Instagram ads for skincare and beauty brands

Partnership ads are Meta's name for influencer and creator ads — it used to call them branded content ads — and they run a creator's own post as your ad, with the partner's account, the advertiser's or both featured in the header. Our skincare Facebook guide covers what they are, what a creator's words commit you to and how to brief one, and that much carries across; the placement and objective lists there were read for Facebook and stay Facebook's. What Instagram adds is a second thing to borrow.

Meta's help centre says that when an advertiser creates partnership ads from a creator's handle, creator audience sharing “allows you to use an audience of people who have engaged with the creator's account on Instagram”, which lets you “share your partnership ads with the creator's engaged audience”. You can also build lookalike audiences from it. There is no Facebook equivalent in anything we have read: on Facebook you borrow the post and the face, and you find your own audience for it.

You build it at the ad set level, in the Audience section, in five steps: Create new beside Custom audiences; Custom audience; Instagram account, under Meta sources; the creator's name from the Source dropdown; Create audience. It is the same dialog that builds an audience from your own Instagram account, where Meta caps the retention window at 365 days and people drop out as the window rolls past their last engagement. Whether a creator's audience is capped the same way is not stated on its page.

Two conditions make or break it. The first is technical and unforgiving: “Custom audiences only work when running ads from the creator's handle on Instagram. The Instagram account selected at the ad level below Identity must match the account used for the custom audience.” An ad running under your own handle cannot use it, which means the ad your customer sees is the creator's post under the creator's name.

The second is a permission the creator grants and can withdraw. Meta's help centre requires account-level partnership ad permissions and notes that “Creators have the ability to remove this access at any time”. Account-level is the broader of the two kinds: content-level permissions cover a named post, while account-level ones let an advertiser create partnership ads and other ad types “from a partner's handle without pre-existing content”. That answers a question most brands ask first — no, they do not have to have posted about you already; the ad can be made from scratch. Requests are accepted in Partnership Ads Hub, the Instagram app or the Facebook app, and where a creator has linked their accounts, accepting once covers both platforms. Nothing we read says what happens to an ad set that is already running when the permission is removed, and the nearest documented case runs the other way — on Facebook, turning an ad code off stops new ads while “any active ads using the code will continue running”. Check the campaign yourself rather than assume either.

Now the collision, which decides the rest of this page. Engagement custom audiences are on Meta's list of audience options withheld from any ad set that can deliver to under-18s, and a creator's shared audience is an engagement audience, so on our reading it stops at 18; Meta's teens page does not mention creator audience sharing by name. Put that beside the mechanism above and the most Instagram-specific targeting available to a beauty brand cannot be used on the age group where Instagram's lead over Facebook is widest. Neither can a lookalike of it. You are not choosing between two settings; you are choosing between two buys.

The optimisation goals narrow in the same direction, and this is the part most accounts discover late. Meta's list of goals unavailable to audiences that include teenagers is “app events, video views, post engagement, value, ad recall lift, landing page views, leads, conversations and event responses”. What remains is impressions, reach, daily unique reach, link clicks, offsite conversions and Instagram profile visits. Read the two lists together and the rule is simple: below 18 you can buy a look, a click or a purchase, and nothing that opens a conversation. Lead forms, click-to-Direct, reminder ads and value optimisation are all 18 and over — which means the formats Instagram built for interaction are barred from the audience Instagram uniquely has.

The examples page has the full lists and their sources, and maps our ten ads against them.

What spans both halves is your profile. Instagram profile visits remain available as a goal when a teen audience is included, which makes them the only way to accumulate anything on that side: not an audience you can target there, but followers who will see what you post, and who are reachable later through an Instagram account custom audience of people who have engaged with you in the last 365 days — in an 18+ ad set. Whether engagement from someone who was 16 at the time shows up in that audience after their eighteenth birthday is not something we found documented.

That profile is also a hole in the age floor you set, and it is specific to this trade. The ad set's minimum age does not follow the tap. If your grid holds the acne gel, the before-and-after and the sunscreen — the three things on the examples page Meta will not let you advertise to a fifteen-year-old — then a teen profile-visit ad delivers them to it anyway. Read your own top rows against the same list you read the ad set against.

Eligibility is decided on the creator's side and it is loose. Meta's Instagram help centre says partnership ads and branded content need an Instagram professional account, or on Facebook a Page or a profile in professional mode, and an “authentic, established presence” with a “sufficient follower base”. No number is published, so treat any follower threshold as unknown and send the request.

