The Paid Media Budget Playbook

01

How Much To Spend On Paid Media

Start with a revenue anchor, then check it against what your unit economics can absorb. The anchor tells you what peers spend. The unit economics tell you what you can afford to spend without losing money on every customer.

Gartner's 2026 CMO Spend Survey put total marketing budgets at 7.8% of company revenue, up from 7.7% in 2025. The 2025 survey found paid media took 30.6% of the marketing budget, or about 2.4% of revenue. Keep in mind who answered: 401 marketing leaders in North America and Europe, most at companies above $1 billion in revenue. Growth-stage companies usually spend a larger share because they are buying growth, not defending share.

For software, SaaS Capital's 2026 survey of more than 1,000 private B2B SaaS companies found a median of 8% of ARR on marketing and 15% on sales, 23% in total. Equity-backed companies spent about twice as much on marketing as bootstrapped ones.

Build The Number Bottom Up As Well

Top-down anchors ignore your pipeline math. Work backwards from the revenue target instead, then compare the two numbers. If the bottom-up figure is far above the top-down one, your targets or your economics need rethinking before you buy media.

Monthly paid budget = (new customers needed per month / lead-to-customer rate) x cost per lead

Example: you need 10 new customers a month, 12% of paid leads become customers and your blended cost per lead is $150. You need 10 / 0.12 = 84 leads, and 84 x $150 = $12,600 a month in media. Add 10% to 15% for testing and you land near $14,000.

02

Channel Benchmarks For 2025 And 2026

Use benchmarks to spot outliers, not to set targets. The datasets below come from different advertiser mixes, objectives and definitions, so compare a channel to its own row, not across rows. WordStream data skews toward lead generation for small and midsize advertisers; Triple Whale data comes from ecommerce brands; Metadata and AdConversion track B2B software.

What The Numbers Say

• Search is expensive per click but converts on intent. WordStream found average cost per lead fell in 2026 for the first time in five years, while conversion rates rose in 87% of industries.
• Meta is the cheapest lead source in this set, but Triple Whale saw Meta CPMs rise 13.24% year over year, so cheap reach is getting less cheap. Meta's own Q3 2026 results, reported October 21, 2026, put the average price per ad up 14% year over year on 10% more impressions.
• LinkedIn costs the most per impression and per lead, yet Metadata found it returned 10.2x pipeline per lead-gen dollar versus 7.97x across all channels. LinkedIn's native lead forms ran at a $193 CPL against $346 for landing-page capture.
• LinkedIn CPCs swing by season. HockeyStack's report on 70+ B2B SaaS companies saw average CPC move from $10.48 in Q1 to $15.72 in Q3.
• TikTok reach is cheap, conversion is harder. Triple Whale saw TikTok CPM fall 28.48% while conversion rate fell 36.79%.

Plan For Auction Inflation, Not Flat Prices

A 14% rise in Meta's average price per ad means a campaign that held its cost per lead flat through 2026 got about 14% more efficient in real terms, and a 2027 budget that assumes 2026 prices is already short. Build a price assumption into the plan: take each channel's trailing twelve-month CPM trend, apply it to next year's impression need, and show finance the gap between flat and trended. If the gap cannot be covered, the answer is conversion rate and creative, not more spend at the old price.

For how each channel works in practice, see our pages on Google Ads, Meta ads, LinkedIn ads and YouTube and TikTok ads.

03

Unit Economics: The Three Numbers That Cap Your Spend

Every paid media plan rests on three numbers: break-even ROAS, target ROAS and maximum cost per acquisition. Calculate all three before launch and write them at the top of your reporting sheet. If a campaign cannot reach them at any spend level, cut it or fix the offer.

1. Break-Even ROAS

Break-even ROAS = 1 / gross margin

Gross margin is revenue minus cost of goods (product, fulfillment, payment fees, delivery labor) as a share of revenue. At a 60% gross margin, break-even ROAS is 1 / 0.60 = 1.67. Every $1 of ads must return $1.67 of revenue just to cover itself on the first purchase.

2. Target ROAS

Target ROAS = 1 / (gross margin - target contribution margin)

Break-even is a floor, not a goal. If your gross margin is 60% and you want 20% of revenue left after ad costs, target ROAS is 1 / (0.60 - 0.20) = 2.5. Check it: $100 of revenue at 2.5 ROAS costs $40 in ads. Gross profit is $60, so $60 - $40 = $20 of contribution, which is 20% of revenue.

3. Maximum CPA From Lifetime Value

Max CPA = (average revenue per month x gross margin x expected lifetime in months) / target LTV to CAC ratio

Worked B2B example: a customer pays $500 a month, gross margin is 80% and the average customer stays 24 months. Gross-margin lifetime value is $500 x 0.80 x 24 = $9,600. At a 3 to 1 LTV to CAC target, max CAC is $9,600 / 3 = $3,200. Payback is $3,200 / ($500 x 0.80) = 8 months.

