Most SaaS growth teams set their channel budget split by some combination of intuition, last-click attribution, and whichever platform rep most recently made a compelling pitch. On a €100k/month budget, getting that split wrong by fifteen percentage points means roughly €180k of misallocated annual spend — money poured into a channel that is over-credited by its own attribution while a genuinely more efficient channel goes underfunded. The fix is to stop treating allocation as an opinion and start treating it as a calculation.
This is a strategy guide for SaaS growth leads who want a concrete budget allocation framework across Google, Meta, and TikTok. We cover what each channel actually does for SaaS, a core allocation formula, the CAC, LTV, and payback inputs that drive it, how the split changes by company stage from PMF to enterprise, why incrementality must replace platform-reported numbers, the triggers that prompt reallocation, the testing reserve, and worked examples across stages. We focus on SaaS specifically because subscription economics — payback periods, retention, LTV skew — make the allocation logic different from e-commerce. For the two-channel deep dive, our Meta vs Google budget allocation for SaaS guide is the natural companion to this three-channel framework.
Every platform over-credits itself, in its own way. Google's models over-credit branded and high-intent clicks that would have converted regardless. Meta's attribution window over-credits view-through and early touches in a long SaaS sales cycle. TikTok claims credit across a generous window too. Sum the three platforms' reported conversions and you will routinely exceed your actual CRM conversions by 20-50% — they are double-counting the same deals. If you allocate budget on those raw numbers, you systematically over-fund demand-capture and under-fund demand-creation. The formula in this guide runs on incrementality-adjusted CAC for exactly this reason: it is the only input that reflects what each channel actually contributes.
Why SaaS needs a budget formula, not gut feel
Three realities make formula-driven allocation worth the effort for SaaS specifically.
The cost of being wrong scales with budget. At €20k/month, a sloppy split wastes a manageable amount. At €100k+/month, a fifteen-point misallocation is six figures annually — real money that could fund headcount or product. As SaaS paid budgets grow, the precision of the split becomes a material lever on growth efficiency, and intuition does not scale with the stakes.
SaaS economics are unforgiving and specific. Subscription businesses live and die by CAC payback and LTV-to-CAC ratios. A channel that looks cheap on cost-per-lead can be expensive on cost-per-retained-customer if its leads churn. The allocation has to be governed by downstream subscription metrics, not front-end ad metrics, and that requires connecting channel spend to CRM revenue and retention — a calculation, not a feeling. Our CAC payback by vertical analysis shows how much these benchmarks vary.
The three channels do genuinely different jobs. Google captures existing demand, Meta and TikTok largely create it. Funding them as if they were interchangeable — comparing their last-click CAC head to head — misunderstands their roles and leads to chronically underfunding demand creation, which then starves demand capture six months later. The formula has to account for what each channel does in the funnel, not just what it costs.
Attribution actively misleads. As covered in the callout, platform-reported numbers are biased in different directions across the three channels. A formula that ingests incrementality-adjusted inputs corrects for this; gut feel anchored on dashboards amplifies it. The discipline of running the calculation is partly a discipline of not being fooled by attribution.
The combined case is straightforward: the stakes are high, the right answer depends on subscription metrics and channel roles that intuition handles poorly, and the most readily available inputs (platform reports) are systematically misleading. A formula does not remove judgment — stage, constraints, and strategy are judgment calls — but it forces the judgment to operate on corrected data within explicit bounds, which is far better than allocating on vibes. The rest of this guide builds that formula and shows it working across company stages.
The three channels and what each does for SaaS
Before allocating, be precise about each channel's role. They are not substitutes; they occupy different funnel positions with different economics.
Google is demand capture. When someone searches for your category or your brand, Google lets you intercept that existing intent. This is why Google typically delivers the lowest CAC and fastest payback for SaaS — you are harvesting demand that already exists. Its ceiling is search volume: once you capture the available high-intent queries, additional Google spend buys progressively lower-quality clicks at rising CAC. Google scales beautifully until it caps, and then it caps hard.
Meta is demand creation and retargeting. Meta reaches your ICP before they are searching, building awareness and intent, and retargets warm audiences toward conversion. Its CAC is higher and payback slower than Google's because it works earlier in the funnel, but it has the inventory to keep growing where Google plateaus. For SaaS scaling past the point where Google demand caps, Meta is the channel that creates the new demand Google will later capture. The post-iOS measurement shifts make incrementality especially important here — see our Meta iOS post-ATT strategy guide.
TikTok is discovery and younger-ICP demand creation. TikTok reaches audiences through discovery rather than intent or social graph, and skews younger. For SaaS with broad or younger ICPs, self-serve or PLG motions, and the ability to produce native non-corporate video, it is a genuine demand-generation channel. For high-ACV enterprise software sold to senior buyers, it usually does not fit. TikTok's economics are the most variable of the three and the most dependent on ICP fit and creative quality, which is why it is often best introduced as a testing-reserve allocation first.
The allocation implication. Because the channels do different jobs, the formula cannot simply rank them by last-click CAC and fund the cheapest — that would always favor Google and chronically starve demand creation. Instead it weights each by incrementality-adjusted efficiency within a stage-appropriate balance of capture and creation. Understanding these roles is the prerequisite for the formula in the next section.
The core allocation formula
The framework is deliberately simple enough to run in a spreadsheet and rigorous enough to defend to a CFO. It has four moves.
Step one — carve out the testing reserve. Before allocating anything to proven channels, ring-fence 10-20% of total budget as a testing reserve for new channels, audiences, creative, and incrementality studies. The remaining 80-90% is your core allocation pool. The reserve is not leftover budget; it is a deliberate bet on discovering the next efficient channel before competitors and avoiding over-concentration. We return to it in section seven.
Step two — compute incrementality-adjusted CAC per channel. For each channel, take platform-reported CAC and divide by that channel's incrementality factor (its true incremental contribution divided by its platform-reported contribution, from section six). This converts misleading platform CAC into the adjusted CAC that reflects reality. A channel that reports €100 CAC but has a 0.6 incrementality factor has an adjusted CAC of €167 — and that is the number the formula uses.
Step three — compute payback efficiency relative to target. For each channel, compare its incrementality-adjusted CAC payback to your stage-appropriate payback target. A channel comfortably inside target is efficient and earns more weight; a channel exceeding target is inefficient and earns less, or gets fixed at a floor. This is where the formula encodes the discipline that you fund channels by how well they actually pay back, not by how cheap they look.
Step four — allocate the core pool by efficiency within stage bounds. Distribute the core pool across channels in proportion to their payback efficiency, but constrained by stage-based bounds (section five) that keep the capture-versus-creation balance appropriate to your company stage. The bounds prevent the formula from, say, dumping everything into Google for a scale-stage company that needs demand creation to keep growing.
The formula does not tell you to fund the cheapest channel — it tells you to fund the channel that pays back best on incrementality-adjusted numbers, within bounds that keep you investing in demand creation before your demand capture caps out. Companies that allocate purely to the lowest last-click CAC always over-index on Google, win the short-term payback, and then stall when search demand runs out and they have built no demand-creation engine to refill the funnel.
The output. The result is a target percentage per channel that is the product of corrected data (incrementality-adjusted CAC), an explicit objective (payback target), and strategic constraints (stage bounds and the testing reserve). It is defensible because every number traces to a source, and it is adaptive because re-running it each quarter with fresh inputs naturally evolves the split. The next three sections detail the inputs — CAC and LTV, stage bounds, and incrementality — that make the formula trustworthy.
CAC, LTV, and payback as the inputs
The formula is only as good as its inputs, and for SaaS the inputs are subscription metrics, not ad-platform metrics.
CAC must be fully loaded and channel-attributed. Customer acquisition cost for the formula means total spend on a channel divided by the customers it actually acquired — reconciled to the CRM, not the platform's conversion count. Fully loaded CAC ideally includes the relevant portion of creative and management cost, not just media. Channel-attributed CAC requires connecting spend to closed-won customers, which is why a tracking foundation that ties clicks to revenue matters so much; the BigQuery data pipeline tutorial describes building exactly that join at scale.
LTV reveals which channels deliver durable customers. Two channels can show identical CAC while one delivers customers who retain for years and the other delivers customers who churn in months. LTV by acquisition channel exposes this, and it should temper the allocation: a channel with slightly higher CAC but materially higher LTV (and thus a better LTV-to-CAC ratio) deserves more weight than CAC alone suggests. For SaaS, where retention is everything, ignoring LTV in allocation is a serious error.
Payback is the governing constraint. CAC payback — how many months of subscription revenue it takes to recover the acquisition cost — is the metric finance cares about most because it determines cash efficiency and runway. The formula uses payback relative to a target as the efficiency measure precisely because it captures both the cost (CAC) and the value velocity (how fast the customer pays it back). A channel can have acceptable CAC but unacceptable payback if its customers monetize slowly.
Reconciliation discipline. The recurring theme is that all these inputs must be reconciled to actual business outcomes, not lifted from platform dashboards. The single most valuable analytical investment a SaaS growth team can make is connecting channel spend to CRM revenue and retention, because it turns CAC, LTV, and payback from estimates into facts the formula can trust. Without that reconciliation, you are running a precise formula on imprecise inputs, which is just sophisticated guessing.
The inputs, properly measured, are what make the allocation defensible. Garbage inputs make even the best framework worthless; reconciled, incrementality-adjusted subscription metrics make a simple formula powerful.
Adjusting the split by company stage
The same formula produces very different splits at different company stages, because stage changes both the payback tolerance and the balance of demand capture versus creation the business needs.
Pre-PMF concentrates on capture. With limited budget and short runway, early-stage SaaS should harvest the cheapest, highest-intent demand — which means Google. Demand creation is a luxury when every euro must pay back fast to extend runway. Meta and TikTok get minimal allocation, reserved mostly for learning. The payback target here is short and strict.
Scale stage introduces creation. This is where most SaaS get allocation wrong by continuing to pour budget into Google because its last-click numbers look great. The reality is that Google's CAC rises as you exhaust high-intent keywords, and without a demand-creation engine you hit a growth ceiling. The formula, with a slightly relaxed payback target appropriate to the stage, naturally shifts meaningful budget into Meta to create demand, plus a ring-fenced TikTok test. Companies that delay this transition stall around the point where Google demand caps.
Enterprise stage tilts to creation. At €10M+ ARR with multiple products and segments, existing search demand is largely already captured, and net-new growth must come from creating awareness for new offerings and audiences. The payback tolerance is highest here (supported by strong net revenue retention), and the split tilts toward Meta, TikTok, and other demand-generation channels. Google remains important for capture but is no longer the growth engine.
ICP overrides stage. Company stage sets the baseline, but ICP can override it. A broad-ICP or PLG product tolerates more Meta and TikTok at any stage because those channels reach its audience efficiently. A narrow high-ACV enterprise ICP tilts capture-heavy and may zero out TikTok regardless of stage, because the buyers simply are not there. The formula's stage bounds should be set with ICP in mind, not mechanically by ARR alone.
The principle: stage and ICP set the bounds within which the efficiency-driven formula allocates, ensuring the split evolves from capture-dominant to creation-balanced as the company grows and as its audience dictates.
Incrementality-informed weighting
Incrementality is the input that makes the whole formula trustworthy, because it is the only way to know each channel's true contribution rather than the contribution it claims.
The attribution problem, restated. All three platforms over-credit themselves, in different directions and magnitudes. Google over-credits branded and high-intent clicks; Meta over-credits view-through and early touches across its window; TikTok claims credit generously too. Because the biases differ, you cannot even correct them with a single blanket adjustment — each channel needs its own incrementality factor. Allocating on uncorrected platform numbers does not just inflate totals; it distorts the relative weighting between channels, which is exactly what the formula is trying to get right.
Geo-holdout testing is the accessible method. The most practical way for most SaaS to measure incrementality is the geo-holdout: turn a channel off in a set of matched regions for a period, keep it running in comparable control regions, and measure the difference in total conversions across all channels. The drop in the holdout regions, normalized for size, estimates that channel's incremental contribution. Run it per channel and you get the incrementality factor each one needs. Our incrementality testing guide and cross-channel attribution guide detail the design.
Marketing mix modeling for larger budgets. Above roughly €100k/month combined spend, marketing mix modeling becomes worthwhile — statistical regression on historical spend and outcomes that estimates each channel's contribution without turning anything off. It complements geo-holdouts and is well suited to disentangling three channels running simultaneously. Google's open-source Meridian framework, covered in our Meridian MMM guide, is built for exactly this.
Converting reads into the formula input. Each incrementality read becomes a factor: true incremental contribution divided by platform-reported contribution. A factor of 0.6 means the channel actually drove 60% of what it claimed. Divide platform CAC by the factor to get incrementality-adjusted CAC. Typical patterns put Google's factor highest (closest to deterministic capture) and Meta's and TikTok's lower (more demand-creation, more attribution inflation) — but measure your own; assumed factors defeat the purpose.
Re-measure on a cadence. Incrementality is not static — it shifts with creative, competition, and channel saturation. Re-running geo-holdouts each quarter (rotating which channel you test) keeps the factors current so the formula stays honest. Stale incrementality factors quietly corrupt the allocation over time.
Incrementality weighting is what elevates this framework above allocating on dashboards. It is more work than reading platform reports, and it is the single highest-value input in the entire formula.
Reallocation triggers and the testing reserve
A budget split is not a one-time decision but a system that responds to change. Two mechanisms keep it alive: reallocation triggers and the testing reserve.
