Second-Sourcing a SKU: When a Backup Supplier Pays for Itself

Most small stores are single-sourced by accident rather than by decision. Someone set the supplier up in year one, the ordering habit formed around it, and nobody has priced the alternative since — which means every price increase arrives as a fact rather than a negotiation, and every stock-out is somebody else's schedule. Second-sourcing isn't about leaving a supplier. It's about knowing what the same part costs from two places, so the number in the price file becomes a claim you can check.

Which lines actually deserve it

Qualifying a second supplier costs real hours, so it belongs on a short list, not the catalog. Four filters, applied in order:

• Spend concentration. Sort your purchases by annual dollars and start at the top. On most catalogs, 30 to 60 lines carry the majority of spend, and those are the only ones where a percentage point is worth a phone call.
• Cost drift. Lines whose cost has climbed fastest over twelve months, weighted by what you buy — the formula is in the supplier scorecard. A line that has moved 14% in a year is telling you the current price is unexamined.
• Margin criticality. Lines already near your margin floor, where the next increase forces a price change you'd rather not make.
• Continuity risk. Lines that would embarrass you if they went out for six weeks — the ones customers come for, the ones that carry other items into the basket, the ones with long lead times or a single manufacturing origin.

A line that clears all four is worth a second source even at parity pricing. A line that clears only the first is worth a quote, nothing more.

The hard part is the common key

You cannot compare two suppliers until you can prove they're quoting the same thing, and part numbers rarely agree. Supplier A sells it under the manufacturer's number, supplier B under its own house code with a hyphen in a different place, and your Shopify SKU is a third string invented years ago. Matching on price and description is how a store ends up buying a similar-looking item with a different spec.

Ranked by how much you can trust them:

1. Manufacturer part number plus brand. The strongest key when both suppliers carry the same brand. Brand matters as much as the number — the same numeric string belongs to different parts across manufacturers, which is why matching on the number alone quietly produces wrong pairs.
2. UPC / GTIN / EAN. Unambiguous when present, and often present in neither file.
3. Published interchange or cross-reference. Standard in auto parts, common in industrial fasteners and electrical. Useful, but it's an assertion by whoever published it, not a fact.
4. Spec-for-spec equivalence. Dimensions, ratings, material, certification. This is engineering work, and it's the only route for private-label or generic goods.

Comparing offers that aren't comparable

Quotes arrive in different shapes: different pack sizes, different freight thresholds, different multipliers, different currencies. Reduce everything to delivered cost per unit at your real order frequency, which is not the same as the unit price in the file.

One part, monthly demand of 48 units, current retail $24.99:

Supplier B has the lowest list price and the highest delivered cost. Fifty units is two packs at $680, which sits under the $1,000 freight threshold, so $38 gets added: $718 ÷ 50 = $14.36. Supplier C looks worse than B on the price list — 30 cents a unit dearer — but five packs of 12 comes to $834 and clears its $750 threshold, so freight is free at $13.90 — plus a small penalty for buying 60 when you need 48. Twelve extra units means roughly six units of extra average inventory, $83 of capital, about $17 a year at a 20% carrying rate, which spread over 576 units a year is 3 cents. Call it $13.93 delivered.

1.8 margin points between the best and worst option, and the ranking inverts against the price list. Before signing anything, add the rest of the delivered stack — duty, brokerage, currency, inbound handling — as set out in landed cost, and check whether the quote is a net price or a multiplier off a list that can move independently, which is standard practice in industrial price books.

What splitting volume costs you

The part most stores miss: moving volume away from an incumbent can raise the price of everything you still buy from them. Volume tiers, annual rebates and freight thresholds all reset at a boundary, and crossing it backwards is expensive.

Say you buy $2,400 a month from the incumbent, and $2,000 a month is the break that earns an extra 3% off everything. The second source quotes 5% under on the lines you'd move.

A 5% saving that costs money. Now size the split against the boundary instead of against a round percentage — move 15%, or $360 a month, and the incumbent keeps $2,040, still above the break:

saving = $360 × 5%  =  $18.00/month  =  $216.00/year
tier    = intact ($2,040 > $2,000)

Half the volume moved, and the result swings by $389 a year. The general rule: find every threshold in the incumbent's agreement — tier breaks, annual rebate bands, freight-free minimums — and treat the largest one as a hard floor on what stays. Then split against that, not against a target percentage someone picked in a meeting.

Qualifying the second source

A quote isn't a supplier. Five checks before any real volume moves:

1. A real test order. Small, paid, on normal terms, with no announcement that it's a test. You're measuring what a routine order looks like: days to ship, condition on arrival, whether the invoice matches the quote, whether the packing list matches the invoice.

2. The part itself. Physically compare against the incumbent's. Same dimensions, same fitment, same certification marks, same packaging quality. On anything customers install or that carries a compliance claim, this is the check that prevents the return wave.

3. The price file. Ask for the standard file they send to dealers, before you commit. A supplier who sends a clean CSV with stable part numbers and an effective date is worth real money against one who sends a PDF — the arithmetic for that is in the supplier scorecard, and what to look for once it arrives is in reading a supplier price list.

4. The written terms. Payment terms, freight rules, minimum order, return and stock-rotation allowance, notice period for price increases. Get the notice period in writing if you get nothing else.

5. Price stability history. Ask how many price changes they issued in the last two years and how much notice went with them. The answer, or the refusal to answer, tells you what the quoted price is worth in twelve months.

Running two sources without doubling the work

The failure mode is not choosing a bad second supplier — it's letting the two drift into an untracked mess where nobody knows which supplier a given unit's cost came from. Three habits prevent that:

• Write down the allocation and the trigger. "70/30, revisit if either delivered cost moves more than 3%, or if fill rate drops below 92% for two consecutive months." A policy with a trigger gets reviewed; a habit doesn't.
• Diff both price files against the same catalog. The whole benefit of a second source is that two quotes on the same part are visible side by side. That only works if both files land in the same process on the same key.
• Know which cost is in Shopify. The cost field holds one number per variant with no history and no supplier attached, so with two sources it's ambiguous by construction. Pick a rule and stick to it — replacement cost from the supplier you'd reorder from today is the most defensible, for the reasons in weighted average, last cost or FIFO — and keep the per-supplier detail outside that field.

The value that doesn't show up in the price

Some of the return on a second source is insurance, and it's worth putting a number on rather than waving at. Take the line above: 48 units a month, $10.79 of gross profit per unit at the incumbent's cost. Suppose the incumbent goes out of stock for six weeks once every two years, and 60% of the demand in that window is genuinely lost rather than delayed.

six weeks of demand  ≈ 66 units
lost                 = 66 × 60%      ≈ 40 units
lost gross profit    = 40 × $10.79   ≈ $432 per event
expected annual cost = $432 ÷ 2      ≈ $216

On one line. And that's before the part nobody quantifies: an incumbent who knows you have an alternative behaves differently when the next price file is drafted. That effect is real, it's unmeasurable, and it's the reason second-sourcing pays back on lines where the second supplier never ships a single order.

The reverse case deserves saying too. If a line fails the filters — small spend, stable cost, healthy margin, easily substituted for the customer — leave it single-sourced. Consolidated volume earns tier discounts, simplifies receiving, and keeps one relationship strong. Second-sourcing everything is as thoughtless as second-sourcing nothing, just more expensive to maintain. And if a line performs badly no matter who supplies it, the question isn't sourcing at all — it's whether to keep the line.