Budget Pacing: How to Manage Ad Spend Across Platforms Without Overspending

"the platform overspent" is usually a pacing model that was never written down meeting a delivery rule that was always documented. google, meta, and tiktok each publish exactly how far above a daily budget they will go and over what period they will even it out. most teams have not read those pages, so they set daily budgets by dividing the monthly number by 30, react to a heavy day by cutting, and reset learning on the way. this guide is the rule set, the math, and the checks.

---

what budget pacing actually is

pacing is the relationship between three numbers: the amount you committed to spend over a period, the amount spent so far, and the amount the platforms will spend if you touch nothing. everything else, including daily budgets, is an instrument.

what most guides get wrong is treating the daily budget as a ceiling. on every major platform the daily budget is an average the delivery system is allowed to exceed on a given day and required to hit on average over a longer window. the ceilings are elsewhere:

so a $1,000 daily budget on google can bill $2,000 on a tuesday and $30,400 in the month. a $1,000 daily budget on meta can bill $1,750 on a tuesday and $7,000 sunday through saturday. neither is a bug. both are the documented product.

---

the fundamental shift: from daily ceilings to period budgets the platform paces for you

before: advertisers set daily budgets as hard limits and used them as the pacing instrument. cut the daily budget to slow down, raise it to speed up, every day if needed. the platforms roughly obeyed.

after: every platform now optimizes delivery within a period and treats the daily number as guidance. google paces to the calendar month and explicitly allows 2x days. meta paces to the week and allows 75% over. tiktok paces a lifetime budget unevenly across days by design and pauses delivery when the allocated budget is exhausted. on top of that, bid strategies designed to spend the whole budget (maximize conversions, maximize conversion value) are, in google's words, "limited by budget by design" (google ads help, maximize conversions).

the consequence is that daily budget edits are no longer a free pacing lever. each edit changes what the platform paces toward for the rest of its period, and a large enough edit restarts learning. pacing now happens at two levels: your monthly model, and the platform's period model. when they disagree, the platform wins on the day and you win only by changing the budget, which costs you learning.

---

what this broke in existing accounts

1. monthly plans built on monthly divided by 30

a $30,000 monthly plan set as $1,000 per day on google is a $30,400 cap. on a 31-day month the plan is short by a day of spend and the cap is over by $400. on meta the same plan is $7,000 per week, which is $30,333 in a 4.33-week month and $31,000 in a five-saturday month. small, but the finance team sees the difference and calls it an overspend.

2. cutting budgets on a heavy day

a 2x day on google is the system taking traffic it expects to pay back later in the month. a team that cuts the daily budget the next morning locks in the heavy day and removes the light days that would have offset it. google then paces the remainder of the month to the new, lower daily budget times the remaining calendar days (google ads help, about spending limits). the month ends under plan, and the campaign spent two weeks in learning.

3. multiple budget edits in one day

google sets the daily spending limit for a day to the highest average daily budget chosen that day. its own example: a $10 budget raised to $50 and lowered to $5 on the same day yields a $100 limit for that day (google ads help, how budget changes take effect). teams that "tested" a budget raise for an hour paid for a full day at the raised limit.

4. meta budget edits that reset learning

meta's learning phase ends after about 50 results in the week after the ad set's last significant edit, and meta warns that "frequent budget changes" can cause an ad set to re-enter learning (meta business help center, about the learning phase). meta does not publish a percentage threshold; it describes budget changes as significant depending on magnitude, with a $100 to $101 change unlikely to reset learning and $100 to $1,000 likely to (meta business help center, significant edits). the practitioner rule of 20% per step is a convention derived from that language, not a documented number. accounts that pace meta by editing ad set budgets daily are permanently in learning.

5. tiktok flights that front-loaded or stalled

a lifetime budget on tiktok is paced unevenly and stops when spent. a flight set with a lifetime budget and no daily floor can spend heavily in the first week around high-traffic days and go quiet at the end. and because you cannot switch a live campaign between daily and lifetime budgets, the fix is a new campaign, which is a new learning phase.