You can run these beside your ordinary ads. Meta's guidance is that partnership ads and business-as-usual ads can sit in the same ad set, or in separate ad sets, which Meta says “must be under the same campaign and have budget fluidity” — Advantage+ campaign budget, one pot at campaign level that Meta moves between the ad sets under it. One sharp edge if you select both kinds at once: “you'll only be able to edit the first identity of the BAU and partnership ads at the same time”. And because a shared audience is tied to one handle, a campaign built on two creators is two ad sets with two identities, with the campaign budget deciding between them. Let it decide between creators; never put the two halves of the age line under one campaign budget, because it will move the money to the buy with the better-looking reported numbers, which is structurally the 18+ one.

Two mechanisms from elsewhere in the cluster apply here, and this page does not re-argue them. A restock or a drop with an hour attached can carry a reminder — a bell people tap to be notified before it starts — which Instagram gives away on an ordinary feed post and sells as a reminder ad; our restaurant Instagram guide sets out what each notifies and when the paid one is worth it. Note what the list above does to it: event responses is unavailable below 18, so the free post is the only version of that mechanism the teen half can have. The second jar is bought on a purchase-window audience, which the skincare Facebook guide builds and measures, and which is 18 and over by construction.

One product limit to plan around. best-in-slot publishes into an ad set you have already made, under your own Page and Instagram account, with a link button and a destination URL. It cannot make a partnership ad, so it cannot run from a creator's handle, and it cannot attach a reminder event or a message destination. Those are Ads Manager builds; what the tool is for is the picture and the words that go into them.

The objective belongs to the campaign and does not change between the apps, so the rule from the skincare Facebook guide still decides it: buy by where the purchase happens and what has to happen first. Instagram changes the answer in two places.

The first is the age floor of the ad set underneath, because the objective you pick has to be one whose goal survives there. Offsite conversions — which is how a purchase-optimised Sales ad survives — link clicks, reach, impressions and Instagram profile visits do. Landing page views, leads, conversations, event responses and value do not. So the same offer can be a Sales ad optimised for purchases in either half; a Traffic ad optimised for landing page views at 18+ and for link clicks below it; and no ad at all on the teen side if it needs a form, a message thread or an event.

The second is Traffic, which gains a destination that does not exist on Facebook: your own profile, bought as profile visits. On the teen side it is the only objective that leaves anything on the platform; a Sales ad there still leaves you a customer and an email address.

The rows are sorted by the decision this platform forces first: whether the ad can be shown to a sixteen-year-old. The three at the top are mechanisms rather than products — the creator's audience, your own profile, and the restock, which is the one skincare offer Instagram sells a purpose-built ad for. Then five offers whose claims keep them eligible for a 13+ ad set, which is the only reason to prefer Instagram to Facebook for them. Then three that cannot go there because of what they contain, and one that cannot because of whom it is aimed at.

Each row pins who the offer is for, which angle family argues it (each is a page in our library of ad angles), which objective buys it and, where the examples page has one, the finished ad — two white pillowcases sold with an acne gel, a self-tan before-and-after, a 24-shade line-up. All ten ads from that page appear once, and the notes under each on the examples page say what went wrong with it; here the rows end on what to do. Twelve rows is not a campaign plan: pick the offer your best product earns on each side of the line.

Health is not one of Meta's special ad categories, so nothing here is category-restricted. What binds a skincare ad set on Instagram is the age floor its ads require, and that floor decides which of the levers below exist at all.

Decide it product by product rather than account-wide, and ask what the strictest ad in the ad set needs. A cosmetic can run from 13; an over-the-counter drug, a before-and-after and an ad aimed at a customer list cannot. Those rules are the United States' and Meta's — other countries draw their own lines, and Meta's own minimum is 16 in Australia — so a brand shipping beyond the US should read its markets before it sets a number.

If the answer is no, run the whole account at 18 and over and use everything. Advantage+ audience on for prospecting, which obeys your location, minimum age, language and custom-audience exclusions strictly while treating interests and included audiences as suggestions it may go beyond. The creator's engaged audience and a lookalike of it in their own ad set, under the creator's identity. The purchase-window reorder audience from the Facebook guide. Your customer list as an exclusion on everything meant to find new people. And on any Sales ad set built on a custom audience, “Further limit the reach of your ads”, because Meta otherwise treats an included custom audience as a suggestion and may deliver past it.

If the answer is yes, build it as a separate buy and expect it to be blunt. Advantage+ audience has to be off, because the youngest minimum age it accepts is 18. What is left is age and location. The location has to be city level or larger with the EU, the EEA and Switzerland taken out entirely, since ads there cannot deliver to under-18s. There are no exclusions, so your own customers are in it. There is no language setting. And the optimisation goal must be one Meta keeps available — purchases, link clicks, reach, impressions, profile visits — which rules out a lead form, a message thread, an event and value optimisation.