Quarterly cadence with trigger-based exceptions. Strategic reallocation happens quarterly — frequent enough to capture real change, infrequent enough to avoid whipsawing on noise and fighting platform learning phases. Between quarters, you monitor a set of trigger metrics and reallocate off-cycle only when one breaches a predefined threshold. This combination gives stability by default and responsiveness when something genuinely shifts.
The triggers that warrant action:
- A channel's incrementality-adjusted CAC breaching a multiple of target (for example, 1.5x) signals it needs reduction or investigation before further spend.
- Blended CAC rising beyond a set percentage quarter-over-quarter signals something broke and scale increases should pause until diagnosed.
- Creative fatigue indicators (rising CPM with flat or falling CTR on Meta or TikTok) signal a creative refresh, not necessarily a budget cut.
- Impression-share signals on Google (high share lost to budget) signal capture headroom worth funding.
The discipline is to distinguish fixable problems (creative fatigue) from allocation problems (a channel genuinely past its efficient point) and respond appropriately rather than reflexively moving budget.
Phase every reallocation. When the formula or a trigger calls for a shift, move no more than 25% of budget between channels in a single step. All three platforms penalize abrupt changes through learning-phase resets, so a large shift staged over several weeks with observation windows outperforms a single dramatic move. Document predicted versus actual impact at each step to refine the model. Our budget pacing guide covers the pacing mechanics.
The testing reserve is strategic, not leftover. The 10-20% reserve carved out before core allocation funds the experiments that find your next efficient channel: a first TikTok test for a SaaS that has only run Google and Meta, new audience and creative tests, and incrementality studies. It is measured by learning and incrementality rather than immediate payback, and it has a clear graduation path — a channel or tactic that proves out moves from the reserve into the core formula-driven allocation, while one that fails is retired. Without a reserve, you optimize the present at the cost of discovering the future, and you over-concentrate in channels that will eventually cap or fatigue.
Triggers and the reserve together turn a static split into a living allocation system — stable where it should be, responsive where it must be, and always probing for the next efficient channel.
Worked examples across stages
The framework is clearest applied to concrete cases. Three worked examples show the formula producing different splits from the same logic.
Example one — pre-PMF SaaS, €15k/month. Stage dictates a strict, short payback target and capture-heavy bounds. After ring-fencing a 15% reserve (€2,250) for learning, the core €12,750 goes overwhelmingly to Google, which delivers the lowest incrementality-adjusted CAC and fastest payback. Meta gets a small allocation for retargeting warm traffic; TikTok is mostly untouched beyond a tiny experiment in the reserve. Indicative result: roughly 85% Google, 12% Meta, 3% TikTok. The logic: protect runway by harvesting cheap intent, defer demand creation until there is budget and proven retention to justify slower payback.
Example two — scale-stage SaaS, €80k/month. Google's incrementality-adjusted CAC is rising as high-intent keywords saturate, while Meta's adjusted payback now sits within the stage's relaxed target. The formula shifts meaningful budget into Meta to create demand, funds a ring-fenced TikTok test from the reserve, and keeps Google as the efficient capture base. After a 15% reserve (€12,000), the core €68,000 splits to roughly 50% Google, 35% Meta, 15% TikTok. The logic: Google is capping, so fund the demand-creation engine that will refill the funnel, validated by incrementality reads showing Meta's true contribution justifies the weight.
Example three — enterprise multi-product SaaS, €300k/month. Existing search demand is largely captured, net revenue retention above 120% supports a longer payback tolerance, and growth requires creating awareness for new product lines. The formula tilts toward demand creation: Meta and TikTok carry more weight, Google remains a strong capture channel but not the growth driver. After a 20% reserve (€60,000) funding new-channel and new-segment tests, the core €240,000 splits to roughly 35% Google, 40% Meta, 25% TikTok. The logic: at this scale you grow by creating new demand across channels, and strong retention economics permit the slower payback that demand creation entails.
The common thread. In all three cases the formula is identical — reserve, incrementality-adjusted CAC, payback efficiency, stage bounds — but the inputs and constraints differ, producing splits that range from capture-dominant to creation-balanced. That is the point: a single defensible framework adapts to any SaaS context, and re-running it quarterly evolves the split as the company moves between stages and as channel economics shift.
For the deeper two-channel cut and the cross-vertical view, see our Meta vs Google budget allocation for SaaS guide and the omnichannel coordination guide for Google, Meta, and TikTok.
If you want AI-driven optimization for the Google half of your stack so your team can spend more cycles on demand-creation creative and incrementality testing, SteerAds runs a free 14-day audit on Google and Microsoft Ads accounts.
Sources
- openviewpartners.com — OpenView SaaS benchmarks on CAC, payback, and growth
- thinkwithgoogle.com — Think with Google measurement and budget insights
- facebook.com/business — Meta Business on incrementality and Advantage+
- tiktok.com/business — TikTok for Business advertising resources
- klipfolio.com — SaaS CAC and LTV reference metrics
FAQ
What is a sensible default budget split across Google, Meta, and TikTok for SaaS?
For most mid-market B2B SaaS, a reasonable starting point is roughly 50% Google, 35% Meta, 15% TikTok — Google captures existing high-intent demand, Meta creates and retargets demand against your ICP, and TikTok is a smaller demand-generation and testing allocation. But this is only a default to be replaced quickly by formula-driven allocation based on your own CAC, payback, and incrementality data. The split shifts heavily with company stage and ICP: pre-PMF leans almost entirely on Google, broad-ICP or PLG products tolerate far more Meta and TikTok, and narrow enterprise ICPs may drop TikTok entirely. Treat the default as a week-one placeholder, not a destination.
Should B2B SaaS even run TikTok ads in 2026?
It depends on ICP and motion. TikTok works for B2B SaaS when the buyer skews younger, the product has a self-serve or PLG motion, and you can produce native, non-corporate video — think productivity, design, marketing, and developer tools with broad appeal. It rarely pencils out for high-ACV enterprise software sold to senior IT buyers through long sales cycles. The right approach is to treat TikTok as a testing-reserve allocation first: commit a small, ring-fenced percentage, measure incrementality and payback against your benchmarks, and scale only if the data justifies it. Do not force TikTok into an ICP that does not live there.
How does CAC payback drive the budget split?
CAC payback is the governor on how aggressively you can fund longer-funnel channels. Google-sourced SaaS leads typically pay back fastest because they capture existing intent; Meta and TikTok create demand and pay back more slowly. The formula weights each channel by its blended-CAC-payback efficiency relative to your target: channels comfortably inside your payback target earn more budget, channels exceeding it get reduced or fixed at a testing allocation. The key discipline is comparing incrementality-adjusted CAC, not platform-reported CAC, because all three platforms over-attribute to themselves in different ways and raw numbers will mislead the allocation.
How often should I rebalance the split across the three channels?
Quarterly is the right cadence for strategic reallocation, with monthly monitoring against trigger thresholds. Monthly strategic shifts create whipsaw on noise and fight the platforms' learning phases; quarterly captures real change. Between quarters, watch trigger metrics — blended CAC, per-channel incrementality-adjusted CAC, creative fatigue, and impression-share signals — and act only when one breaches a predefined threshold. Most well-run SaaS accounts shift five to ten percentage points per quarter; a shift above fifteen points usually signals something broke (creative fatigue, tracking issue, or a competitive shock) rather than a genuine strategic recalibration.
Why use incrementality instead of platform-reported numbers for allocation?
Because every platform systematically over-credits itself, and the biases differ by channel, so allocating on raw platform numbers misfunds your channels. Google's models tend to over-credit branded and high-intent clicks that would have converted anyway; Meta's attribution window over-credits view-through and early touches; TikTok similarly claims credit across a generous window. Summed platform conversions routinely exceed actual CRM conversions by 20-50% because they double-count. Geo-holdout tests and marketing mix modeling recover each channel's true incremental contribution, and applying that as an incrementality factor to platform CAC gives you the adjusted CAC the formula should actually use. Without it, you over-fund demand-capture and under-fund demand-creation.
How should the split change as a SaaS company scales?
It shifts from demand-capture toward demand-creation as the company grows. Pre-PMF (sub-€1M ARR) concentrates almost entirely on Google to capture early high-intent demand cheaply and protect runway. Scale stage (€1-10M ARR) introduces meaningful Meta and a TikTok test as Google demand starts capping and you need to create demand. Enterprise and multi-product stage (€10M+ ARR) pushes further toward Meta, TikTok, and other demand-generation channels because existing search demand is largely already captured and net-new growth must come from creating awareness for new products and segments. The formula encodes this by adjusting target payback tolerance and channel weights by stage.
What share of budget should I reserve for testing new channels and creative?
A practical guideline is to ring-fence 10-20% of paid budget as a testing reserve, separate from your proven-channel allocation. This reserve funds new-channel experiments (such as a first TikTok test), new audience and creative tests, and incrementality studies, without destabilizing the core allocation that drives current pipeline. The reserve is what lets you discover the next efficient channel before competitors and avoid over-concentrating in channels that will eventually fatigue or cap. Treat reserve spend as bets measured by learning and incrementality, with a clear graduation path: a channel or tactic that proves out moves from the reserve into the core formula-driven allocation.