

After media coverage on 6 to 7 May, Bristol City Council published a tranche of additional traffic material on its SBLN page. Read carefully against its own documents, none of it supports the claim that vehicles cut through Southville in high numbers. What the Council itself paid for says the opposite.

10 to 13% The Council's own consultant (Arup, Nov 2024) put Southville through-traffic at 10 to 13%, marked as an upper-end estimate, with the verdict that "through-traffic does not appear to make up a large proportion of all traffic". The analysis was withheld from public-forum responses for 17 months.

53% Floow's hourly figures sum to just 53% of what the Council's own pneumatic-tube counts physically recorded on the same streets in the same hours (86 matched sites, Weekday PM peak). On the busiest road, Floow reports 3% of the cars the tubes counted.

45.5% Of 1,077 physical road segments where Floow reports values for ≥2 of its 5 time windows, 45.5% swing by ≥30 percentage points across the day. The metric is unstable across time of day on the same road.

0.07% Independent ANPR cross-check (Part One): only 10 to 11 vehicles per day were detected at both Camera 14 and Camera 16, the two cameras nearest the Southville Zone. That is 0.07 to 0.14% of traffic at those cameras.

14 Seventeen documented inconsistencies between what the Council says in public and what the Council's own data actually contains. Every claim sourced. The full list is below.

Seventeen documented inconsistencies

Each row pairs a Council claim or data point with what the Council's own documents actually say, and the source file you can check.

The Council says…

The Council's own data says…

Source you can check

"High numbers of vehicles from outside the area use residential streets in Southville as a cut-through to other destinations."

Arup's own number for Sector 5 (Southville) is 10 to 13%, and Arup themselves call it an "upper-end estimate" that "does not appear to make up a large proportion of all traffic".

Arup Through Traffic Analysis.pdf · pp. 1, 3
bristol.gov.uk/.../10865-arup-through-traffic-analysis

Two consultancies were paid to measure "through traffic". The same word, used in public communications, must mean the same thing.

Arup defined through as "trips not made by residents or visitors". Floow defined it as "trip endpoints both outside the local zone polygon". They are not the same metric. The Council never tells readers which one a given percentage is measuring.

Arup Through Traffic Analysis.pdf · p. 1 (glossary)
floow_all.geojson · field percent_through

The Floow map shows percentages "by SBLN zone".

The dataset references 12 zone codes; only 10 zone polygons are published. Two zones used to classify trips are not visible to the public.

floow_all.geojson field od_region ∈ {10, 11, 12, 13, 14, 18, 38, 47, 48, 49, 55, 57}
sbln_zones.geojson · 10 features

"The SBLN" is a coherent area divided into "modal-filter zones".

The Council's own documents use three different zone systems: Arup's 8 sectors, Floow's 10 to 12 zones, and the engagement material's 4 modal-filter zones. There is no published cross-walk; a Southville percentage is not the same area as a Sector 5 percentage.

Arup Through Traffic Analysis.pdf · Figure 1
sbln_zones.geojson
BD17291 - SBLN Engagement Report A4_a_v8.pdf

Floow data is shown "by neighbourhood".

The published zone polygons carry no name field, only an integer OBJECTID. A resident cannot tell which neighbourhood their own address belongs to, and therefore cannot tell whether a given through-traffic claim is even about their road.

sbln_zones.geojson · attributes: OBJECTID, Shape__Area, Shape__Length

"The values for each time of day are averaged across the period September 2024 to May 2025."

Only 5 time-windows are present: weekday AM peak, weekday PM peak, weekend AM peak, weekend midday, weekend PM peak. No weekday inter-peak, no weekday evening, no overnight. Any "the data shows X%" claim is a peak/weekend statement, not a 24-hour one.

floow_all.geojson · field day_time · 5 distinct values

"…where there is a sufficient sample size."

Some segments report 100% through-traffic, which under any sample-size filter implies zero local trips on the segment. The "sufficient sample size" threshold is not published. The dataset also contains duplicate rows for the same segment-bearing-period with identical attributes but different Shape__Length, so naïve length-weighted averages will double-count.

floow_all.geojson · 4,866 features ↔ 2,963 unique segment_ids
OBJECTIDs 3 and 4 (same segment_id 9, identical fields, lengths 60.78 m vs 57.12 m)

"December 2025 Traffic Surveys".

The Council page says December 2025; the underlying ArcGIS layer name says Autumn 2025 (RequestName 2025-10-30-SBLN-before-counts-Autumn-2025); the spreadsheet filenames inside the zip say "Nov 25". Three different labels for the same dataset.

