PPC for e-commerce companies that still waste query spend

How to Stop Wasting PPC Budget in E-Commerce (Shopping, Search & PMax)

Your best SKU can sit at $1.84 CPC in Shopping while the same product in branded Search costs $4.10 and still wins on contribution margin. That split is the whole job of PPC for e-commerce companies: you are not buying “traffic,” you are buying a mix of product feed quality, query intent, creative, and a product page that does not leak the click. Google Ads still averages about 4.2x ROAS across retail verticals (WordStream 2025/2026 Ads Benchmark Report). Meta sits closer to 2.8x. Neither number is a promise for your catalog. Both numbers collapse if you send paid clicks to a slow homepage, if 20-30% of spend hits junk queries, or if smart bidding is starved of conversion signal.

Search CPC averaged $2.96 in 2026, up 12% year over year. Mobile already takes 63% or more of paid search clicks, and those sessions convert 30-40% worse than desktop when the page is heavy. Only 55.9% of CrUX origins pass all three Core Web Vitals. If your PDP fails INP on a mid-range phone, you are paying Google and Meta for sessions that never reach add-to-cart. This piece is the operating version of that problem: how e-commerce marketers structure Shopping, Search, Performance Max, and paid social so the catalog, not the dashboard vanity, pays the bill.

Where catalog PPC actually burns money before the first sale

Most waste is not “the algorithm.” It is a feed that still uses manufacturer titles, a Search campaign that still bids on informational queries, and a Performance Max asset group that mixes hero SKUs with clearance junk. You will see it as rising CPC with flat add-to-cart, or as ROAS that looks fine last-click while contribution margin after returns is ugly. High CPC pressure then tempts teams to yank new accounts back to manual bidding. That can feel safer for a week. It usually just hides a feed and landing-page problem under a lower bid cap.

Privacy and ad blockers already strip 20% or more of conversion data. If you only fire a browser pixel on purchase, Enhanced Conversions and the Conversions API never get a clean purchase value, and smart bidding starts optimizing for the people who still allow cookies. Last-click still double-counts the same order across Shopping, branded Search, and remarketing. CM360-style paths from Google Ads have been getting noisier since late 2025. Your CFO is not wrong to distrust the in-platform ROAS. They are wrong if the response is to freeze spend instead of fixing the signal and the page.

Creative fatigue on Meta shows up in CPM before volume dies. UGC that worked on Monday can be exhausted in 5-6 days. Broad targeting plus weak creative is how you buy cheap, low-quality sessions. Broad targeting plus a sharp offer, size/fit proof, and a fast PDP is how 2026 auctions actually work. “Creative is your targeting” is not a slogan. In Meta’s current signal-based delivery, creative is the targeting. If they tell you they will just send Shopping traffic to the homepage, that is not a media plan. It is a leak.

If your Shopping and Search mix is already live but contribution per SKU is a mystery, a focused Google Ads and paid search program is usually cheaper than another month of blended ROAS theater.

Shopping, Search, and Performance Max as one catalog system

Treat Shopping as the product index, Search as the intent filter, and Performance Max as the inventory expander that only earns budget after the first two are honest. PMax for e-commerce accounts often drives the bulk now. It is also how you hide branded cannibalization and low-margin SKUs if you never split asset groups by margin band, availability, and seasonality. Optimized PMax accounts often land in a 4x-8x ROAS band. Unoptimized ones look busy and still lose money on shipping and returns.

Keep branded Search separate so you can see what you would have captured anyway. Use exact and phrase around commercial modifiers: buy, price, size, colorway, in stock, near me for local pickup. Negate research language that never buys: how to, DIY, vs, review roundup, free, PDF. AI Max style campaigns can run 12-18% lower CPCs than standard Search when the conversion setup is clean. They will happily spend into irrelevant queries if your negatives and account-level brand exclusions are lazy. Google still needs keywords as guardrails even while delivery leans on signals.

On Meta, stop stacking 2020 lookalikes as if they were a moat. Run broader, feed first-party purchasers and high-value add-to-carts as signals, and rotate short-form video that stops the scroll without wrecking the feed experience. Advantage+ is not a strategy by itself. It is a delivery mode that punishes generic lifestyle cuts and rewards SKU-level proof: unboxing, fit on a real body, 3-second price hook, returns policy in the first frame if that is the objection. TikTok can beat Meta on cold traffic in some catalogs in 2026. Do not add it until the PDP and purchase event quality on the channels you already pay for are trustworthy.

Feed hygiene is media. Titles should read like a search query a shopper would type, not like an internal SKU code. GTINs, color, size, material, and availability have to be accurate or Shopping will show the wrong variant and you will pay for the bounce. Custom labels for margin, bestseller, and clearance let you bid and exclude without rebuilding the whole account every season. If 20-30% of SEM budget is still leaking to irrelevant queries, neglected negatives, or mismatched landings, that is not “the market.” That is unfinished account work.

A working PPC checklist for e-commerce marketers

Use this as an audit you can run on the live Google Ads for e-commerce account, not as a theory deck. The order matters: signal and page first, then structure, then creative volume. Constant structure changes reset learning. Daily bid tinkering on a healthy PMax campaign is how you teach the system nothing.

