Should a beauty account build ad groups around ingredients or around products?

Around ingredients, in most cases. Several SKUs usually share one hero active, so a per-SKU structure splits the same intent across more ad groups than the traffic can support. Grouping by active pools the clicks, makes the copy specific to what was typed, and keeps the number of ad groups proportional to how people actually search. Confirm it with your own search terms report first: if your queries are dominated by product and brand names rather than actives, the reverse structure may fit better.

Do own-label cosmetics need a GTIN in Merchant Center?

Products from manufacturers that assign identifiers normally have one and should carry it. If your own-label product genuinely has no GTIN, you submit the brand and MPN combination and set the identifier-exists flag to reflect reality rather than to clear a warning. Identifier requirements differ by category and change over time, so check the current Merchant Center documentation for your product category rather than copying another store's feed setup.

Is a brand campaign worth it if my own page already ranks first organically?

It depends on whether anyone else is in that auction. Run auction insights on a brand-only campaign to see who appears on your name, and use the paid and organic report alongside Search Console to see whether total clicks on brand queries changed when the campaign started. If paid clicks rose while the total held flat and nobody else is bidding, you are mostly moving free traffic into paid. If a reseller or marketplace listing sits above you, the defence is usually worth the cost.

What negative keywords should a cosmetics account add first?

Four families cover most of the early waste: informational suffixes attached to ingredient names, queries naming a marketplace or pharmacy where you are not the seller, dupe and equivalent-product phrasing, and DIY or homemade recipe wording. Add them before the ingredient ad groups go live, then review the search terms report monthly, because new phrasing appears as soon as an ingredient trends.

Google Ads for Beauty Brands: Ingredient Queries

Google Ads for beauty brands breaks the usual retail pattern in one specific way: a large share of the demand never arrives as a product name. It arrives as an ingredient. Someone who has decided they need niacinamide types that word long before they know which brand they want, and the account that owns that moment gets the sale. Three things follow from it, and they are what this guide covers: building a keyword layer around actives rather than SKUs, the Merchant Center attributes cosmetics feeds get wrong, and running a brand campaign that defends your name without simply repurchasing clicks your organic listing already had.

Why google ads for beauty brands starts with the ingredient

Test the premise on your own data before restructuring anything. Open the search terms report for the last 90 days and count how many of your top queries name an active rather than a product. If your range is actives-led, vitamin C, retinol, niacinamide, hyaluronic acid, salicylic acid, that count will tell you plainly where the decision is being made and how much of your account is aimed somewhere else.

Two details specific to the Turkish market change the keyword list. First, transliteration. Shoppers type both spellings, and niacinamide against niasinamid, or hyaluronic against hyaluronik, are not close enough for a match type to bridge reliably. Both belong in the account as explicit terms. Second, concentration modifiers carry commercial intent that the bare ingredient does not: someone searching a specific percentage of niacinamide has already read about it and is now comparing bottles.

Ad groups built on actives, not SKUs

The instinct is one ad group per product. Resist it. Say you sell 12 SKUs built on 5 hero actives. One ad group per SKU gives you 12 groups competing over overlapping intent, each holding a thin slice of the data. One ad group per active gives you 5, each carrying both spellings, the concentration variants and the category phrasing that maps to the same shelf. The 5-group version becomes readable sooner because the same clicks are pooled rather than spread across more than twice as many groups.

Copy then has a job it can do. Name the active first, the concentration and format second, and let the third line answer what the ingredient shopper is actually weighing, which is usually skin type suitability or whether it can be layered with what they already use. Which of those angles pulls is a targeting and creative question rather than a keyword one, and that layer is covered on the AI targeting page.

The negatives to add before the first click

Ingredient queries drag in traffic that will never buy. Add these families before you switch the groups on rather than after the first invoice: informational suffixes such as nedir, ne ise yarar, nasil kullanilir and yan etkileri; other-channel shopping such as eczane or a marketplace name where you are not the seller; dupe hunting such as muadil, dupe and benzeri; and DIY phrasing such as evde or tarifi. Revisit the list monthly.

Merchant Center attributes that decide whether a shade shows

Shopping and Performance Max both read the same feed, and cosmetics feeds fail in ways general retail feeds do not. Four attributes carry most of the weight.

Item group ID with colour. A foundation in thirty shades is one product with thirty variants, not thirty products. Grouping them stops the range competing against itself and keeps the chosen shade attached to the click.
Size and size system for volume. A 30 ml and a 50 ml serum at different prices are routinely submitted with identical titles, which makes the price shown in the listing look inexplicable to a shopper.
Product identifiers. Stocked third-party brands normally carry a GTIN. Own-label products that genuinely have none need the brand and MPN combination, with the identifier-exists flag set honestly rather than to whatever clears the error.
Category against product type. Google's own taxonomy is not the same field as your internal product type. The taxonomy governs how Google classifies you; the product type is yours, and it is what you will segment campaigns by later.

Title order is the cosmetics-specific call. Because the active is what people type, a title running brand, product, active, volume matches more real queries than the brand-first pattern most feed templates default to. The general mechanics of feed quality are a separate subject and product feed ads covers that side. Check the current required and recommended attribute list in Google's official Merchant Center documentation before committing to a rebuild, because the specification moves.

Where cosmetics copy meets policy

The line that gets beauty accounts disapproved sits between a cosmetic claim and a medical one. Reducing the look of blemishes and treating acne are different sentences to a policy reviewer, and only one of them stays in an unregulated category. Before-and-after imagery is handled differently again. Read the current healthcare and medicines policy and the personalised advertising policy in Google Ads Help rather than reasoning from what a competitor's live ad appears to get away with. An ad running is not evidence that its claim is permitted.

Brand defence without paying for clicks you already had

Beauty brands routinely find other advertisers standing on their own name. Authorised resellers, marketplace sellers, affiliates and competitors on broad match all end up in your brand auction, and the sharpest version is the brand-plus-ingredient query, where a marketplace listing for your own product can sit above your own page. Auction insights on a brand-only campaign shows who is genuinely there; for the wider picture of who is spending against you, competitor ad analysis sets out the method.

Whether brand spend is additive at all is a separate question, and the honest answer comes from a holdout test, not from the brand campaign's ROAS, which always looks excellent because the buyer had already decided. Settle that on its own. What matters structurally is that when you do run brand terms, the spend stays contained and visible.

Keep brand in its own campaign, on exact and phrase, with its own budget. It must never share a budget with prospecting, because it will win that internal competition every time.
Add brand terms as negatives in every non-brand Search campaign and use the brand exclusion setting in Performance Max. Without both, generic budget quietly buys brand clicks and your acquisition reporting flatters itself.
Report brand and non-brand separately, permanently. A blended figure that includes brand tells you nothing about whether acquisition works.

On cannibalisation specifically, the number to watch is total clicks, not paid clicks. Google Ads has a paid and organic report showing queries where you appeared organically, with an ad, or both. Pair it with Search Console and ask whether total clicks on brand queries moved when the brand campaign launched. If paid rose and the total stayed flat, you moved cost from free to paid. That is sometimes the right trade, and when a reseller is sitting above you it usually is, but you should know you are making it.

The margin number that comes before any budget change

Beauty carries enough gross margin to hide a poor account for a while, which is exactly why the break-even figure belongs at the front. Take a serum listed at 900 TL. Product cost, packaging, fulfilment and payment fees total 315 TL, so gross margin is 900 - 315 = 585 TL, which is 585 divided by 900, or 65% of revenue. Break-even ROAS is 1 divided by 0.65, which is 1.54.

Now the ingredient ad group that looks like it is working. At a ROAS of 1.20 it returns 1.20 TL per lira spent, and 1.20 multiplied by 0.65 leaves 0.78 TL of gross margin. You spent 1 TL to keep 0.78 TL. It is producing orders and losing money, and scale does not fix it. Run your own inputs through the break-even ROAS calculator and write the result next to every ad group before you touch budgets.

Where to start if the account already exists

Order matters here, because each step makes the next one cheaper. Pull the search terms first and count the ingredient share, since that decides whether a restructure is worth doing. Fix variant grouping and titles in the feed next, because Shopping and Performance Max both inherit them. Then split brand out and add the negatives and exclusions, so your non-brand numbers become honest. Only then rebuild the ingredient ad groups, and only after that compute the break-even figure and start moving money. Any other sequence means every measurement you take is taken against an account that is still moving underneath it.