TikTok sells attention at $4.08 per thousand impressions. Meta sells it at $14.19. That gap looks like a decision — until you read the next column. TikTok's median ecommerce click-through rate is 0.61%; Meta's is 2.19%. TikTok's median cost per purchase is $17.07 with a 1.51x return on ad spend; Meta's implied cost per purchase is $43.75 with a 1.86x return. The cheap impressions buy fewer clicks, the clicks convert at different rates, and the customers they produce spend different amounts. This guide prices both platforms on the same $10,000 month, from the same benchmark cohorts, so the choice is arithmetic instead of fashion.
The benchmark table both platforms share
Every figure below is a reported median from a named cohort, not a vendor promise. The TikTok column comes from Triple Whale's ecommerce benchmarks (August 2025 to July 2026, 5,900+ brands) and DigitalApplied's Q1 2026 in-feed retail data. The Meta column comes from Triple Whale's full-year 2025 data (nearly 35,000 brands) and WordStream's 2025 benchmarks (1,180 campaigns, April 2024 to June 2025). Medians from different cohorts do not multiply into each other — the CPA median and the ROAS median describe different advertisers — so the scenario math in this guide derives one consistent set per platform and says so where it does.
| Metric | TikTok (ecommerce median) | Meta (ecommerce median) | Source |
| :--- | :--- | :--- | :--- |
| CPM | $4.08 | $14.19 | Triple Whale 2025–26 |
| CTR | 0.61% | 2.19% | Triple Whale 2025–26 |
| CPC (traffic) | $0.79 in-feed | $0.70 | DigitalApplied / WordStream |
| Purchase CVR | 1.56% | 1.60% | Triple Whale |
| Cost per purchase | $17.07 | $43.75 (derived) | Triple Whale / WordStream |
| ROAS | 1.51x | 1.86x | Triple Whale |
| Implied average order value | $25.78 (derived) | $81.38 (derived) | CPA × ROAS |
The implied order values are the first thing to notice. TikTok's $17.07 CPA times its 1.51x ROAS implies a $25.78 average order. Meta's $43.75 derived CPA times its 1.86x ROAS implies an $81.38 average order. TikTok wins the cheap purchase; Meta wins the expensive one. An advertiser selling a $25 impulse item and an advertiser selling an $80 considered purchase are not running the same campaign, and the platform medians say so.
> [!KEY TAKEAWAY]
> TikTok is the cheaper impression and the cheaper purchase; Meta is the higher return. At the reported medians, $10,000 buys about 586 TikTok orders worth $15,100 and about 229 Meta orders worth $18,600. The right platform follows the average order value, not the CPM.
The $10,000 month, priced both ways
The scenario uses each platform's own reported medians, kept internally consistent. Spend is $10,000 in both cases.
$
\text{TikTok orders} = \frac{\$10,000}{\$17.07} = 586
$
$
\text{TikTok revenue} = 586 \times \$25.78 \approx \$15,100 \quad (1.51\times)
$
$
\text{Meta orders} = \frac{\$10,000}{\$43.75} = 229
$
$
\text{Meta revenue} = 229 \times \$81.38 \approx \$18,600 \quad (1.86\times)
$
Why it matters
The two platforms do not sell the same customer. TikTok's feed is built for discovery: a user who did not intend to buy sees a product, and the purchase is an impulse. Meta's feed is built for intent and retargeting: a user who has already shown interest sees the product again, and the purchase is a decision. The medians prove the difference — a $25.78 implied order on TikTok against $81.38 on Meta — and an advertiser who buys TikTok traffic expecting Meta order values will conclude TikTok "doesn't convert," when the truth is that TikTok converted exactly the customer it reaches. The platform choice is a customer choice, and the benchmark table is the cheapest way to make it.
The impression math explains why the CPM gap does not decide anything on its own. TikTok's $4.08 CPM buys 2,450,980 impressions for $10,000; Meta's $14.19 buys 704,722. But TikTok's 0.61% CTR turns those impressions into 14,951 clicks, while Meta's 2.19% CTR turns its fewer impressions into 15,433 clicks. The click counts are nearly identical — the CPM gap is almost exactly cancelled by the CTR gap. What differs is what happens after the click: TikTok's 1.56% purchase rate against a $25.78 order, Meta's 1.60% against an $81.38 order.