6. campaign budgets and ad set budgets mixed inside one pacing model

on meta, a campaign budget (advantage+ campaign budget) distributes spend across ad sets and the campaign is the pacing unit. an ad set budget makes the ad set the unit. an account with both, reported at the campaign level, shows campaign totals that hide which ad sets are actually consuming the month. the same is true of shared budgets on google.

---

the documented rules

no public dataset exists on how often accounts overspend a monthly plan or by how much. what follows is what the platforms document, which is the only thing a pacing model should be built on.

google ads.

• single day: up to 2x the average daily budget. calendar month: no more than 30.4x the average daily budget. you never pay more than the spending limits even if served cost exceeds them. (google ads help, overdelivery.)
• after a mid-month budget change, the rest of the month paces to the new average daily budget multiplied by the remaining calendar days, not the remaining scheduled days. (google ads help, about spending limits.)
• multiple same-day edits: the daily limit is based on the highest budget chosen that day. google's worked example: on november 24 with $103 spent and 7 days left, raising to $10 a day sets a new monthly cap of $173. (google ads help, how budget changes take effect.)
• ad schedules do not change the 30.4x monthly limit; the system paces to the full monthly limit across the active days. (google ads help, about spending limits.)
• learning status is triggered by a new or reactivated strategy, a setting change to the bid strategy, a composition change, or an ad group target change on shopping. budget changes are not on google's list of learning triggers (google ads help, learning period). the pacing consequence of a budget change is the recalculated monthly cap, not a learning reset.

meta.

• daily budget is an approximate amount per day over a week running sunday to saturday. on days with better opportunities meta may spend up to 75% over the daily budget. for every week ending saturday at midnight, spending will not exceed 7x the daily budget. (meta business help center, about daily budgets.)
• learning exits after about 50 results in the week after the last significant edit; frequent budget changes can re-enter learning; budget significance depends on magnitude. (meta business help center, learning phase and significant edits.)
• a campaign budget lets meta distribute spend across ad sets toward the best results; an ad set budget fixes each ad set's spend. the pacing unit follows the budget level.

tiktok.

• daily budget is the average per day; lifetime budget is the total for the flight, allocated unevenly across days but never exceeding the total, with delivery pausing when the allocated budget is exhausted. (tiktok ads manager help, about budget.)
• minimums: campaign daily and lifetime budgets above $50; ad group daily above $20; ad group lifetime at least the number of days times the minimum daily budget, so a 31-day ad group needs at least $620. (same.)
• budget type cannot be switched after launch. tiktok recommends limiting budget increases to 40% during the learning phase and 30% after it, no more often than every two days. (same.)
• learning volatility typically declines after about 25 results or 7 days from entering learning; pausing, edits that retrigger learning, and unreasonable budgets are the listed things to avoid (tiktok ads manager help, about learning phase).

---

what changed in 2026 specifically

google ads, august 17, 2026: budget-limited campaigns on target cpa and target roas now deliver to the stated target rather than beating it (google ads help, changes to target based bid strategies). for pacing this means a campaign that was under-spending because the target was loose will now spend to the target, and a campaign with a tight target will under-spend more visibly. google's suggested remedy is a bigger budget. the bidding guide covers when that is right.

meta, advantage+ campaign budget as the practical default: consolidated structures with campaign-level budgets exit learning faster because signal pools at the campaign. the pacing unit is the campaign; ad set spend inside it is meta's decision.

tiktok, smart+ and cbo: campaign budget optimization is the default on smart+ campaigns, so the ad group level is no longer where pacing happens for those campaigns.

all platforms: none of the overdelivery rules above changed in 2026. the mistakes did not change either.

---

what actually works now

1. run one monthly pacing model per platform, with the platform's own period inside it

for each platform and each brand:

• plan: the committed monthly amount, M.
• expected to date: M x d / D, where d is days elapsed and D is calendar days in the month.
• pace: spend to date divided by expected to date. 1.00 is on plan.
• projection: spend to date plus the trailing 7-day average daily spend times remaining days.

then set the platform budgets to the platform's period, not to M / 30. google: average daily budget = M / 30.4 is exact for the monthly cap. meta: daily budget = M / (7 x weeks in the month) where weeks is the number of saturdays; or use a lifetime budget on flighted campaigns. tiktok: for ongoing campaigns, daily budget = M / D; for flights, a lifetime budget and accept uneven days.