SBLN-December-2025-Traffic-Surveys.zip
sbln_traffic_count_extents_autumn2025.geojson

The "December 2025" surveys help support the through-traffic claim.

The dataset has no through-traffic field. It records 7-day average vehicle volume, mean speed, 85th-percentile speed, walking, cycling, and nothing else. By construction it cannot say anything about through-traffic.

sbln_traffic_count_extents_autumn2025.geojson · 7 attribute fields, none of them through-traffic

The "December 2025" surveys help justify the September 2025 statement.

The data was collected in October to November 2025, after the Council had already made the September 2025 claim. Even if it had a through-traffic field, it could not retroactively support a claim made before it existed.

SBLN-December-2025-Traffic-Surveys.zip · spreadsheet filenames tagged "Nov 25"

"We have published all the relevant evidence." (Implied in repeated public-forum responses through 2025 and early 2026.)

The Arup November 2024 analysis, the most directly relevant document, addressing the exact question, was withheld from public-forum responses for 17 months and surfaced only in late April 2026, after media coverage forced the issue. Separately, the Floow telematics map covering data through May 2025 was not published as an ArcGIS Online item until 27 April 2026 — almost a year after the data-collection window closed.

Arup Through Traffic Analysis.pdf · Nov 2024, first cited late April 2026 · Floow ArcGIS item 701b8df988cf4480b9d80522f3f81a01 created 27 April 2026

"High numbers" justify physical street closures across Southville.

The single highest absolute through-traffic count Arup identifies anywhere in the SBLN study area is 229 vehicles per AM peak on Palmyra Road westbound, about 1.3 vehicles per minute, in a different sector (Sector 4, Bedminster Central). Not "high numbers", and not Southville.

Arup Through Traffic Analysis.pdf · Table 1, p. 2

Floow's per-segment %through is a stable property of the road.

It isn't. Of 1,077 physical road segments where Floow reports values for ≥2 of its 5 time windows, 45.5% swing by ≥30 percentage points in percent_through between any two periods, and 15.4% swing by ≥50 points. Median swing across the day on the same road is 28 points. Examples: one segment goes 22.9% (WD AM) → 85% (WD PM); another 81% (WD AM) → 19.8% (WD PM). The same physical road, the same direction, swings of 60+ percentage points across the day. A measurement that fluctuates this much cannot be a property of "through traffic"; it is sample noise plus algorithm artefacts at the polygon boundary.

floow_all.geojson · 1,077 multi-period segments matched on (lat, lon, bearing). Reproducible from code/dataset_deep_dive.py.

Floow's estimatedCarsPerHour is the actual vehicle count on each segment.

It is not, it's an estimate from a sample of vehicles whose sample rate is unpublished. Cross-checked against the Council's own pneumatic-tube ATC counts on 86 matched streets in PM peak: Floow's hourly figures sum to 53% of what the tubes physically counted. Site-level ratios range from 0.03 (Floow undercounts BCC-583 by 33×, on a road carrying 12,509 vehicles/day) to 3.4 to 3.6 (Floow overcounts BCC-538, BCC-683, BCC-609 on quiet residential streets by 3 to 4×). The disagreement is structural, not noise: Floow systematically undercounts on busy main roads (sample-rate effect) and overcounts on quiet residential streets (extrapolation artefact).

floow_all.geojson vs sbln_traffic_count_extents_autumn2025.geojson. Cross-check in code/floow_vs_atc_crosscheck.py; results CSV at code/floow_vs_atc_crosscheck.csv.

"The average daytime speed along this section is 16 mph. This means a 20 mph limit should improve safety without adding to journey times." (20 mph consultation, Coronation Road and Clift House Road entry, and near-identical wording on Redcatch Road, Parson Street, Bedminster Down Road and Winterstoke Road.)

The 16 mph figure does not reproduce on any Council ATC counter we checked. On every consultation road with a direct counter, the actual mean daytime speed (07:00 to 19:00) is between 3.8 and 7.8 mph higher than the quoted figure. The 85th-percentile speed, one of the two commonly used measures listed in DfT Circular 01/2013 paragraph 36, is between 24.0 and 29.2 mph on every road; the Council have published neither the mean nor the 85th-percentile in DfT-aligned form. Between 38% and 85% of daytime vehicles on these roads are currently above 20 mph and would be slowed by the cap, contradicting the "no effect on journey times" claim.

