Skip to content

Data Analytics

Understand what is happening, where it comes from, and what needs attention.

Reporting tells you that something moved. Analysis tells you whether the change is real, where it came from, and whether it deserves a response.

Understand where growth is coming from, see which categories need attention, and know where your market is moving, in a view your team can rerun every period.

Discuss a projectSee the case studies

Case studies. Real data, real findings.

Explore the charts, then request the full report. Each one shows an analysis we can build for your market.

Understand competitive shifts

Who is gaining ground in South Africa's vehicle market?

Jetour added +2.25 pp of market share in a year: more than GWM/Haval and Chery combined.

Chinese brands now sell 19.6% of new light vehicles. The biggest seller is rarely the biggest mover, and knowing who is taking share tells you where to focus.

Request the full reportDiscuss a project

SA Informatics analysis of naamsa monthly new-vehicle sales. Explore it on the site

Chinese brands' share of new light-vehicle sales
0%5%10%15%20%25%202420252026

Biggest contributors to the gain

  • Jetour+2.25 pp
  • BYD+0.80 pp
  • Omoda & Jaecoo+0.75 pp

Separate price from volume

Sales are growing. What is happening to volume?

Food specialists' sales rose +4.9% in rands while the volume they sold fell −1.9%.

Rand growth can hide shrinking demand. Separating price from volume shows whether a category is really growing, and whether it needs a volume plan or a pricing plan.

Request the full reportDiscuss a project

SA Informatics analysis of Stats SA retail trade sales. Explore it on the site

Last 12 months: growth in rands against growth in volume
Growth in rands+4.9%
Growth in volume−1.9%
Price effect: the gap between the two+7.0%

Food and drink specialists sold less, for more: rand sales rose +4.9% while volume fell −1.9%. Revenue is being carried by price.

Plan with a tested forecast

South African retail: the outlook for the year ahead

About R1,232.6 bn of retail sales over the next 12 months, peaking at R138.8 bn in December 2026.

December carries more than a tenth of the year, and the margin for error is thin: a 1% shift in sales moves the chance of a weaker year from 50% to 78%. Plans need the range, not just the number.

Request the full reportDiscuss a project

SA Informatics analysis of Stats SA retail trade sales. Explore it on the site

Retail trade sales, R billion at constant 2019 prices
  • Observed (Stats SA)
  • 12-month outlook
  • 95% range
  • 80% range
10012014020232024202520262027Outlook

1 / 3

Four questions that come back every period. Each one answered on real data.

Did performance really change? Where did the change come from? Does anything need a closer look? Where is the market moving? Each tab answers one of them on real South African data, with the decision it informs and what you would receive if the question were yours.

Stats SA retail sales · naamsa vehicle sales

Every figure is computed by script from the published releases, pinned to the release it came from, and reruns when the next one arrives.

Analysis of Stats SA retail trade sales

Comparative analysis

Growth in rands can hide falling volume.

The business problem

Sales reported in rands mix two things: how much was sold, and the price it sold for. When prices rise, a category can report growth while selling less. A comparison is only meaningful when periods, definitions and price effects are treated the same way on both sides.

What the data shows

Over Aug 2025 to Jul 2026, against the 12 months before, South African retail sales rose +4.3% in rands but +2.6% in volume. About 1.7% of the growth was price.

Food specialists sold less, for more: rands +4.9%, volume −1.9%. Furniture and appliances did the opposite: volume grew +9.2%, faster than rands (+4.9%), because prices fell.

Last 12 months' growth: rands against volume
  • Growth in rands (current prices)
  • Growth in volume (constant 2019 prices)
  • Gap = price effect
−4%0%+10%Total retail+4.3% → +2.6%Food and drink specialists+4.9% → −1.9%Pharmacy, health and beauty+6.9% → +3.7%General dealers+4.1% → +1.6%Hardware, paint and glass+3.0% → +1.7%Clothing, footwear and textiles+3.1% → +2.5%All other retailers+5.6% → +7.3%Furniture and appliances+4.9% → +9.2%
Show the data
Growth in rands and in volume by type of retailer
Retailer typeRandsVolumePrice effect
Total retail+4.3%+2.6%+1.7%
All other retailers+5.6%+7.3%−1.6%
General dealers+4.1%+1.6%+2.4%
Textiles, clothing, footwear and leather+3.1%+2.5%+0.5%
Household furniture, appliances and equipment+4.9%+9.2%−4.0%
Pharmaceutical, medical, cosmetics and toiletries+6.9%+3.7%+3.1%
Hardware, paint and glass+3.0%+1.7%+1.3%
Food, beverages and tobacco (specialised)+4.9%−1.9%+7.0%

Aug 2025 to Jul 2026 against Aug 2024 to Jul 2025. SA Informatics analysis of Stats SA retail trade sales.