Now push it down the funnel. If 10% of leads become customers, max cost per lead is $3,200 x 0.10 = $320. If 5% of landing page visitors become leads, max CPC is $320 x 0.05 = $16. Against the table above, LinkedIn's $202 median CPL fits inside that ceiling; a $346 landing-page CPL does not.

Worked ecommerce example: average order is $80 at a 55% gross margin. First-order break-even CPA is $80 x 0.55 = $44. If customers place 1.8 orders in 12 months, 12-month gross profit is $80 x 1.8 x 0.55 = $79.20. Keeping 30% of that as profit sets max CPA at $79.20 x 0.70 = $55.44.

04

How To Split The Budget

A common practice is the 70/20/10 split: 70% on proven campaigns, 20% on scaling promising ones, 10% on new experiments. It is a planning habit, not a law. The right ratio depends on how much proof you have. A new account might run closer to 50/30/20 until something works; a mature account might hold 80/15/5.

Channel Count By Budget Tier

Size each campaign so it can buy enough conversions to optimize. A simple check: campaign budget / target CPA = conversions per month. If a $3,000 campaign has a $200 target CPA, it can only buy 15 conversions a month, which is thin for automated bidding. Consolidate before you split.

Automation Moved Upstream: What To Keep Manual

Google spent 2026 pushing control up a level. AI Max, which expands search campaigns to queries you did not bid on and picks landing pages for you, extended to Shopping and travel in April 2026, and Google's AI Brief lets you describe the audience and constraints in plain language instead of building keyword lists. Meta's Advantage+ does the same for audiences and placements. The pitch is reach; the cost is that a mistake now compounds across more queries and more inventory before anyone sees it.

Keep five things manual whatever the platform suggests. Brand exclusions, so the automation cannot report your own brand searches as incremental growth. A weekly search terms and placement review with a negative list that grows every week. Landing page controls, so URL expansion cannot send a CFO prospect to a careers page. One primary conversion action per campaign goal, because broad matching optimised to a soft event buys soft traffic. And a monthly incrementality check, since the systems that expand your reach are also the systems grading it.

05

Creative Testing Cadence

Creative is the largest lever you control. NCSolutions' 2023 analysis of nearly 450 campaigns found creative drove 49% of incremental sales, more than brand, reach, targeting or recency. On platforms where targeting is increasingly automated, the ad itself does more of the targeting work.

A Cadence That Fits The Budget

1. Test concepts before variants. A concept is a distinct message or angle; a variant is a new hook, headline or format of the same idea.
2. Write a hypothesis for each concept: the audience, the problem, the proof and the expected metric change.
3. Give each test a fixed budget and a stop rule, for example two times target CPA spent with no conversion.
4. Judge on the business metric (CPA, cost per qualified lead), not on CTR alone.
5. Move winners into the core bucket and retire losers. Log every result so you do not retest dead ideas.
6. Watch for fatigue: rising frequency with falling CTR and rising CPA over two to three weeks means refresh.

Strong ads need strong destinations. Pair creative tests with landing page and conversion rate work, since doubling landing page conversion halves your cost per lead at the same CPC.

06

Tracking Foundations

Bad tracking makes every other section of this playbook useless. Fix it before you scale spend.

UTM Naming

Google Analytics supports utm_source, utm_medium, utm_campaign, utm_id, utm_term, utm_content and utm_source_platform. Values are case sensitive: Google treats utm_source=google and utm_source=Google as different sources. Pick one convention, write it down and enforce it.

GA4 Default Channel Groups

GA4 sorts traffic into channels such as Paid Search, Paid Social, Paid Video, Display and Paid Other. Paid channels require a medium matching the pattern cp, ppc, retargeting or anything starting with paid. A medium like social_paid or ads falls outside the rule and lands in Unassigned or the wrong group, which quietly breaks channel reporting.

Server-Side And Offline Conversions

• Meta Conversions API: sends events from your server or CRM to Meta. Meta recommends running it alongside the pixel and deduplicating by matching the pixel eventID to the API event_id and event name.
• Google offline conversion import: store the GCLID with each lead, then upload the conversion when the deal closes. Google recommends enhanced conversions for leads, which uses hashed first-party data such as email; Google reports a median 10% increase in conversions versus standard imports. From June 15, 2026, these uploads move to the Data Manager API.
• LinkedIn: Campaign Manager also offers a Conversions API for sending CRM events.
• Optimize to revenue, not form fills. Send qualified-lead and closed-won events back to each platform so bidding learns which clicks become customers.