> [!TIP]
> Price the test on the learning phase, not the floor. A $50-a-day TikTok campaign buys about three purchases a day at the median CPA and never exits learning — the "test" produces noise, not a verdict. Fund 50 × CPA ÷ 7 per ad set or do not test at all.
The break-even ROAS test
Neither median ROAS is good or bad on its own. It is good or bad against the gross margin. The break-even return is one divided by the gross margin:
$
\text{Break-even ROAS} = \frac{1}{\text{Gross margin}}
$
| Gross margin | Break-even ROAS | TikTok 1.51x | Meta 1.86x |
| :--- | :--- | :--- | :--- |
| 30% (physical goods, resold) | 3.33x | Loses money | Loses money |
| 50% (branded DTC) | 2.00x | Loses money | Loses money |
| 70% (digital, high-margin) | 1.43x | Profitable | Profitable |
A physical-goods brand at 30% margin needs 3.33x to break even on ad spend alone — neither platform's median gets there, which is why median advertisers at 30% margin are either losing money on first purchase and recovering it on repeat, or not actually at the median. A 70%-margin digital product breaks even on both. The honest question is not "which platform has the higher ROAS" but "which platform's ROAS clears my margin" — and the answer changes with the margin, not with the platform.
> [!FOUNDER]
> "Every founder I have watched pick TikTok because the CPM is a third of Meta's made the same error: they priced the impression and ignored the order value. The platform that wins is the one whose median customer spends what your customer spends. Price the order, not the impression." — Enow A. Jovial, Founder & Chief Executive Officer
The learning-phase budget neither platform advertises
Both platforms need about 50 conversion events a week per ad set before the algorithm exits the learning phase and stabilizes delivery. That requirement sets a real minimum budget far above the published floors. The formula is 50 times the cost per purchase, divided by seven days:
$
\text{Daily budget} = \frac{50 \times \text{CPA}}{7}
$
| Platform | CPA | Learning-phase daily budget | Monthly equivalent |
| :--- | :--- | :--- | :--- |
| TikTok | $17.07 | $121.93 | $3,658 |
| Meta | $43.75 | $312.50 | $9,375 |
The published floors are much lower — TikTok accepts $50 a day at campaign level and $20 a day at ad-group level; Meta accepts $1 a day for impression-charged ad sets and $5 a day for conversion-charged ad sets in US-tier accounts, with $40 a day for low-frequency events. The floors are what the system accepts. The learning-phase budget is what the system needs. A $50-a-day TikTok campaign buys about 3 purchases a day — 21 a week — and never exits learning. Budgeting the floor and wondering why performance never stabilizes is the most common paid-social mistake in ecommerce, and both platforms' documentation says so in different words.
What the headline CPA hides on TikTok
TikTok's $17.07 headline CPA is a platform-reported figure, and part of it is not real buyers. Lunio's analysis of 414 million retail ad clicks (Q4 2025 to Q2 2026) puts TikTok's invalid-traffic rate for retail at 5.56%, climbing from 3.94% to 6.61% across the period. Invalid clicks, views, and add-to-carts inflate the funnel before the purchase event, so the cost of a genuine customer sits above the headline:
$
\text{Genuine CPA} = \frac{\$17.07}{1 - 0.0556} = \$18.07
$
The $1.00 gap is small at retail scale and large at volume: on 586 orders a month it is $586 of spend reaching no real shopper. Meta is not immune — every auction platform carries invalid traffic — but the published retail-specific figure exists for TikTok, and the honest budget adds the adjustment rather than pretending the dashboard number is the whole cost. Server-side tracking (Meta's Conversions API, TikTok's Events API) is the fix both platforms recommend, and it is a one-time engineering cost against a permanent 5% tax.
Seasonality: the Q4 tax
Benchmarks are annual medians; auctions are seasonal. TikTok CPMs rise 40–80% in Q4, and Meta's median CPM already rose about 20% year over year through full-year 2025, with Meta's average ad price up 12% year over year in both Q1 and Q2 2026. A TikTok CPM of $4.08 becomes $5.71 to $7.34 in Q4, and the $17.07 CPA rises with it unless creative quality rises faster. January and February are the cheapest months on both platforms. The planning rule is to test creative in January, scale in Q2 and Q3, and defend margin in Q4 — not to launch a new account in November and judge the platform by Black Friday auction prices.
