# Methodology — South Africa E-Commerce Import Trends v1.0 (2026) ## 1. Data source All figures derive from the **South African Revenue Service (SARS) Trade Statistics** detailed import data, downloaded from https://tools.sars.gov.za/tradestatsportal/data_download.aspx: - Trade type: **Imports** only. - Granularity: tariff line (8-digit HS code) × country of origin × month. - Measure: **customs value in South African rand (ZAR)** — the value declared for customs purposes. Nominal values; no inflation adjustment. - Period: January 2021 – December 2025 (full years), plus available 2026 months (partial, flagged, and excluded from all growth and index calculations). - Chapters downloaded: 33, 61, 62, 64, 84, 85, 95 (84 is needed solely for heading 8471; the basket then filters to the exact scope below). UN Comtrade (DESA/UNSD) is **not used in v1.0**; an international-comparison chart is planned for a future update. ## 2. The e-commerce basket — exact definition Five consumer-goods categories most associated with online retail: | Category | HS scope (exact) | |---|---| | Apparel & clothing | **Chapters 61 + 62**, all tariff lines | | Consumer electronics | **Headings 8471, 8517, 8518, 8519, 8521, 8528 only** (see below) | | Beauty, cosmetics & fragrance | **Headings 3303, 3304, 3305, 3306, 3307 only** (see below) | | Footwear | **Chapter 64**, all tariff lines | | Toys, games & sporting goods | **Chapter 95**, all tariff lines | A tariff line belongs to a category if its HS code starts with one of the listed chapter (2-digit) or heading (4-digit) prefixes. The basket total is the sum of the five categories; no line is counted twice (the scopes do not overlap). ### Beauty, cosmetics & fragrance — headings included and why not all of chapter 33 | Heading | Covers | |---|---| | 3303 | Perfumes and toilet waters | | 3304 | Make-up and skincare preparations (incl. sunscreen, manicure/pedicure) | | 3305 | Haircare preparations | | 3306 | Oral and dental hygiene preparations | | 3307 | Shaving preparations, deodorants, bath preparations | HS Chapter 33 bundles consumer cosmetics with **beverage syrup concentrates** — SARS itself labels the chapter "Cosmetics, Toiletries and Beverage Syrup". The excluded headings are 3301 (essential oils — largely industrial inputs) and 3302 (odoriferous mixtures and flavour concentrates for the food and beverage industry). Heading 3302 alone accounted for roughly half of chapter 33's import value in 2025 (≈R11bn), dominated by beverage-flavour concentrate — including large volumes from a regional concentrate-manufacturing hub — none of which is a consumer beauty product. Restricting the category to consumer headings 3303–3307 removes that distortion; like the electronics definition, this makes the category **conservative**. ### Consumer electronics — headings included | Heading | Covers | |---|---| | 8471 | Computers, laptops, tablets and other automatic data-processing machines | | 8517 | Smartphones and other telephones; network/communication apparatus | | 8518 | Headphones, earphones, microphones, loudspeakers, audio amplifiers | | 8519 | Sound recording / reproducing apparatus | | 8521 | Video recording / reproducing apparatus | | 8528 | Monitors, projectors and television receivers | Note: **8471 sits in HS Chapter 84**, not 85 — the category therefore spans Chapters 84 + 85. (The common shorthand "electronics = Chapter 85" would miss every computer and laptop.) ### Consumer electronics — deliberately excluded To keep the category defensible, headings with heavily mixed consumer/industrial or B2B content are excluded, notably: 8504 (power supplies/chargers — largely industrial transformers), 8523 (storage media — includes commercial software and smart-card volumes), 8525 (cameras — includes broadcast equipment), 8507 (batteries — dominated by automotive/industrial cells) and 8544 (cables). This makes the electronics figure **conservative**: it understates rather than overstates consumer electronics. ## 3. Calculations All computations are sums and ratios over SARS rows — nothing is estimated, imputed or modelled. - **Annual category value** = sum of customs value across the category's tariff lines, all origins, all 12 months. - **Year-on-year growth %** = (value_year ÷ value_previous_year − 1) × 100. - **CAGR 2021–2025** = ((value_2025 ÷ value_2021)^(1/4) − 1) × 100. This is the compound annual rate across the five-year window 2021–2025, i.e. four compounding steps. - **SA E-Commerce Import Basket Index** = (basket value_year ÷ basket value_2021) × 100, so 2021 = 100. Full years only. - **Top source countries** = origins ranked by customs value within a category for the latest full year (2025). Non-country origin codes (Unclassified, Unknown, Ship/Aircraft, "Gold, Petroleum and Other") are excluded from the ranking but retained in category totals so totals reconcile with SARS. - **Monthly seasonality** = each month's share of that year's category value, averaged across 2021–2025. An even spread would be 100/12 = 8.33% per month; the peak month is the one with the highest average share. ## 4. Limitations — read before citing 1. **This is not a measure of online sales.** Customs data records goods crossing the border, not the sales channel. Wholesale, retail and B2C shipments in these categories are indistinguishable. The basket is a transparent proxy: "consumer-goods import categories associated with e-commerce". 2. **Customs value ≠ retail value.** Figures are declared customs values, excluding duties, VAT, freight mark-ups and retail margins. 3. **Low-value consignment coverage.** Small parcels under de minimis-style simplified clearance may be recorded differently from full declarations, so direct-to-consumer parcel flows can be under-captured. 4. **Country of origin ≠ country of purchase.** Goods bought on an international platform may be recorded against the manufacturing origin. 5. **Nominal rand values.** Growth reflects both volume and price/exchange-rate effects; no deflator is applied. 6. **2026 is partial** and excluded from growth rates and the index. 7. **Tariff structure changes** (reclassified lines) can shift small amounts between headings across years. ## 5. Reproducibility The full pipeline (validation, processing, report build) is deterministic Python (pandas) run over the SARS download; the scripts define the basket in one shared module so definitions cannot drift. Contact info@jlog.co.za for the pipeline. --- JetLog (Pty) Ltd, trading as JLog · Unit 12C, 10 Railway Street, Woodstock, Cape Town · info@jlog.co.za · +27 21 300 6099. Analysis layer licensed CC BY 4.0; underlying figures © SARS.