The decision it informs

Whether growth targets were met in real terms, and where price rather than volume is carrying performance. That tells you which categories need a volume plan rather than a pricing one.

To act on it, you'd also need your own unit and price data, the product mix within each category, and promotion timing.

What you would receive

A comparison of each category, region or channel in rands, in volume and in price effect, every period, built on written definitions so that next month's figures mean the same as this month's.

Why the question recurs

Prices keep moving, so the gap between rand growth and volume growth changes every period. The question comes back at each month-end, budget review and pricing decision.

Questions like this

  • “Is our category growing faster than the market once price is taken out?”
  • “How do our stores compare once size, location and trading days are accounted for?”

Data and method

Definitions and sourceWhat rands, volume and price effect mean here
Rands
Sales at current prices: what was actually rung up, in R million.
Volume
Sales at constant 2019 prices: the same sales valued at 2019 prices, so price changes are removed.
Price effect
The implied price change: (1 + growth in rands) ÷ (1 + growth in volume) − 1.
Source
Stats SA retail trade sales, July 2026 release. Actual (not seasonally adjusted) monthly values, total retail and seven types of retailer.
How the numbers are calculatedTwelve-month totals, not single months

For each series, the sum of the 12 months Aug 2025 to Jul 2026 is divided by the sum of the 12 months Aug 2024 to Jul 2025, minus one. Growth is calculated separately at current and at constant prices; the price effect follows from the two.

Every figure is produced by a script from the pinned release. None is typed by hand.

Why this methodLike-for-like periods, and a price adjustment we didn't choose

Comparing full 12-month totals covers the whole seasonal cycle, so December trading, Easter timing and month length fall on both sides of the comparison. Stats SA's own constant-price series apply a price adjustment for each type of retailer, so the result does not depend on an index we picked.

For a business, the same comparison is built from its own unit and price data.

Analysis of Stats SA retail trade sales

Diagnostic analysis: contribution

The biggest category isn't always the biggest contributor.

The business problem

When a total moves, the first question is where the movement came from. Large categories dominate the conversation, but size alone says little about how much of the change they account for. Splitting the change into contributions points follow-up work at the right place.

What the data shows

Of the +2.6% change in retail volume, general dealers, 44% of sales, contributed +0.72 pp. “All other” retailers, 11% of sales, contributed slightly more: +0.77 pp.

Food specialists took 0.16 pp off the total. The seven contributions add up exactly to the total change.

That is where the follow-up starts: in the categories that moved the total, not the ones that are simply the largest.

Contribution to the +2.6% change in retail volume
All other retailers+0.77 ppGeneral dealers+0.72 ppClothing, footwear and textiles+0.45 ppFurniture and appliances+0.39 ppPharmacy, health and beauty+0.27 ppHardware, paint and glass+0.13 ppFood and drink specialists−0.16 ppTotal change in volume+2.56 pp
Show the data
Contribution of each type of retailer to the change in total volume
Retailer typeContributionShare of salesOwn growth
All other retailers+0.77 pp11.0%+7.3%
General dealers+0.72 pp43.9%+1.6%
Textiles, clothing, footwear and leather+0.45 pp17.8%+2.5%
Household furniture, appliances and equipment+0.39 pp4.5%+9.2%
Pharmaceutical, medical, cosmetics and toiletries+0.27 pp7.3%+3.7%
Hardware, paint and glass+0.13 pp7.4%+1.7%
Food, beverages and tobacco (specialised)−0.16 pp8.3%−1.9%

Percentage points of total change, Aug 2025 to Jul 2026 against Aug 2024 to Jul 2025. SA Informatics analysis of Stats SA retail trade sales.

The decision it informs

Where to direct the follow-up: which categories' plans, ranges or performance deserve a closer review, and which large categories are holding steady.