2024 ATC speed-summary spreadsheets in data/SBLN 2024 Traffic Survey Results/ATC surveys/. Cross-check in code/all_consultation_roads.py; results in sheet Consultation_compare of the Excel database.

The 20 mph consultation states an "average daytime speed" for approximately 100 roads across Bristol, in the 16 to under-18 mph range.

No consultation page checked cites an ATC site number, survey date or methodology document for any road. For four roads that overlap our SBLN data set (Clift House Road, Bishopsworth Road, Bedminster Down, Bedminster Down Road) we can also confirm there is no matching counter in the published SBLN dataset. For the remaining roads the absence of source citation makes the figures unverifiable in either direction. The Council are making numerical claims about ~100 roads without publishing the supporting data for any of them.

Bristol City Council 20 mph consultation pages (bristol.gov.uk/20mph and consultation hub id=281), all areas checked. No matching counter found for the four SBLN-overlap roads in data/SBLN 2024 Traffic Survey Results/ATC surveys/.

Arup's 10 to 13% Sector 5 (Southville) through-traffic figure remains a current published Council figure.

Arup computed that figure against a Sector 5 perimeter along East Street and Bedminster Parade in November 2024, with Dean Lane inside the sector. The September 2025 SBLN consultation map redrew the perimeter along Dean Lane, putting Dean Lane on the boundary. Trips on Dean Lane that crossed the November 2024 perimeter and counted as through-traffic cease to be through-traffic under the new geometry. Arup's 10 to 13% has not been recomputed and is still cited. JTC Site 6 (Dean Lane south end) recorded 2,486 vehicles in the AM peak alone on 11 June 2024, approximately 3.78 times Arup's entire Sector 5 through-traffic count for the same period.

Arup Through Traffic Analysis.pdf · Figure 1 (sector map); September 2025 SBLN consultation map. JTC Site 6 spreadsheet in data/SBLN 2024 Traffic Survey Results/JTC surveys/; calculation in code/dean_lane_calculation.py.

The Floow interactive map is the Council's central evidence for the "high numbers" cut-through claim.

The Floow ArcGIS Online item (701b8df988cf4480b9d80522f3f81a01) was not created until 27 April 2026. The item title is TheFloow_though_traffic_SBLN_Sep24_May25, so the underlying data collection ended in May 2025. The Council held the map for approximately 11 months before publishing it, and only released it in the same week as the Arup analysis, after media coverage.

ArcGIS item JSON at arcgis.com/sharing/rest/content/items/701b8df988cf4480b9d80522f3f81a01?f=json, created timestamp 1777290791000 (= 2026-04-27 14:33 UTC).

Same pattern · Parallel consultation closing 17 May 2026

The Council is doing it again, on a different consultation, with the same data.

In parallel with the SBLN scheme, Bristol City Council is consulting on 20 mph speed limits for around 100 roads, closing 17 May 2026. For each road, the consultation page asserts an "average daytime speed" (16, 17 or 18 mph) and uses it to claim "a 20 mph limit should improve safety without adding to journey times." The reasoning is statistically false. The numbers themselves do not reproduce on the Council's own ATC counters.

"The average daytime speed along this section is 16mph. This means a 20mph limit should improve safety without adding to journey times." Bristol City Council 20 mph consultation, Coronation Road / Clift House Road entry

Average means the centre of a distribution. A 20 mph cap slows every vehicle currently above 20. Their journey times rise. Nobody's falls. We checked every consultation road that has a direct counter in the SBLN ATC dataset, including the slowest, most queue-affected, already-20-mph site (Site 7, Bedminster Parade), which is the Council's strongest possible defence. Even there, the figure does not match.

Road and ATC site

Council quote

Day mean

24h mean

85th %ile

Bedminster Parade, Site 7already 20 mph, queuing zone, the Council's best case

(implied 16 mph)

