Most sourcing reviews start and end with the price on the PO. That’s the number everyone negotiates, reports and defends in the meeting.

In twenty-plus years of sourcing, I’ve found it is rarely where the margin goes. It goes earlier, in decisions nobody revisits, and later, in costs that land on someone else’s budget line.

This week covers the four places it leaks, and a read on what Bangladesh, Pakistan, Turkey and China shipped.

01 · Origin Watch: Q3 and Year-to-Date 2026

Public figures only, with sources. Read them as signals, not forecasts.

Bangladesh

RMG exports for Jul–Sep (Q1 FY27) were US$10.58bn, up 6.10% YoY (EPB). The EU took US$4.90bn (+3.33%), but its share slipped from 47.60% to 46.36%. Shipments to the UK rose 5.94% and to the US 10.40%. Exporters attribute the squeeze to the energy crisis and rising production costs.

Source: The Financial Express (BD), 10 Oct 2026.

Pakistan

Textile exports for Jul–Sep (Q1 FY27) were US$5.026bn, up 5.26% (PBS). Ready-made garments rose 15.69% to US$1.223bn and knitwear 4.14% to US$1.483bn, while cotton cloth fell 5.34%.

Source: Business Recorder, 16 Oct 2026.

Turkey

Apparel exports for Jan–Sep 2026 were US$12.37bn, down 2.6%, and September fell 5.4% to about US$1.41bn. Exports to the EU fell 6.1% to US$7.31bn and to Germany 12.9%. Knitwear was the weakest category.

Source: Tekstil Rehberi, citing İHKİB and TİM data, 16 Oct 2026; İHKİB September report.

China

Apparel exports for Jan–Sep 2026 were US$120.69bn, up 4.8%. September alone rose 24% to US$15.42bn, which the chamber attributes partly to a low base last year.

Source: China Chamber of Commerce for Import & Export of Textiles, via Chinanews, 14 Oct 2026.

My Read (Opinion)

These four origins are not moving together:

  • Bangladesh is growing, but faster into the US than into Europe.
  • Turkey is contracting, and hardest into the EU.
  • Pakistan’s growth is coming through finished garments rather than cloth.
  • China’s September jump is partly a base effect, and shows capacity ready for fast orders.

If your origin mix hasn’t changed while the map has shifted, that alone is worth a conversation. Re-test origin by style and fabric base, not by habit.

02 · The Four Leaks

None of these shows up as a line on the cost sheet. All of them show up in the P&L.

  • Origin: buying out of habit, not landed value
  • Specs: loose tech packs that create rework
  • Timing: a two-week slip that forces markdowns
  • Suppliers: scorecards that don’t move orders

Origin

What it looks like: A country chosen years ago and never re-tested. The decision was right once. The fabric base, duty position and lead time have all moved since.

Where it lands: In landed cost, in lead time and in the flexibility you no longer have when a season turns.

What to check, per style, each season:

  • Duty and tariff exposure into your markets
  • Energy and labour cost trends at origin
  • Where the yarn and fabric actually come from: local or imported
  • The transit window the style really needs

A style that doesn’t need five-day road freight into Europe shouldn’t carry the cost of a nearshore origin. A technically demanding style moved to the cheapest hub can cost more in rework than it saves on FOB.

The question: When did you last re-test origin for your top 10 styles?

Specs

What it looks like: A tech pack that leaves room for interpretation. Small deviations in weight, wash or trims are each "within reason" on their own.

Where it lands: Rework, claims and markdowns. It rarely gets booked as a spec problem. It gets booked as a quality problem, a returns problem or a sell-through problem.

What helps: Complete measurement tables with tolerances for every size, physical wash and colour standards approved before bulk, and the bill of materials locked to supplier references before the first sample.

The question: Which spec deviation repeats most often, and who owns it?

Timing

What it looks like: A two-week slip. On paper, it’s a minor delay. On the shop floor, it turns a full-price drop into a promotional one.

Where it lands: In margin at the till, months after the slip, when nobody connects the two.

What helps: Track the slip where it starts, at fabric intake, not where it ends, at ex-factory. A three-day delay at dyeing is a warning. The same delay found at packing is a markdown.

The question: Do you track delivery against the selling window, or against the PO date?

Supplier Performance

What it looks like: A scorecard filled in after the season, filed and forgotten. Allocation for next season gets decided on relationships and price.

Where it lands: You keep paying for the same problems, season after season, with the same partners.

What helps: Tie allocation to the scorecard, with rules agreed in advance. Suppliers in the top tier on on-time delivery and right-first-time samples earn first call on core styles and the best terms. Those in the bottom tier lose volume until corrective actions are verified. Suppliers change their behaviour quickly once they see volume actually move.

The question: Does last season’s scorecard actually change this season’s allocation?

03 · Where Data Actually Helps

The pattern across all four is the same. The information exists, but it lives in different places: origin data with sourcing, specs with product, the calendar with merchandising, supplier history with quality. By the time anyone puts them side by side, the money has already gone.

The fix isn’t another dashboard. It’s connecting those four views early enough to change a decision: before the PO, not after the markdown.

The real margin of a programme is simple to write down and hard to see:

True margin = net revenue − (landed cost + rework + expedites + markdowns)

Only the first term in brackets is on the cost sheet.

Closing

Margin is not lost in the negotiation. It is lost in the decisions nobody reopens.

Harvics Global Ventures builds trade and operating infrastructure for organisations preparing their supply chains for this transition.

The Trade Intelligence Brief is a series on the operating frameworks reshaping global commerce.