If you sell expensive products across thousands of SKUs, one Shopping campaign will always spend your budget on the same handful of items. Here's the structure we use to fix that — and how to build it in an afternoon.
Put 14,000 products into a single campaign and you've handed Google one instruction: spend this budget wherever conversions come easiest. That is almost never where your profit is.
What happens in practice is a concentration spiral. Your proven best-sellers already have click history, so they win the internal auction, take the impressions, generate the conversions and get rewarded with more impressions. Meanwhile 90% of your catalogue never gets enough data to prove it can sell — so it looks like dead weight, and gets treated like dead weight forever.
| Symptom | What's actually happening |
|---|---|
| A handful of SKUs eat the budget | Google is optimising for conversion volume, not margin. Your cheapest, easiest-selling items win by default. |
| Long-tail products get no impressions | They never accumulate enough click data to compete inside the same campaign. |
| ROAS looks fine but profit doesn't | A $200 accessory at 5x and a $4,000 fire table at 5x are not the same business outcome. |
| You can't bid differently by price | One campaign, one target. A single tROAS across a 20x price range is a blunt instrument. |
| Seasonal items never ramp in time | By the time they'd earn impressions organically, the season is over. |
Two campaigns over the same product feed, separated by campaign priority and shared negatives. Tier 2 catches everything first at low bids; anything that proves itself gets promoted into Tier 1, where you pay properly for it.
Priority is inverted on purpose: Tier 2 is HIGH priority, Tier 1 is LOW priority.
That inversion is the trick most people get backwards. In Shopping, a higher-priority campaign gets the first look at every query. So Tier 2 — the cheap, catch-all discovery campaign — sees the traffic first at a low bid. You then use shared negatives to push the queries you actually want into Tier 1, where the bids are aggressive and the target is set to your real margin.
Priority: LOW
Products: SKUs with ≥ 15 clicks & ≥ 1 conversion in 60 days
Bidding: tROAS at your true break-even + margin
Budget: ~70–80% of Shopping spend
Priority: HIGH
Products: everything else in the catalogue
Bidding: Manual CPC or low tROAS
Budget: ~20–30%, capped
You can't split a catalogue you can't segment. Before you build a single campaign, get four custom labels into your feed — they're what turn "14,000 products" into groups you can actually bid on.
| Label | Use it for | Example values |
|---|---|---|
| custom_label_0 | Performance tier — this is what drives the Tier 1 / Tier 2 split | proven testing dead |
| custom_label_1 | Margin band, so bids follow profit rather than revenue | high_margin mid_margin low_margin |
| custom_label_2 | Price band — essential when AOV ranges from $80 to $8,000 | under_500 500_2000 2000_plus |
| custom_label_3 | Seasonality or stock depth, so you can ramp ahead of demand | spring evergreen clearance |
Set custom_label_0 from a rule, not by hand. Whatever generates your feed — a Shopify app, a Google Sheet, a feed tool — should recalculate it weekly against the promotion rule in section 4. If you're maintaining labels manually across a big catalogue, the structure will quietly rot within a quarter.
custom_label_0 with a weekly automated refresh. Everything else in this guide is built on top of it. Our Feed & Custom Label Cheat Sheet covers the title and label rules in more depth.A tiered structure is only as good as the rule that moves products between tiers. Write it down, automate it, and let it run — the point is to remove your opinion from the decision.
dead and exclude after two consecutive demotions, or when stock depth drops below your reorder point.dead gets one fresh run at the start of its season before it's excluded for good.Roughly two hours of work if your feed is already healthy. Do it in this order — the negatives step is the one that breaks the structure if you skip it.
custom_label_0 = proven, campaign priority Low, tROAS set to your break-even plus target margin.Tier 2 = High priority, low bid, capped budget
Negatives on Tier 2 route queries into Tier 1
Promotion rule refreshed weekly, automatically
Tier 1 set to High priority (Tier 2 never sees traffic)
Same negative list applied to both campaigns
Manually moving SKUs when you "feel" they're ready
This structure is not a quick win — it's a compounding one. The first fortnight usually looks flat or slightly worse, because you've deliberately redirected budget toward products with no history.
| Timeframe | What you should see |
|---|---|
| Days 1–14 | Both campaigns in learning. Blended ROAS dips slightly. Don't touch anything. |
| Days 15–30 | Tier 2 starts producing first conversions on long-tail SKUs. Impression share spreads across the catalogue. |
| Days 30–60 | First promotion cycle lands. Tier 1 ROAS climbs as it inherits proven products with real data behind them. |
| Days 60–120 | The number to watch: percentage of SKUs generating sales. On the accounts we run, this is where it typically moves the most. |
We ran exactly this build on a 14,000-SKU outdoor retailer — the numbers are in our case studies.
We do this for a small roster of high-ticket e-commerce brands. Book a free audit and we'll tell you whether a 2-tier structure is actually right for your catalogue — or whether something simpler would do the job.
Apply To Work With UsFree, no obligation. The plan is yours to keep either way.