Paid Search vs Paid Social for DTC: Where to Spend

The paid search vs paid social debate is the wrong frame for a DTC budget, because the two channels are not competing for the same job. Social interrupts a passive scroll and manufactures a want; search waits at the bottom of the funnel for a want that already exists. The split question is really a funnel-position question, and most DTC brands get it wrong because last-click hands social's downstream revenue to Google and makes search look like the hero.

The split decision comes down to how to weight the two by stage, margin, and category, and how to stop trusting in-platform ROAS as the arbiter.

Paid Search vs Paid Social: Harvest vs Manufacture

Paid social manufactures demand for a cold audience; paid search harvests demand that already exists. You need both, and most DTC brands underweight social early because last-click makes search look like the hero.

• The split is not a platform preference, it is a funnel-position decision.
• Search ROAS is inflated by branded queries; social ROAS is suppressed by cold prospecting.
• The real KPI for the split is incremental contribution margin, not platform ROAS.

The Core Split: Demand Generation vs Demand Capture

Reframe the question. The difference between Meta and Google is not the platform, it is the user's intent at impression time.

On social, the user is scrolling for entertainment. The ad is an interruption. The creative itself has to do the persuasion work, manufacture the want, and earn the next five seconds within the first two. That is why paid social is the engine of demand generation: it can introduce a category nobody was searching for and make people want it.

On search, the user has already raised a hand. They typed "best clean beauty foundation" or your brand name. They have moved past awareness and consideration on their own. Paid search is unmatched at fulfilling that intent, but it cannot create it.

Taylor Holiday's framing for Common Thread Collective is the cleanest version of this: Google exists "post-demand." You cannot raise the search budget to "get more demand" because search volume is fixed by what the market is already asking for. Lift on search comes from upstream activity that puts your name into people's heads.

For the broader category frame, see performance marketing.

Why Social Feeds the Search You Win (The Halo Effect)

This is the single most under-appreciated mechanism in DTC media buying. A consumer sees a Meta or TikTok ad, does not click, and three days later opens a tab and searches the brand name. Google Search gets credit under last-click. Social looks inefficient. The brand cuts social to "improve efficiency," and three to six weeks later the branded search volume that was carrying the apparent ROAS collapses with it.

The documented cases are blunt. A DTC skincare brand paused Meta prospecting on weak platform ROAS; subsequent analysis showed those exact ads were lifting Sephora sell-through by 18%, and the pause slowed Sephora sales immediately. A fast-casual brand took a 10-day paid blackout and saw site visits fall 17% and leads fall 54%; relighting the spend drove a 224% rebound in leads. A separate brand aligned Meta creative with its existing Google strategy and saw Meta CTR move from 2.46% to 8.20% while Google Ads revenue climbed 26.45% in the same window. The channels work as one organism.

This is also why the right diagnostic metrics are not Meta-reported ROAS or Google-reported ROAS. See MER vs ROAS for the blended truth metric, and attribution stack for the plumbing.

The 2026 Economics: What CAC and ROAS Actually Look Like

Set expectations honestly. CAC has risen an estimated 40-60% between 2023 and 2025 across DTC, and industry data puts median ecommerce ROAS at roughly 2.04:1 in 2026. Half of all ecommerce businesses are running under a 2:1 platform return.

Inside that median, the structural pattern is consistent: social prospecting prints lower numbers than search, because it is doing the harder job.

These are directional bands, and each channel's minimum budget and learning floor sits in our PPC platform comparison. The Google Search row is inflated by branded queries; the Meta row is suppressed by cold prospecting. Both are doing exactly what their funnel position predicts.

The real decision metric is not platform ROAS at all. It is CAC payback period and contribution margin. A beauty brand acquiring a $40 customer on a $50 first order is at 1.25x day-one ROAS and looks terrible on the dashboard. If that customer subscribes monthly and the LTV:CAC settles at 4:1 or 5:1, the same low day-one number is a winning unit economic. The full math lives in CAC and payback and contribution margin.

The Search ROAS Illusion (And Why It Burns Brands)

Branded keywords routinely hit 10x to 20x ROAS. The reason is simple: the demand was already created, and Google is the tollbooth collecting at the last step. Branded clicks blend into the channel's average and pull the apparent return into the 3.0x - 8.0x band you see in the table above.

The trap is treating that blended number as the marginal return on the next search dollar. It isn't. The next dollar usually buys non-brand inventory, which performs nothing like brand.

Two rules fix it:

• Separate branded from non-branded in your reporting before you judge any "search ROAS" number.
• Treat branded search as a defensive line. It is harvesting demand, not creating it. The demand-creation work happens upstream.

For the deeper Google build, see Google Ads for DTC.