18.79

19.32

23.4

Coronation Rd, Site 8near roundabout, currently 30 mph

16 mph

20.30

21.11

26.0

Coronation Rd, Site 9long free-flow stretch, currently 30 mph

16 mph

22.23

23.40

28.4

Redcatch Rd, Site 5currently 30 mph

16 mph

24.07

24.47

29.1

Parson Street, Site 3currently 30 mph

below 18 mph

23.06

23.76

28.6

Winterstoke Rd, Site 12currently 30 mph

around 17 mph

24.50

25.51

29.9

Every consultation road with a published Council counter shows that the Council's own 7-day daytime mean is 2.8 to 8.1 mph higher than the figure quoted in the consultation. The day-mean and 85th-percentile columns above are read directly from row 67 column 21 and row 70 column 22 of each ATC site's published Speed Summary Two-Way sheet, so the comparison is between two figures the Council itself has published. The 7-day 24-hour 85th-percentile speed (one of the two commonly used measures of actual traffic speed listed in DfT Circular 01/2013 paragraph 36) is between 23.4 and 29.9 mph on every road. DfT Circular 01/2013 paragraph 100 says a 20 mph signing-only case is supported where the mean is at or below 24 mph. The Council's own means are at or near that threshold, so the policy direction is supportable; the 16 mph figure on the consultation page is what does not reproduce on the Council's own data.

Between 38% and 85% of daytime vehicles on these roads are currently above 20 mph and would be slowed by a 20 mph cap. The "no effect on journey times" claim is mathematically false on the Council's own data.

Roads in the consultation with no published counter: Clift House Road, Bishopsworth Road, Bedminster Down, Bedminster Down Road. The Council are making numerical claims about these without publishing the supporting data.

Full analysis in §8 of the consolidated Report v1.4, reproducible from code/all_consultation_roads.py. Download Report v1.4 (PDF) →

Key Finding (Part One, April 2026)

The June 2024 survey data does not contain evidence sufficient to demonstrate that vehicles are using Southville's residential streets as a cut-through between Coronation Road and North Street. The ANPR camera network had no cameras placed within the Southville zone. The two nearest cameras recorded only 10-11 vehicles per day appearing at both locations, representing approximately 0.07-0.14% of the traffic at those cameras.

The Numbers, Part One

ANPR cameras deployed across South Bristol. None within the Southville Zone.

15,092

vehicles detected at Camera 14 (Coronation Road) on 11 June 2024.

vehicles detected at both Camera 14 and Camera 16. That's the entire cross-zone count.

90%

of Camera 14 traffic went to Camera 1 (Clift House Road) — along the corridor, not through Southville.

45%

of Camera 16 traffic also went to Clift House Road — along the perimeter of the Southville Zone.

94%

of Luckwell Road traffic can be explained by local residential demand, schools, deliveries, and visitors.

Visual Evidence, Part One

$Pie charts showing cross-zone detections as fraction of total traffic$

Cross-zone detections as a fraction of total camera traffic. The red slices are so small they are barely visible.

Where Camera 14 traffic actually goes. 90% travelled to Clift House Road along the corridor.

Where Camera 16 traffic actually goes. 45% went to Clift House Road along the perimeter.

The cross-zone flow (6 vehicles) compared to the dominant corridor flows (3,600-5,600 vehicles).

What every term means

ANPR camera: Automatic Number Plate Recognition camera. A fixed camera that records the number plate, time, and direction of every vehicle passing it. The June 2024 SBLN survey deployed 25 of these around the SBLN study area, but none inside the Southville Zone.

The Floow: A telematics company. Their dataset doesn't see number plates; it sees anonymous GPS traces from vehicles whose owners use a connected app or insurance device.

Telematics: Vehicle GPS data, sampled at intervals. Knows where a journey started and ended (within GPS error of a few metres) and which roads were used.

"Through traffic" (Arup version): Trips not made by residents or visitors. Verbatim from the Arup glossary, page 1. This is the common-sense reading of "cut-through".

"Through traffic" (Floow version): Trips whose start AND end addresses are outside a small SBLN sub-zone polygon. Has nothing to do with whether the car drove the whole length of any road.

SBLN zone polygon: A small area drawn on a map. The Council uses 10 such polygons in the published zones layer; the Floow data refers to 12 zone codes, so two polygons exist that the public cannot see. The polygons carry no name field, only ID numbers.

Segment: One row of the Floow data, a short directional piece of road, typically 6 to 60 metres long. A normal residential road has many segments; the percentage is reported on each one separately.

percent_through: The Floow number that's been reported as "85%" or "81%". Share of vehicles on a segment whose journey endpoints both lie outside the local SBLN zone polygon.

percent_internal: Trips that started AND ended inside the local SBLN zone polygon.

percent_inbound: Trips arriving from outside the zone, visitors, deliveries, residents returning home.

percent_outbound: Trips leaving the zone for elsewhere.

estimatedCarsPerHour: Total vehicles per hour through the segment, regardless of category. The absolute number that "85%" is taken of. Often surprisingly small, typically 30 to 60 on residential streets.