What Meta, Google, and TikTok ads cost for beauty brands

More spend can expose a weaker next group of customers, tired creative, a product-page problem, or economics that were only workable at a smaller volume. It can also expose inconsistent measurement. Start by reconciling actual orders, refunds, spend, and first-time customers over the same period. Check tracking changes and attribution windows before diagnosing a campaign from a dashboard alone.

Then separate the symptoms. If impressions cost more while click-through and purchase conversion are steady, investigate delivery conditions and audience mix. If click-through falls, review the message and creative. If clicks hold but purchasing falls, inspect product availability, price, shipping, mobile checkout, and the match between the ad promise and page. These are investigation paths, not proof of causation; change one meaningful variable at a time where practical.

Increase spend only within a loss limit and review period the business can fund. Look for a repeatable offer, credible conversion measurement, sufficient creative and inventory, and acquisition economics that fit the contribution and cash available. There is no universal revenue milestone, ROAS, or budget-increase percentage that makes a beauty brand ready to scale.

Illustrative first-order economics: an order with $60 of net revenue and $24 of product and variable order costs leaves $36 before marketing and overhead. At $30 of media-only acquisition cost, $6 remains before agency fees, production, and overhead; that is not $6 of profit. If additional spend acquires the next customer for $40, the first order loses $4 before those other expenses. Repeat purchases may change the picture, but use observed cohort evidence rather than assumed lifetime value.

Pause an increase when tracking breaks, stock cannot support the promise, the buying experience fails, or the agreed loss limit is reached. Where evidence is inconclusive, reduce the scope of the test rather than describing uncertain performance as a proven winner or loser.

For the auction-based campaigns discussed here, cost depends on the buying method, competition and delivery conditions rather than a fixed beauty rate card. CPM measures cost per 1,000 impressions; CPC measures cost per click. CPA and CPL require an explicit acquisition or lead definition before they can be compared.

A published average or median describes its source’s sample, not your next campaign. Objective, market, audience, placement, creative and buying experience can all differ. Lebesgue places Beauty and Health at the top of its reported CPM categories; that finding applies to its dataset rather than every beauty advertiser.

Lebesgue’s March 2026 report lists Beauty and Health CPM at $12.46 and a metric it calls customer acquisition cost at $34.29. These are vendor-reported category figures; confirm the outcome and customer definition before comparing them with your store’s new-customer CAC.

Influee separately attributes a $37.92 beauty CPA and 1.57 beauty ROAS to Triple Whale’s full-year 2025 data. Its $14.19 CPM is a platform-wide median, not a beauty CPM. These figures are secondary reporting and do not establish why costs changed or what your next campaign will earn.

LocaliQ separates traffic and lead objectives. Its 2026 Facebook benchmarks report Beauty and Personal Care CPC of $0.50 for traffic campaigns and $2.97 for lead campaigns, with a $50.91 cost per lead. A cheaper traffic click is not evidence of a cheaper purchaser; evaluate the action the campaign was bought to produce.

LocaliQ’s 2026 search benchmarks cover customer campaigns across Google Ads and Microsoft Ads. For Beauty and Personal Care, the source reports $4.62 average CPC and $39.25 CPL. They are search-advertising references, not Google-only ecommerce purchase costs.

The reported 10.35 percent conversion rate measures leads divided by clicks. A contact or form submission is not a paid order, and a beauty-services lead sample may differ substantially from an online cosmetics store. Do not use that rate to forecast store purchases.

A brand with little branded search can still investigate relevant nonbrand demand. Examine the actual queries, market, product fit and available volume before choosing Search. Google also sells Shopping and video placements, so choose the campaign type around the question you are testing.

Influee attributes a $5.28 beauty CPM and $18.82 beauty CPA to Triple Whale’s January–December 2025 data. Retain these as secondary historical references rather than proof that TikTok will be your cheapest acquisition channel. The underlying account mix and acquisition definition need to match your use before the comparison becomes useful.

Influee also reports 0.58 percent beauty CTR and 0.74 beauty ROAS, while acknowledging uncertainty about the click definitions behind some comparisons. A low attributed ROAS does not prove that missing sales occurred elsewhere. Investigate attribution coverage and actual order economics separately.

TikTok’s July 2026 budget guidance says daily budgets must exceed $50 at campaign level and $20 at ad-group level. Three separately budgeted groups above $20 each imply more than $60 daily, or more than $1,800 over 30 days. That is a selected configuration, not a universal monthly minimum or a guarantee of delivery or conversion volume; confirm the settings available for your campaign before funding it.

The rows describe different samples and outcomes. LocaliQ publishes its own campaign benchmarks; Influee reports figures attributed to Triple Whale. A source’s CAC or CPA label does not establish your definition of a first-time customer. Use the table to identify a relevant reference, then inspect its method before applying it.

Benchmarks describe the middle of a wide distribution. The spread around every median comes from a short list of factors, most of which a brand controls:

Choose the decision, affordable loss allowance and review window before setting daily spend. Budget separately for production, rights and management. Historical CPA can inform a scenario, but it does not establish how many orders the test will receive or whether that volume will distinguish two ideas.

Illustrative sensitivity check: a $2,000 media allowance corresponds to 80 attributed purchase events at $25 CPA, 50 at $40, or 25 at $80. These are arithmetic scenarios, not forecasts or sample-size requirements. Decide in advance what you will do if delivery is low or the result remains inconclusive.

A limited budget can become uninformative when it is divided among too many simultaneous questions. There is no standard test length; set the scope and duration from the buying cycle, delivery timing and available cash. Watch for these avoidable problems: