← Back to blog

Google Ads for E-commerce — Where to Start and What Results to Expect

The logic of this funnel is not aesthetics, it is economics. Demand Gen and YouTube are expensive and have a long payback cycle — a store with a 2,000 PLN monthly budget cannot finance them while testing sales at the same time. Shopping and branded Search hit a customer with clear purchase intent, deliver conversions from the first week and feed data to the entire system. That is where 100% of your initial budget should go — only when those layers run at ROAS > 3 do you expand up the funnel.

What this article will NOT do for you

It will not tell you whether your store is ready for Google Ads. That depends on you — on margin, product range, traffic and your marketing experience. But it gives you a concrete list of questions you should know the answer to before spending the first PLN. From my experience, in roughly 7 out of 10 accounts that come to me with "Google Ads doesn't work", the problem is not in the campaign itself — it is in the foundations underneath. The campaign is the last 30%.

Step 1 — what to prepare before you launch the Google Ads account

This is the section people skip most often. And without it, you start blindly and then spend two months wondering why nothing is working.

Margin per product / category

You have to know how much you earn on a single product after deducting the cost of goods, shipping, packaging, payment platform fees and operations (usually 5–10% of revenue). Without this you do not know whether ROAS 3.0 is profit or loss.

• Margin 60–70% (cosmetics, supplements, niche dropshipping) — break-even ROAS around 1.5–1.7. Comfortable: 2.5–3.5.
• Margin 40–50% (mid-segment apparel, accessories) — break-even ROAS around 2.0–2.5. Comfortable: 4–5.
• Margin 20–30% (electronics, equipment, larger appliances) — break-even ROAS around 3.3–5. Comfortable: 6+.
• Margin 10–15% (cheap fast fashion, deep discounts) — break-even ROAS around 6.7–10. Google Ads is brutally tough here — one mistake and you eat the entire monthly profit.

Conversions with transaction value in GA4 + Google Ads

A standard "purchase" conversion only counts the number of purchases. That is not enough — the system has to know how much you earned on each one in order to optimise for ROAS instead of for volume. Check Google Analytics 4: the purchase event with the value and currency parameters. Then import it into Google Ads as a value-based conversion. If the "Conv. value" column is empty in Ads — that is the first thing to fix.

Merchant Center and product feed

Without this there is no Shopping or PMax with your catalogue. The most common feed problems I see in new accounts:

• Missing GTINs — products without a barcode get lower visibility and are sometimes disapproved.
• Titles like "Dress 2548" instead of "Women's summer dress — Lemon Sorbet, black, size M" — the first will not win on any query, the second has a chance.
• Out-of-sync stock levels — the system advertises a product you do not have. Click, "out of stock", the competitor wins.
• Missing attributes (size, color, material, brand) — in fashion this is practically a requirement.

If you run Shoper, WooCommerce, Shopify or PrestaShop — there is a plugin for that. If your store is custom-built — you need to generate the XML/CSV feed manually or via an API.

Website — speed and trust

If your landing page loads longer than 3 seconds on mobile, half of your customers will bounce before they see anything (Google: 53% of mobile users abandon a site that takes longer than 3 seconds to load, source below). Quality Score also pays the price — more expensive clicks. Check on PageSpeed Insights. Targets: LCP < 2.5 s, CLS < 0.1.

The second topic is trust: reviews (Trustpilot, Google, product ratings), a return policy (essential — in the EU, 14 days minimum), company details in the footer, clear shipping costs before the cart. Missing even one of these elements lowers conversion rate by 10–30% (my own observation from A/B tests across different clients).

Budget — what makes sense

A practical minimum for a Polish online store is 1,500–3,000 PLN per month on media spend alone, that is what actually goes to Google. You can launch with less, but two problems show up:

• Smart Bidding needs data — a rule of thumb is 30 value-based conversions in 30 days. In practice I see tROAS stabilise from around 20 conversions per month when the conversion value data is clean. The threshold depends on the CPA in your category: at a CPA of 15–30 PLN (cosmetics, supplements) 1,500 PLN per month is enough, but at a CPA of 60–100 PLN (outdoor gear, niche B2B) you need 2,500–3,000 PLN to clear that bar.
• Auctions are brutal — at 20 PLN per day in a moderately competitive niche you get 10–20 clicks per day. That is too little to optimise anything.