Budget Split by Brand Stage

Three concrete recipes. Each is a starting point, not a law, and each shifts with margin and product searchability.

Early Stage (under $5M)

Skew hard to social, 70-80% of paid. Nik Sharma's framing of the "minimum mix" puts social on upper and mid-funnel work and search strictly on the lower-funnel intent that exists.

There is no branded search volume yet, so a non-brand search budget is not buying intent that recognizes you. The 20-30% on search exists to catch what social creates and to capture any exact-match category queries the product happens to map onto.

Anti-pattern: piling non-brand search spend on a brand nobody has heard of. You are paying premium prices to compete with bigger players inside their established categories.

Growth Stage ($5M-$50M)

The split typically stabilizes at roughly 50-60% social and 40-50% search and Shopping. Total paid often runs 8-18% of revenue.

The mechanic has changed. Social is now manufacturing meaningful branded search volume, and Performance Max plus branded RSAs on search are required to capture it before competitors do. Inside paid, the common allocation pattern is the 50/30/20 rule: 50% to direct acquisition (the Meta/Google core), 30% to retention (email, SMS, retargeting), 20% to brand and innovation.

The social-side mechanics for getting here are in scaling paid social, and the build sits in account structure.

Mature Stage ($50M+)

The diversified 70/20/10 model becomes common: 70% to proven core channels (often a near-even social/search split inside that), 20% to emerging growth bets (CTV, retail media), 10% to pure experimentation.

Cody Plofker's framing at Jones Road is the warning: you can direct-response your way to an 8-figure brand, but you cannot direct-response your way to a 9-figure brand. The DR ceiling is real, and crossing it requires brand investment that does not look like a ROAS on the dashboard.

Media Mix Modeling becomes the budget arbiter rather than in-platform ROAS. MMM is a time-series regression that estimates each channel's incremental contribution to total sales, which lets you compare social and search on the same axis without trusting last-click. The testing layer for verifying MMM outputs is incrementality via geo holdouts.

Budget Split by Category and Margin

Unit economics override stage. This is the single sharpest split rule on the page.

Break-even ROAS is just 1 divided by gross margin. A 30% gross margin requires 3.3x ROAS to break even. A 70% gross margin needs only 1.43x. The margin ceiling decides whether you can absorb a 1.5x day-one social return at all.

High-margin, replenishable categories tolerate a 1.25x day-one social return because the second and third purchases bring the cohort to a 4:1 LTV:CAC. Low-margin or low-frequency categories cannot. A 20% gross margin requires a 5x break-even and cannot afford a 60-day payback window; the budget has to go where intent is already present.

The deeper LTV math lives in customer lifetime value and subscription LTV.

The Searchability Test (The Fastest Allocation Heuristic)

One diagnostic question cuts through most of the budget debate for a new SKU.

Do people already search for what you sell? If no, the budget goes to social. If yes, the budget goes to search.

Novel product with zero search volume (a new posture wearable, an aesthetic emergency kit, a category that does not exist as a query yet): 80%+ to social to educate the market and create the search you will eventually capture.

Commoditized necessity with established demand (replacement toothbrush heads, printer ink, refurbished laptops): the demand exists and visual storytelling adds limited lift. Majority to search and Shopping, fight for the click at the moment of intent.

The trap on both sides is symmetrical. Spending non-brand search budget on a category nobody knows to search for is lighting money on fire. Spending pure social budget on a high-intent commodity is paying interruption prices to win clicks that were free to capture at the search bar.

Hybrid Funnel: How Search and Social Run Together Inside Google

Even inside Google alone, the right answer is not "PMax vs Search." It is both, with negative-keyword discipline so they do not cannibalize.

Performance Max runs across Search, Shopping, YouTube, Display, Discover, Gmail, and Maps. It does not use keywords; it uses search themes, audience signals, and an asset library, and the algorithm actively reads the creative to decide who sees what. Pure Search runs on the SERP only, on exact-match and phrase-match keywords, with full negative-keyword control and explicit search-query reports.

The hybrid play: account-level negative keyword lists exclude your high-value Search keywords from the PMax campaign, so PMax hunts incremental discovery and cross-channel remarketing while pure Search owns the known-good intent. Without this exclusion, PMax cannibalizes the cheaper, more transparent Search inventory.

PMax has hard anti-use cases. The 30/30 rule requires at least 30 conversions in 30 days before automated Target CPA bidding can stabilize; below that the learning phase never closes. Daily budget must run 3x to 5x Target CPA or the algorithm starves and chases junk auctions. Highly regulated industries (pharma, restrictive healthcare, financial services) struggle because PMax dynamically mixes thousands of headline-description-image combinations that no compliance team can pre-approve. And any account with bot-contaminated conversion data will see the AI optimize toward the contamination.