The PDP is the conversion event, not a courtesy click

PPC for e-commerce companies dies in the 8 seconds after the ad. Message match is not a slogan on a slide. If the RSA promised “free returns on trail shoes” and the PDP opens on a lifestyle hero with no shipping line, you trained the auction that your clicks bounce. One apparel brand we modeled moved add-to-cart rate from 33% to 43% after putting size proof and reviews above the fold. That is a CRO lift, not a bid trick. Another pattern: 70% PPC conversion-rate lifts show up when the landing repeats the exact SKU, price, and CTA from the ad instead of dumping people on a collection of 240 items.

Use HeyLead Insights when the in-platform CPA looks healthy and the order book does not. Session recordings and heatmaps show the size chart that never opens, the sticky ATC that covers the price, the review widget that loads after the shopper has already hit back. You do not need a lecture on “user experience.” You need evidence of where paid intent leaked. Pair that with title tags and H1s that match the query so organic and paid are not fighting each other on the same SKU.

AI Overviews now reach on the order of 2 billion monthly users. Plenty of category searches never click. That does not mean you abandon Search. It means branded and SKU-level queries, Shopping, and remarketing carry more of the close, while your content has to earn citation rather than empty volume. Per Originality.ai’s 2024 analysis, though methodology is debated, 86.5% of top-ranking pages already contain AI-generated copy. Thin generated PDPs will not save a paid program. Specific fit notes, return data, and real photos will. Zero-click is a SERP fact. It is not an excuse to send paid traffic to a 4MB homepage.

A mid-market home-goods operator in Austin had Search converting at 2.1% while Shopping sat at 4.8% on the same SKUs. The Search ads pointed at a blog “buying guide.” They pointed Search at the matching PDP, added a 48-hour ship badge, and left bids alone. Search conversion moved to 3.7% in 11 days. One URL change. That is the mechanism. Not a new platform.

What marketing leaders are seeing

A pattern we see repeatedly in DTC home goods accounts: teams keep raising Shopping bids because blended ROAS looks like 4x, then finance shows contribution after returns on two custom-label groups was under 1.2x. The feed labels were the fix, not another campaign type.

A pattern we see repeatedly in apparel e-commerce: Meta CPMs climb for four days while purchases stay flat. Killing the UGC cut that peaked on day five, instead of waiting for CPA to blow up, is the move. Creative is the targeting. Treating it like decoration is how spend leaks.

FAQs

Should e-commerce companies still use keywords if Performance Max does the volume?

Yes. PMax needs clean product signals and brand controls. Search keywords and negatives still stop you from buying research queries and protect the terms that actually attach to SKUs. Signals over keywords is a delivery shift, not a reason to delete the intent layer.

Is automated bidding just going to inflate spend?

It will if purchase value is wrong, if you optimize to add-to-cart instead of qualified purchase, or if the landing cannot convert mobile. With Enhanced Conversions, accurate values, and stable structure, it usually beats panic-manual bidding on mature catalogs. New accounts with thin data are the exception, not the rule for forever-manual.

How do we know ad fatigue is real and not just a bad week?

Watch CPM and thumb-stop rate before CPA. If frequency and CPM rise while CTR falls on a specific asset, that cut is dying. Rotate hooks on a cadence of days, not quarters. Diagnostics like Event Match Quality help signal quality. Event Match Quality (found in Meta’s Events Manager) scores how well your pixel or CAPI events match Meta user profiles - low scores mean smart bidding is flying blind. They do not, by themselves, predict profitable ROAS.

What ROAS should we put in a board deck?

Use contribution after discounts, shipping, and expected returns, by product group, not a single last-click number. Google’s 4.2x and Meta’s 2.8x averages are directional, not your target. No serious partner should guarantee a CPL or ROAS before they have seen the account, the feed, and unit economics. Three to six months is a fair rebuild window.

Do we need another ad network before we fix Google and Meta?

Usually no. Tighter positioning, clearer offers, and decent attribution beat adding TikTok or Reddit while the PDP is slow and the purchase event is leaky. Add a channel when the current ones have clean signal and a creative engine, not because a pitch deck said diversification.

Putting it to work

Start by exporting 14 days of Shopping item IDs, Search terms, and Meta asset performance, then mark every row as profitable SKU, junk query, or dying creative. That single pass usually shows the 20-30% you can stop buying this week.

This week

• Turn on or repair Enhanced Conversions and CAPI with order-id deduping.

• Rewrite titles for the top 20 SKUs and point every non-brand ad at the matching PDP.

• Add a negative list for research and wholesale language.

• Run the Core Web Vitals checker on three money PDPs and fix INP on add-to-cart.

• Build UTMs so Shopping, Search, and Meta cannot hide inside a blended channel.

• Kill one Meta asset whose CPM is up while CTR is down, and ship two new hooks.

Next 30 days

• Split PMax asset groups by margin and availability.

• Import contribution or shipping-adjusted values into bidding.

• Stand up a weekly search-term and creative review, not a monthly PDF.

• Reconcile in-platform revenue to the order system and drop last-click as the board metric.

• Only then test a new cold channel if the PDP and signal quality hold.

If the gap is the loop between feed quality, query waste, and PDPs that drop mobile add-to-cart, HeyLead can own that catalog PPC work so you are not restitching Shopping, Search, and conversion signal every week. Chat with us on WhatsApp