Attribution: the same window, different trust
Both platforms default to a 7-day click, 1-day view attribution window. The difference is not the window but what it counts. TikTok's feed is a 99% mobile environment, so every TikTok dollar is effectively a mobile dollar, and its attribution leans on view-throughs from fast-scrolling feeds. Meta splits across Facebook, Instagram, Messenger, and Audience Network, with desktop still present. The practical consequence is that TikTok's reported ROAS overstates incrementality more than Meta's when both are read at face value — a view-through on a feed the user scrolled past in 1.5 seconds is not the same evidence as a click. Holdout tests (geo-split or PSA) are the only way to know the true incremental ROAS on either platform, and the article that skips them is quoting the platforms, not measuring them.
The decision table
| Attribute | TikTok Ads | Meta Ads | Authority |
| :--- | :--- | :--- | :--- |
| Median CPM | $4.08 | $14.19 | Triple Whale 2025–26 |
| Median CTR | 0.61% | 2.19% | Triple Whale |
| Cost per purchase | $17.07 | $43.75 derived | Triple Whale / WordStream |
| Median ROAS | 1.51x | 1.86x | Triple Whale |
| Best order value | Under ~$40 | Over ~$60 | Implied AOV derivation |
| Learning budget / month | $3,658 | $9,375 | 50 × CPA ÷ 7 |
| Creative cadence | New creative every 1–2 weeks | New creative every 3–4 weeks | Platform guidance |
| Attribution default | 7-day click / 1-day view | 7-day click / 1-day view | Platform defaults |
Who should NOT pick TikTok first
TikTok first is wrong for three advertisers. First, a high-AOV considered purchase above $100: TikTok's median customer spends $25.78, and the platform's impulse mechanics fight the research cycle. Second, a B2B or regulated advertiser: TikTok's B2B CPA median is $104.69 with a 0.84% CTR, and finance verticals pay $13.20 CPMs for 0.72% CTRs. Third, an advertiser with no creative engine: TikTok creative fatigues in one to two weeks against Meta's three to four, and an account that cannot ship a new concept every week will pay the fatigue tax in rising CPMs. Meta first is wrong for the mirror image: a $20 impulse product with strong short-form creative and a founder who can film daily. The platform follows the product, not the trend.
> [!WARNING]
> Do not judge either platform on a $500 test. $500 buys 29 TikTok purchases at the median CPA and 11 Meta purchases — neither sample is large enough to exit the learning phase, and the "winner" of a $500 test is noise. The minimum honest test is one learning-phase month: $3,658 on TikTok, $9,375 on Meta, at the median CPAs above.
The five-week test both platforms deserve
1. Week 1: set the break-even. Compute break-even ROAS from the gross margin and write it on the test plan. If the margin is 30%, the target is 3.33x, and the test is judged against that number, not against the platform median.
2. Week 2: fund the learning phase. Set the daily budget to 50 times the expected CPA divided by seven. On TikTok that is $121.93 a day at the $17.07 CPA; on Meta, $312.50 at $43.75.
3. Week 3: ship three creatives per platform. One founder-led video, one product demo, one customer proof. Kill any creative below the platform's median CTR after 3,000 impressions — 0.61% on TikTok, 2.19% on Meta.
4. Week 4: read incrementality, not the dashboard. Compare attributed revenue against the holdout or the pre-test baseline. The dashboard ROAS is the claim; the holdout is the evidence.
5. Week 5: decide on profit, not ROAS. A 1.51x ROAS at 70% margin is a profitable channel. A 1.86x ROAS at 30% margin is a loss. Scale the profitable one; fix the margin or the offer on the other.
Execution checklist
The cheap platform is the one whose median customer buys what you sell at a return your margin can survive. For a $25 impulse product with daily creative, that is TikTok at $17.07 a purchase. For an $80 considered purchase with a longer funnel, that is Meta at $43.75 a purchase and 1.86x. Run the break-even first, fund the learning phase second, and let the order value decide. The Google Ads vs Meta Ads ROAS playbook prices the search alternative on the same terms, the LinkedIn vs Google B2B lead cost playbook covers the B2B side of the same decision, and the email marketing pricing comparison prices the retention channel that makes a 1.51x first purchase profitable on the second. Estimate the funnel with the pricing calculator before the first dollar goes to either auction.