If you have less than 1,500 PLN per month — consider starting with Search without Shopping (smaller reach but more precise traffic from keywords) or shift the money to content/SEO and Meta Ads, where the entry threshold is lower.

Source: page-speed data — Google web.dev — Why does speed matter? (53% mobile abandonment when LCP > 3 s). Practical minimum budget — my own observation from accounts I run plus recommendations from StoreGrowers — Google Ads Cost 2026.

When Google Ads does NOT make sense

Because it does not always make sense. Three situations where you have to be brave enough to say "don't launch":

• Margin 10–15% with AOV under 50 PLN. A killer combination — to break even you need ROAS 7–10. In practice impossible to balance without raising the margin or AOV (cross-sell, free shipping threshold, product bundles).
• No demand in search. Products that people do not actively search for — a brand-new category, heavy technical B2B, creative services delivered to individual briefs. Search will not work because there are no queries to bid on. PMax will burn budget trying to find "your" audience on Display and YouTube, where there is no purchase intent.
• Shop with fewer than 30 products or an unapproved feed. Shopping with 3–5 products is wasted budget — the system has nothing to rotate, click-through drops, the campaign dies. At that stage focus on content marketing, SEO and Meta Ads, where the entry threshold is lower. Come back to Google Ads when the catalogue grows.

This is not a "sales pitch on no" — these are real cases in which I run a reality check with the client and tell them that Google Ads is not their channel. Better to say it now than after three months of burned budget.

Step 2 — the campaign structure I recommend for a starter

Not everything at once. A store that is just kicking off with Google Ads should run a maximum of 3 campaigns in the first 30 days:

Campaign	Why	% of budget	When
Branded Search	Captures customers who already know your brand. Cheapest traffic, highest conversion rate, saves attribution.	10–20%	From day 1
Standard Shopping or PMax (feed only)	Main sales engine. Shows products from the feed on purchase queries.	50–70%	From day 1
Search non-brand (precise)	Queries like "<product> <attribute>", phrase / exact match, carefully.	15–25%	From day 1, but with a small budget
Dynamic remarketing	Recovers abandoned carts. Requires traffic (min. 100 people / 30 days on the list).	5–10%	After 4–6 weeks
Demand Gen / YouTube	Brand awareness building — expensive, long cycle.	0%	Only when ROAS on the above is stably > 3

Structure based on my own observations from 30+ e-commerce launches. The philosophy: start at the bottom of the funnel (purchase decision), move up only when the basics pay off. If your budget is below 2,000 PLN per month — launch Shopping/PMax first and add non-brand Search in week 2 (splitting 1,500 PLN across 3 campaigns leaves none of them with enough data).

Standard Shopping or PMax — what to pick at the start?

The most common question from new clients. Short answer: at the start — Standard Shopping. Reasons:

• Transparency. You see which products are advertising, which queries generate clicks, where the budget is going. PMax hides half of that data.
• Control. You can exclude weak products, add negatives, adjust bids. PMax does this for you — not always the way you want.
• Learning the system. Standard Shopping has a simpler optimisation scope (one ad type, one channel) — you see the effects faster and can interpret them better. PMax combines several channels at once, so the learning period is longer.

I usually turn PMax on after 30 days of Standard Shopping — once I have conversion data, audience lists, product exclusions and clarity on what works. PMax delivers on average 10–20% higher ROAS than Standard Shopping, but that number applies to mature accounts with conversion data. On a new account PMax often starts worse than Standard Shopping for the first 4–8 weeks — until it gets the full context (audience signals, asset groups, value-based conversions). That is why I turn it on after Standard Shopping, not instead of it. Phase 3 in the chart above (Month 4+) describes a mature account — PMax with asset groups itself usually kicks in from Month 2.

PMax advantage data — Search Engine Journal — Performance Max for ecommerce (10–20% ROAS lift vs Standard Shopping with a hybrid strategy) plus my own observations from accounts.

Step 3 — what to expect in the first months

The most common trap: expecting ROAS 6× in the first week. That will not happen — the Google Ads system has a learning period (usually 7–14 days from launch or from a strategy change). Conclusions drawn before that are noise, not signal.