• One written UTM convention, lowercase, shared with everyone who builds links
• Every paid medium maps to the right GA4 channel group
• Pixel and server events deduplicated
• Click IDs (GCLID, fbclid, li_fat_id) captured in hidden form fields and stored in the CRM
• Qualified-lead and closed-won stages synced back to ad platforms
• Conversion actions audited: one primary action per campaign goal

If this list has gaps, start with analytics and tracking and CRM integration before adding budget.

07

Reading Results Without Fooling Yourself

Every ad platform grades its own homework. Each claims credit using its own attribution rules, so adding platform-reported conversions together almost always exceeds the real total. Three tools keep you honest.

Know The Attribution Window

Meta's default setting is 7-day click, 1-day engage-through and 1-day view-through. In March 2026 Meta narrowed click-through to link clicks and moved likes, comments, shares and saves into engage-through. A report with view-through included can look far better than one without it. Compare periods only on the same setting, and report platform numbers next to CRM numbers.

Run Incrementality And Holdout Tests

Incrementality asks what would have happened without the ads. A holdout test withholds ads from a random audience or set of regions and compares outcomes. In November 2025 Google said experiments that once cost upwards of $100,000 can now run for $5,000, which puts lift testing within reach of growth-stage budgets.

Incremental lift = (test group conversion rate - control group conversion rate) / control group conversion rate

Example: regions with ads convert at 2.4% and holdout regions at 2.0%. Lift is (2.4 - 2.0) / 2.0 = 20%. If the platform claims 500 conversions but the test implies only 150 extra, your incremental CPA is spend / 150, not spend / 500. At $30,000 spend that is $200, not $60.

Size The Test Before You Run It

Most holdout tests fail before launch because the groups are too small to detect the lift that would change a decision. A quick sizing rule for a conversion-rate test at roughly 80% power and 5% significance:

Users needed per group = 16 x p x (1 - p) / d^2 p = baseline conversion rate, d = smallest absolute lift worth detecting

Example: landing pages convert at 2.0% and you only care about a lift of at least 20% relative, so d = 0.4 percentage points, or 0.004. Users per group = 16 x 0.02 x 0.98 / 0.004^2 = 19,600. Two groups, so about 39,000 users in the test window. At 20,000 paid visitors a month that is a two-month test, which tells you whether the test is affordable before you spend a dollar on it. If the required size is out of reach, test a bigger lever (a channel on or off, not a creative tweak) or move to a geo test with matched markets, where the unit is the region and the arithmetic is the same.

Marketing Mix Modeling Basics

A marketing mix model (MMM) uses statistics on historical spend and sales to estimate each channel's contribution, including offline effects and seasonality. Google made its open-source MMM, Meridian, available to everyone in January 2025. Its documentation recommends weekly data and at least two years of history for geo-level models, three years for national models. Many companies spending under $50,000 a month lack the channel variety and history a model needs, so start collecting clean weekly spend and outcome data now.

08

Monthly Review Template

Hold one 60-minute review each month with marketing and finance in the room. Use the same sheet every month so trends are obvious.

1. What changed: spend, CPA and volume by channel versus last month and target.
2. Why: creative, auction pressure, landing page, seasonality or tracking.
3. What we learned: tests concluded and their result.
4. Decisions: budget moves between buckets and channels, with the reason written down.
5. Next month: tests to launch, owners and dates.

Our marketing budget calculator helps set the next month's plan, and a periodic paid media audit catches drift the monthly review misses.

09

A 30/60/90 Day Plan

1. Days 1 to 10

   Calculate break-even ROAS, target ROAS and max CPA. Audit tracking, UTM convention and GA4 channel mapping. Capture click IDs in the CRM.

2. Days 11 to 30

   Launch one intent channel and one prospecting channel. Put 70% in the most proven campaign type, the rest in tests. Launch the first 2 to 4 creative concepts.

3. Days 31 to 45

   Connect Conversions API and offline conversion imports. Start sending qualified-lead events back to platforms. Cut tests that hit the stop rule.

4. Days 46 to 60

   Shift budget to campaigns within target. Add retargeting. Run the first monthly review with finance and reset the split.

5. Days 61 to 75

   Design a holdout or lift test on the largest channel. Refresh fatigued creative. Test one new channel from the 10% bucket.

6. Days 76 to 90

   Read the lift test, adjust CPA targets to incremental numbers, and set next quarter's budget from real CAC and payback data.

• Unit economics written down and agreed with finance
• Tracking checklist complete
• Budget split and stop rules documented
• Creative test log running
• First incrementality test designed by day 75
• Quarter-two budget based on measured CAC, not benchmarks