Then reason about the top of the range as well as the bottom, which is the half of this lever most accounts forget. Age is one of only two things left in that ad set, and it has two ends. Capping it at 17 makes the buy a clean test of the audience you opened it for, and shrinks the customers-you-cannot-exclude problem to the share of your customers who are minors. Raising the floor is worth as much: Pew puts Instagram at three-quarters of 15- to 17-year-olds against 44% of 13- to 14-year-olds, so a floor of 15 or 16 costs little of the reach you came for and drops the band where a skincare ad is hardest to defend.

Whether to open it at all is a separate question from whether you may, and it is a budget question. An ad set has to be fed: Meta says one usually leaves the learning phase after about 50 results in a week, so a teen Sales buy has a weekly floor of roughly 50 × what a purchase costs you. A brand shipping across Europe also has to split its map before it can touch the age lever, which is another ad set with its own 50 results to find. If that arithmetic does not clear, the honest answer is that the audience is real and the buy is not — run the profile-visit line instead and let the grid do the work.

In the 18+ ad set the targeting does the qualifying and the creative does the persuading. In the teen ad set the creative does both. That is why the teen buy wants more ads inside one ad set rather than more ad sets beside it, and why the render is the thing to check hardest before it runs.

The default is the same as on Facebook: the product page, written in the same cosmetic words as the ad, with the size, how long it lasts and the full ingredient list as text, and a Purchase event carrying its value. The FDA counts claims made “in advertising, on the Internet”, so a page that says more than the ad does is the version that gets read back to you.

A partnership ad is the exception worth planning for, and it matters more on Instagram than on Facebook: the ad runs under the creator's handle, so the profile a curious reader taps first is theirs, not yours. If the ad is doing the work of introducing your brand, the destination has to finish the introduction, because the header did not.

For an ad that has to explain first — a schedule, a shade quiz, a habit — send it to a page under your own name that makes the argument, optimised for landing page views at 18+ and link clicks below. A lead form is not an option below 18 at all, so a restock list aimed at younger buyers is a sign-up page bought as Traffic.

Your profile is a destination in its own right, and the only one that leaves anything behind in a teen ad set. Buy the visit when the grid can answer a stranger's questions and survive the age floor the ad set was set to, as the offer row above describes.

Instagram Direct messages are where shade and suitability questions belong, through Engagement or Leads — and at 18 and over, since conversations is withheld from teen audiences. Open it only when someone is answering, and write the replies in advance in cosmetic terms. A reply is your writing as much as a post is, and “will this help my rosacea?” invites a yes that is a drug claim, typed by your brand, in public.

For a subscription, the rules are federal and Meta's, and the skincare Facebook guide sets both out: the price and interval in the ad, the material terms and express consent before the checkout takes a card.

Every brief in our batch named an audience — people who wash after training, people who lose a lip balm a week, people who have bought the wrong shade online. In a 13+ ad set not one of those is expressible as targeting. They become creative direction: the picture and the first line have to find the person the ad set can no longer ask for. That is the biggest difference between writing for this platform and writing for Facebook, and the clay mask shows what it costs. Its brief was a mask, $18, with friends on a Friday night. Its render is three adults on a sofa, and in an ad set with age and location only, the render is the targeting.

So cast the ad: decide who is in the frame, what they are doing and at what time of day, and check the render against that decision. Where nobody was described, our batches keep drawing the same person — both undescribed faces in the Facebook run came back light, and so did the hands in this one.

Casting young is the hard case, and this trade has no easy version of it. A generated teenager is the FTC problem the ankles and the invented handle already are: a person shown using a product has to be a real user. A real one needs consent a parent gives. The clean answers are the product alone, hands, or a real customer photographed with permission — and if none of those is available, the honest conclusion is that the ad is for the 18+ buy.

The claim rules decide the age floor, so read every draft for the two sentences that move it: a treatment claim that turns a cosmetic into a drug, and a comparison shown as a before-and-after. Then read for the sentence that points at the reader. Four of the ten ads told the reader what was happening to their own skin — “your lipid barrier”, “If your face feels tight after showering at the gym”, “your rough skin” — which is the shape Meta's personal-attributes standard is written about, though it does not say whether cracked or breaking-out skin counts as physical health. Treat those lines as a risk. Move the condition into the product description and describe the reader by what they do.

Check where the sentence lives before you plan to fix it. In the cleanser ad the diagnosis is set in type at the top of the poster, in the accent colour, so a caption edit does not touch it. Anything burned into a picture — a claim, a price, a date, a label — is a re-render, and it is worth reading the render for those before the copy is written.

Read the labels. The tool draws packaging with words on it, and across two batches it has printed an ingredient list nobody chose, an internal freight calculation, a treatment name and a jar volume on a spatula. On the sunscreen the opposite problem cost more: the stick came back with no SPF, no broad-spectrum statement and no water-resistance claim at all, in a drug ad. Match every rendered label to your real pack.