What goes wrong most often in the first month

1. Strategy changes every 3 days. The client sees high CPC, panics, halts campaigns. The algorithm starts from zero. Give it 14 days with one defined strategy, then judge.
2. No conversion value. The system optimises for the number of purchases, not for revenue. It chases cheap products instead of large orders. Fix: value parameter in GA4 → import as a value-based conversion.
3. Maximize Conversions instead of tROAS. The default strategy in new campaigns is often "Maximize Conversions" without a ROAS target. That means "spend the entire budget" — regardless of return. After 30 conversions, switch to Target ROAS.
4. No remarketing exclusions. A customer just purchased and 2 weeks later sees an ad for the same product. Annoying and expensive. Add a "purchased in the last 14 days" list and exclude it from acquisition campaigns.
5. An unprepared shop. Everything is set up perfectly, but the site loads in 8 seconds. ROAS 0.8. You fix the site, ROAS 4.1. Often it is not the campaign — it is the website.

What I see after 11 years — observations by category

These ranges come from real accounts I have run or am running, anonymised. They are not "market averages" — they are concrete intervals:

• Natural cosmetics — Shopping CPC 0.40–1.20 PLN, target ROAS 4–6, branded campaign ROAS 12–18× (returning customers). A 60–70% margin allows for flexibility.
• Diet catering (subscription e-commerce — meal-box plans ordered online like in a shop, not lead-gen) — Search CPC 2.50–5.00 PLN, CPA 30–80 PLN on the first order, ROAS 2.5–4. We also watch LTV, not just first order. Seasonality: January + September = costs up by 30–50%.
• Fashion (mid-segment store) — Shopping CPC 0.60–1.80 PLN, ROAS 3–5 (after factoring in returns of 25–35% — because the nominal ROAS in the report is higher than the real one). Returns have to be subtracted, otherwise the picture is misleading.
• Supplements — Shopping CPC 1.00–2.50 PLN, ROAS 3–5, LTV is the key (the customer buys every month). The first transaction is often at ROAS 1.5–2, the second and third make it profitable.
• Sports / outdoor gear — Shopping CPC 0.80–2.00 PLN, ROAS 2.5–4 (lower margins), strong seasonality (summer for bikes / camping, winter for skiing).
• B2B tools / products — CPC 3–8 PLN, ROAS 5–12 (high order values), but a low number of transactions = hard for smart bidding to optimise.

Source: JP Ads accounts (anonymised). Supporting data — Lumo — E-commerce benchmarks 2026 (PL), Bluerank case studies, OwlClaw Google Shopping Benchmarks 2026.

If your numbers are off these ranges by more than 50% — something needs checking. The problem is not always in the campaign. Sometimes it is the competition, the season, the assortment, the margin. But it is good to know which way to look.

How to check after 30 days whether things are going well

Do not look at ROAS as the only indicator. Three diagnostic levels, 20 minutes of work:

Level 1 — is there even any data?

• Conversions > 20–30 in the last 30 days (value-based — those feed Smart Bidding)
• Conversion value populated (the "Conv. value" column is not empty)
• Each campaign has at least 7 conversions in the last 30 days (otherwise turn it off or change it)

Level 2 — efficiency

• ROAS within the range for your category (see the table above)
• Quality Score ≥ 6 for the main keywords
• Search impression share > 40% on branded queries
• Search lost IS (budget) < 30% — is the budget covering demand

Level 3 — site and conversion

• Conversion rate > 1% for average AOV stores (50–300 PLN). High-ticket categories (electronics, furniture, B2B gear) typically run at 0.3–0.8% — the signal is a drop below your category average, not the absolute number
• Mobile load time < 3 s (PageSpeed Insights)
• Engagement rate > 50% for paid traffic (content typically runs higher; 55%+ suggests good feed → page relevance)
• Average order value (AOV) in line with expectations

If levels 1 and 2 are fine but level 3 is not — the campaign is doing its job, the problem is on the site. If level 1 is not met — fix that first, only then optimise the campaigns.

About the sources: Polish ROAS and Shopping benchmarks — Lumo, Bluerank, Double Digital. Global — OwlClaw Google Shopping Benchmarks 2026, Search Engine Journal — PMax. The rest is observations from 11 years of running Google Ads campaigns at JP Ads — numbers from my own practice are flagged "from my experience".