To act on it, you'd also need store, product and promotion detail within the contributing categories, before drawing any conclusion about causes.

What you would receive

A breakdown of each period's change by category, region or channel that reconciles exactly to the reported total, with a route down to the next level of detail.

Why the question recurs

Every set of results prompts the same question: what moved the total? A breakdown that reconciles to the published number answers it the same way each time.

Questions like this

  • “Which regions account for the fall in volume while revenue held up?”
  • “How much of the change in average selling price came from the product mix?”

Data and method

Definitions and sourceContribution and share of sales
Contribution
A category's change in volume between the two 12-month periods, divided by total retail volume in the earlier period, in percentage points.
Share of sales
The category's share of total retail volume over the last 12 months.
Source
Stats SA retail trade sales, July 2026 release. Constant 2019 prices, actual values.
How the numbers are calculatedAnd how we check they reconcile

Contribution = (category volume, last 12 months − category volume, previous 12 months) ÷ total volume, previous 12 months × 100.

Check: the seven contributions sum to 2.56 pp, equal to the total change of 2.56pp. These constant-price series add up, so there is no residual. Where components don't add up to the total, we show the residual rather than spreading it.

Why this methodWhy growth rates alone mislead

Ranking categories by their own growth rate rewards small categories that grow fast without moving the total. Contribution weights each change by the category's size, and because it reconciles to the total, nothing is left unexplained or counted twice.

Analysis of Stats SA retail trade sales

Investigative analysis

A flag tells you where to look, not what happened.

The business problem

Every reporting cycle has numbers that look unusual. Chasing each one wastes time; ignoring them risks missing a real break. A consistent rule for what counts as unusual, followed by a set of checks, separates the movements worth investigating from routine variation.

What the data shows

A simple rule flagged 7 of 139 months between January 2015 and July 2026, all in 2020 and 2021.

The most extreme, April 2021's +87.0% year-on-year rise, was not a boom. Against April 2019, sales were 2.0% lower: the month it was compared with was the April 2020 lockdown. Since July 2021, no month has crossed the threshold. The closest was July 2022 (+8.8%, a score of 2.96 against a threshold of 3.5).

Year-on-year change in retail sales volume, with pattern breaks flagged
  • Year-on-year change in volume
  • Month flagged as breaking the pattern
-50%0%+50%20192020202120222023202420252026April 2020: −47.6%May 2020: −11.2%June 2020: −6.9%July 2020: −8.3%April 2021: +87.0%May 2021: +15.0%June 2021: +10.6%
Show the data
Months flagged as breaking the pattern
MonthYear-on-year changeAgainst two years earlier
April 2020−47.6%−46.2%
May 2020−11.2%−9.5%
June 2020−6.9%−4.5%
July 2020−8.3%−6.3%
April 2021+87.0%−2.0%
May 2021+15.0%+2.0%
June 2021+10.6%+2.9%

Months flagged where the change departs sharply from the usual pattern. Chart from January 2019. SA Informatics analysis of Stats SA retail trade sales.

The decision it informs

Which movements deserve investigation time this cycle, and which are routine variation that needs no response.

To act on it, you'd also need for each flag, a set of checks: the comparison base (is it a base effect?), calendar and trading days, the category and regional breakdown, and whether the source data was revised.

What you would receive

An exception list each period: the flagged movements, the checks run on each and what they found, so an investigation is recorded rather than repeated.

Why the question recurs

Each new period of data brings movements that look unusual. Applying the same rule every time means attention goes to the same kind of signal, not to the loudest number.

Questions like this

  • “Margins dropped at one dealer group. Is it one product line, one branch, or everywhere?”
  • “Is a category's growth accelerating, or is this month an echo of last year?”

Data and method

Definitions and sourceYear-on-year change, and what counts as unusual
Year-on-year change
Total retail volume in a month against the same month a year earlier, at constant 2019 prices.
Usual change
A median of +2.0% a year, with a typical spread of 2.3 percentage points.
Flag
A month whose change sits more than 3.5 times the typical spread away from the usual.
Source
Stats SA retail trade sales, July 2026 release. Total retail, constant 2019 prices, actual values.
The checks behind the April 2021 findingTesting for a base effect

A year-on-year change compares two months. If the earlier month was itself unusual, the change mostly reflects that month. Comparing with two years earlier removes the 2020 base:

April 2021
+87.0% on a year earlier; −2.0% on two years earlier
May 2021
+15.0% on a year earlier; +2.0% on two years earlier
June 2021
+10.6% on a year earlier; +2.9% on two years earlier

On a two-year basis none of them looks like a boom. April 2021 was still below April 2019. That is a reason to look at recovery, and at Easter timing, not at a surge in demand.