For a creator's ad the brief is the Facebook guide's and none of it changes here: what the product does in cosmetic words, what they may say about their own skin, how long they will have used it, and the words that are off limits. Instagram adds two things. Their post is the creative, so it has to be in a format Meta accepts for a partnership ad, and the ad is reviewed before it runs. And because the ad carries their identity, the first 125 characters read as theirs — which is where the FTC's disclosure advice bites hardest: add your own, at the start, whatever the header shows.

Then the shapes and the lengths, which the examples page measured: 4:5 for the feed, 9:16 for Stories and Reels with the top 14%, the bottom 35% and 6% at each side kept clear, and a caption whose first 125 characters have to work as the whole ad. Ten of our fifty captions fit that, all from the two ads whose argument was a single fact. None of the fifty was short enough for an image ad in Reels, so either write a second caption for that placement or leave it out of the ad set.

The age floor splits the account, so the structure has four lines rather than three: the borrowed audience, ordinary prospecting, the teen buy if you want it, and the customers. The first, second and fourth are 18 and over by construction. The third is a separate buy with its own job, because nothing in it can be compared with anything in the others — and if you have no creator yet, the first line simply does not run and the second carries the prospecting until it does.

Start where the Facebook guide starts, because the arithmetic does not change between apps: a new customer is worth the margin on their first order, plus the share who reorder times the margin on a reorder times the reorders a returning customer makes over a period you are willing to wait for. That figure is the ceiling on what you can pay for a first order, and it is worked out in full there.

The teen buy needs a second ceiling, and the reason is structural rather than behavioural. A teenager who buys a lip balm may well buy another; what you lose is the channel for prompting it, because that customer cannot be put in a reorder window, a customer list or a lookalike until they turn 18. The refill ad is the clearest case — an ad that is nothing but a customer list — and on that side of the line it does not exist. So price the buy on the first order's margin plus what your email and your profile recover. If you are already running the reorder holdout the Facebook guide describes, you have a live measurement of the second half of that: the held-out customers are your own margin per customer without paid retargeting.

Read that ceiling against the right number. An ad set that includes under-18s will show your first-order ad to existing customers it cannot exclude, so its reported purchases flatter it. The test is new customers in your store, not purchases in Ads Manager.

Then count the ad sets, because each has to be fed. Meta says an ad set usually leaves its learning phase after about 50 results in the week following its last significant edit, so plan on roughly 50 times your cost per purchase per Sales ad set per week. A creator ad set, a prospecting ad set, a teen ad set and a reorder ad set is four; a second creator makes five; shipping across Europe makes six, since the teen buy's map has to be split. A brand that cannot feed them should run fewer rather than starve them all, and the teen buy is the one to defer, because it is the only line whose number you cannot borrow: you may not import a cost per purchase from an 18+ ad set that differs from it on four axes. Fund it as a fixed discovery budget for a fixed period, decided before it starts.

Creator audience sharing also changes what a creator's fee is buying. You are paying for the post and for reach into an audience that already trusts them, and you are paying it in a form you cannot keep: the audience is theirs, the permission can be withdrawn at any time, and the ad has to run under their handle for the targeting to work. Two consequences. What you keep from the campaign is the customers it produced and whatever they left on your own profile, not the audience. And on our reading, every paid impression you buy under their handle feeds the rolling engagement audience you are renting — which is also why an inactive creator's audience shrinks while you are paying for it.

Then judge the creator on the store rather than on the comparison Meta offers you. As the Facebook guide puts it, you do not choose how much each ad in an ad set is shown, so a creator's ad beside your own is a record of what the system chose to show, not a test. Give them their own ad set, alternate it on and off over equal periods — each long enough to produce a readable number of orders in your store — and look at the second order as well as the first.

The profile is the one line with no reported value attached, and on the teen side that is not a gap but the point: for a buyer you cannot retarget for years, the follow is the retargeting. Ads Manager will tell you what a visit cost. What it is worth is two numbers of your own — followers gained, and what your account's traffic does to store sessions and first orders in the weeks after. Agree in advance how long you will fund it before that second number has to move.

The reorder line is measured with the holdout the Facebook guide builds, and nothing about Instagram changes it — except that no version of it is available on the teen side, where you cannot hold anyone out in the first place.

One last caution on reading the numbers. A partnership ad's likes and comments land on a post under someone else's handle, so that engagement is not yours to count as brand lift; the orders are.

The first four are about the age floor and what people assume it costs. The next three are about the borrowed audience, which is rented rather than owned. The last two are about words that go out unread, in a picture and in a message thread.