The full Google build sits at Google Ads for DTC and the video extension at YouTube ads.

Why Creative Now Decides the Split

Privacy changes made the creative asset the primary targeting variable on both sides of the line. That re-weights the entire budget conversation.

On Meta and TikTok, the algorithm reads engagement on the creative to decide who sees it next. A UGC skincare demo attracts and retains skincare enthusiasts; the algorithm watches and amplifies. Cody Plofker's repeated point is that scaling on Meta is a continuous search for new creative angles and new buyer personas, not new targeting toggles.

On Google PMax and Demand Gen, the same logic now applies. The AI literally reads the visual cues, audio, captions, and context of an uploaded asset to determine the audience. Featuring a dog in the video is the targeting signal; you no longer need to select "pet owners" as a demographic.

Format matters. UGC-style creative (real environments, natural lighting, direct-to-camera, authentic product interaction) generates roughly 3-5x higher engagement and up to a 2.8x higher conversion rate than polished studio work. Ecommerce sites featuring UGC retain visitors 90% longer, and shoppers rate UGC as 2.5x more authentic than brand-produced assets. Documented agency case data shows CPAs dropping anywhere from 9.2% to 50% on migrations from studio to native creator-led 9:16 video with burned-in captions, with CTRs clearing 1.2%.

The operational change that closed the social/search creative gap is the Social Video Upload beta inside PMax and Demand Gen. It lets brands import up to five social-style 9:16 vertical videos directly into the Asset Library, bypassing the old requirement to host every video on a YouTube channel first. The winning Meta creative ports straight into Google.

This forces a real creative pipeline. The mechanism is detailed in creative-as-targeting and the strategy side in creative direction.

What This Costs You to Run

Brutal honesty on production. Scaling paid social past roughly $50,000 per month in spend requires 15-30 (often 20-40) new ad concepts per month to fight creative fatigue. Without that pipeline, performance decays and the social-heavy split stops working on its own merits.

Three production models, with 2026 cost structures:

• Freelance creators / AI UGC. Individual creators cost $100-$500 per video deliverable, with usage rights typically adding 25-100% on top. AI-generated UGC scales the volume cheaply but tends to underperform on hero testimonial slots where raw authenticity is doing the work.
• Performance creative agencies. Retainers of $5,000-$25,000 per month, with effective per-video costs in the $300-$1,000+ range. The premium covers creator vetting, brief management, and variation production for each platform.
• In-house teams. Breakeven is around $20,000-$50,000 per month in media spend with a 15-30 asset monthly volume. A fully loaded in-house setup (strategist at $65k-$90k, freelance editors, studio costs) runs $83,000-$120,000 annually.

The decision is volume-driven. Below the threshold, an agency's per-video economics win. Above it, in-house iteration speed and brand knowledge pay back. If neither is in place, the social-heavy split is not viable yet, and the budget defaults to search until the pipeline exists.

The build-vs-buy call is at agency vs in-house, and the conversion-side service at performance creative.

The Decision Framework: A Four-Step Process

1. Calculate break-even ROAS and CAC ceiling from gross margin. Low margin pushes you to search; high margin plus repeat unlocks social-heavy. See contribution margin.
2. Run the searchability test. Existing demand goes search-led. Net-new category goes social-led. No exceptions for the first 90 days.
3. Set the stage-appropriate split as a starting point, not a target. 70/30 social/search early, 55/45 in growth, roughly even at maturity. Adjust off margin from step one.
4. Prove the split with geo holdouts, not platform ROAS. Match test and control markets on historical conversion rate and growth trajectory, pause or scale a channel in the test markets only, run 4-6 weeks avoiding seasonal events, and analyze with Difference-in-Differences or Synthetic Control. See incrementality testing and attribution.

The Comparison Matrix

Neither column is a complete strategy. Social introduces the brand; search ensures the transaction closes when the introduced consumer is ready.

Common Failure Modes

• Cutting social on flat day-one ROAS, then watching branded search collapse three to six weeks later.
• Judging search performance on blended ROAS without separating branded from non-brand inventory.
• Running PMax under the 30/30 conversion threshold and starving the algorithm of signal.
• Pushing social-heavy without a creative pipeline that can deliver 20+ concepts per month.
• Trusting in-platform reporting end-to-end instead of running periodic geo holdouts to verify the split.

Where to Go Next

If your blended ROAS dropped without an obvious cause, the first read is why ads stopped working. For a structured look at the current account against this framework, the paid media audit is the entry point, and our paid media service runs both sides of the split from there. And when the pipeline that makes a social-led split actually viable is what's missing, the conversion page is performance creative.