Why this methodWhy median and MAD, and why this threshold

The rule measures “usual” with the median and a typical spread that extreme months can't distort, so the months we want to find can't hide themselves. A threshold of 3.5 times that spread is a widely used convention.

It is a transparent rule rather than a trained model. That suits the question, which is which months to examine, and anyone can recompute it.

Worked example: South Africa's new-vehicle market

Market and competitor performance

The biggest seller is rarely the biggest mover.

The business problem

Where is market share moving, and which competitors account for the change? In any competitive market, from vehicles to groceries to personal care, a shift in share changes the plan: pricing, range, stock and where to focus sales effort.

A competitor's growth rate alone can mislead. What matters is how much of the market's movement each one accounts for. Here is that analysis on South Africa's new-vehicle market, where Chinese brands now sell one in five new light vehicles.

What the data shows

In the 12 months to September 2026, Chinese brands held 19.6% of new passenger and light commercial vehicle sales, up from 13.4% a year earlier and 8.7% the year before. Their sales grew +67% while the market grew +14%.

The biggest contributor was not the biggest seller. Jetour added +2.25 pp of share, more than GWM/Haval (+0.49 pp) and Chery (+0.43 pp) combined. Brands that entered naamsa's reports during the year (BYD, Changan, LDV) added +1.20 pp.

Chinese brands' share of new light-vehicle sales, by month
0%5%10%15%20%25%202420252026
Show the data
Chinese brands' monthly share of new light-vehicle sales
MonthShareChinese brandsAll brands
September 202619.4%11,39558,652
August 202619.3%10,58854,943
July 202620.5%11,21454,622
June 202621.8%11,21751,564
May 202622.8%10,99448,122
April 202622.9%10,40445,380
March 202619.5%10,73354,927
February 202618.7%9,50650,794
January 202619.1%9,19848,186
December 202518.5%8,68146,875
November 202517.6%9,16852,206
October 202516.1%8,50652,971
September 202515.8%8,16751,681
August 202516.1%7,95149,240
July 202514.7%7,12148,604
June 202515.1%6,75044,699
May 202514.7%6,28442,679
April 202514.6%5,85840,062
March 202512.8%5,97246,775
February 202512.1%5,50945,559
January 202511.9%5,30244,431
December 202411.5%4,46538,911
November 202410.7%4,91745,928
October 202410.3%4,63945,019
September 202410.2%4,19341,132
August 20249.7%3,95240,731
July 20249.5%3,95741,488
June 20248.5%3,18437,480
May 20248.8%3,06034,701
April 202410.6%3,75935,618
March 20248.1%3,37641,447
February 20248.3%3,50942,163
January 20248.4%3,31439,661
December 20237.9%2,96937,787
November 20237.8%3,15540,531
October 20236.6%2,80742,273
September 20236.7%2,85942,838
August 20238.0%3,40342,603
July 20238.5%3,46240,505
June 20237.4%3,23743,740
May 20237.9%3,17840,226

SA Informatics analysis of naamsa new-vehicle sales.

Who drove the +6.22 pp gain in share
Jetour+2.25 ppBYD+0.80 ppOmoda & Jaecoo+0.75 ppMG+0.75 ppGWM/Haval+0.49 ppChery+0.43 ppFoton+0.37 ppChangan+0.28 ppJAC+0.11 ppLDV+0.11 ppBAIC−0.13 ppAll Chinese brands+6.22 pp
Show the data
Units and share by Chinese brand group, last two 12-month periods
Brand groupOct 2024 to Sep 2025Oct 2025 to Sep 2026Share change
GWM/Haval24,73931,200+0.49 pp
Chery24,04330,034+0.43 pp
Jetour5,61120,353+2.25 pp
Omoda & Jaecoo10,56216,657+0.75 pp
MG1,0105,779+0.75 pp
BYD04,970+0.80 pp
Foton2,2194,806+0.37 pp
JAC2,2973,322+0.11 pp
BAIC2,4532,022−0.13 pp
Changan01,767+0.28 pp
LDV0694+0.11 pp
FAW10+0.00 pp

Percentage points of the market, Oct 2025 to Sep 2026 against Oct 2024 to Sep 2025.

The decision it informs

Where competitive pressure is building and from whom, so plans, pricing and sales effort go where the market is actually moving.

To act on it, you'd also need share data at the level you compete on: your segments, channels or regions, from your own sales or an industry source.

What you would receive

A consistent view of your market share and your competitors', rebuilt from each new release of the data, with every movement broken down by who drove it.

Why the question recurs

Market data arrives every month or quarter, and the competitive picture keeps moving. Each release raises the same question: who gained, who lost, and does it change our plan?

Request the full market reportDiscuss your own market

Questions like this

  • “Which competitors are taking share in our categories, and how fast?”
  • “Is our share loss a market shift, or a gap in our own range?”

Data and method

Definitions and sourceWhat is counted, and which brands are Chinese
Measure
Local (domestic) sales of new passenger cars and light commercial vehicles, as reported to naamsa.
Chinese brands
GWM/Haval, Chery, Jetour, Omoda & Jaecoo, MG, BYD, Foton, JAC, BAIC, Changan, LDV, FAW
Source
naamsa monthly flash reports, January 2021 to September 2026, Total Vehicle Sales by Manufacturer.
How the numbers are calculatedTwelve-month windows and share points

Share = units sold by Chinese brands ÷ units sold by all brands, over each 12-month window ending in September 2026. A brand group's contribution is its own change in share between the last two windows.

Check: the contributions sum to 6.22 pp, equal to the total change of 6.22 pp.

Why this methodWhy 12-month windows, and why share points

Monthly shares jump with model launches and fleet deals; 12-month windows show the shift underneath. Measuring each brand's change in share points, rather than its own growth rate, shows who actually moved the market: a small brand doubling its sales can matter more than a large one growing slowly.

What matters for the decision.

Most organisations measure more than they can act on. We help decide which indicators deserve leadership attention, define each one precisely, set baselines, and show where exceptions come from, so the monthly pack supports a decision rather than describing one.

A definition is only useful if the next person computes the same number. Here is the one behind the examples above.

Indicator specification

Real retail sales growth

Question it answers
Are we selling more, or just charging more?
Definition
Retail sales at constant prices over the last 12 months, against the 12 months before
The full specificationCalculation, source, refresh, revisions and what to watch for
Calculation
Sum of the latest 12 months at constant 2019 prices ÷ sum of the previous 12 months − 1
Source
Stats SA retail trade sales: total retail at constant 2019 prices
Refresh
Monthly, on each Stats SA release
Revisions
Pinned to the release used (July 2026); recent months are preliminary, so each release is a new run
Read with
The same measure at current prices; the gap is the price effect
Watch for
Base effects (April 2021 against the 2020 lockdown) and the January 2027 rebase

The foundation underneath. Most analytical failures start before any chart is drawn.

Sources that don't reconcile, definitions that drift, a spreadsheet nobody can rebuild. We do the engineering the analysis needs as part of the work, not as a separate project.

  1. Sources

    Systems, spreadsheets, documents and public data, gathered with their origin recorded.

  2. Validate

    Inputs checked for gaps, duplicates and breaks before anything is calculated.

  3. Define

    Measures written down precisely, so this month means the same as last month.

  4. Analyse

    Comparison and diagnosis, with the evidence for each finding kept alongside it.

  5. Decision view

    Delivered into your reporting environment or a purpose-built view.

The analysis and its evidence

Findings stated plainly, with the data and method behind each one.

Definitions you can keep using

Indicator specifications, so next month's numbers mean the same thing.

A view built for the decision

In your existing reporting environment or a purpose-built view.

A way to rerun it

Where the question recurs, a repeatable process or system, not a one-off report.

When understanding isn't enough

Sometimes you need an estimate of what comes next.

That is a different route to the decision, not a more advanced one.

Data Science
How an engagement runs
The Johannesburg skyline at golden hour

Tell us the question you need to keep answering.

Describe the question, the data you have, and whether you'll need the answer again. We'll tell you whether we're the right fit.

Start a conversation