2. alert on the projection, not on the day

a 2x day on google or a 175% day on meta is inside the rules. the alert conditions that matter:

• pace outside 0.90 to 1.10 after day 7.
• projection outside plan by more than 5% after day 15.
• meta week-to-date spend above 7x the daily budget with a day remaining in the week, which indicates a budget edit mid-week rather than a delivery anomaly.
• google month-to-date above 30.4x the lowest daily budget that ran this month, which indicates a raise that reset the cap.

3. change budgets on the platform's clock and in steps

google: change once, expect the rest of the month to re-pace to new budget times remaining calendar days, never edit twice in a day. meta: change no more than once per week per ad set or campaign, in steps small enough not to be a significant edit; 20% is the working convention. tiktok: follow the documented 40% in learning, 30% after, no more often than every two days. write the step size into the brand's rules so an agent and a human follow the same one.

4. pick the budget level that matches the pacing unit

if the monthly plan is per campaign, use campaign budgets (advantage+ campaign budget on meta, cbo on tiktok, a single campaign or shared budget on google). if the plan is per audience or per test cell, use ad set or ad group budgets and accept that you own the distribution. do not mix within a campaign group that reports as one line.

5. handle month-end explicitly

from day 25, freeze budgets unless the projection is outside 5% of plan. if it is over, lower the daily budget by the exact amount the remaining-days formula needs and no more. if it is under, raise once, knowing google's cap for the rest of the month becomes new daily budget times remaining days and meta will still hold the week at 7x. never zero out a campaign for the last two days to hit a number; on the first of the month it re-enters learning.

6. separate pacing findings from tracking findings

a spend drop with a matching conversion drop is pacing. a conversion drop with steady spend is tracking. check spend first, every time. an agent that reports "conversions down 40%" without reading spend by day is generating false tracking alarms.

7. write the rules where the agent can read them

per brand: monthly plan per platform, allowed step size, maximum change per day, and the actions that always need a human. those four numbers are the whole pacing policy. orbit's onboarding audit checklist captures them at kickoff.

---

what to stop doing immediately

• stop setting google daily budgets as monthly divided by 30. it is 30.4.
• stop cutting a budget the morning after a 2x day on google or a 175% day on meta.
• stop editing a google budget more than once in a day.
• stop editing meta ad set budgets daily as a pacing lever.
• stop launching tiktok flights on a lifetime budget without a plan for uneven days.
• stop reporting campaign-level totals when the budgets live at the ad set level.
• stop calling a conversion drop a tracking problem before you have read spend by day.
• stop zeroing out campaigns for the last two days of the month.

---

the orbit agent instructions

{
  "guide": "budget-pacing-daily-management",
  "agent_tasks": {
    "monitor_month_to_date_pace": {
      "check": "each morning, for every platform with a monthly plan, pull spend by day for the month to date, compute pace as spend divided by plan times days elapsed over calendar days, and the projection as spend plus the trailing 7-day average times remaining days",
      "flag": "pace outside 0.90 to 1.10 after day 7, or projection outside plan by more than 5% after day 15",
      "recommend": "state the projected month-end number and the single budget change that brings it to plan using the platform's remaining-days formula, sized within the brand's step limit"
    },
    "monitor_overdelivery_within_rules": {
      "check": "compare yesterday's spend per campaign to its daily budget on google (2x limit) and per ad set or campaign on meta (1.75x limit), and compare meta week-to-date spend to 7x the daily budget",
      "flag": "any day above the platform's documented single-day multiple, or meta week-to-date above 7x with a day still remaining in the sunday-to-saturday week",
      "recommend": "if inside the rules, report it as normal pacing and do not propose a cut; if outside, look for a mid-period budget edit in change history before anything else"
    },
    "audit_budget_edit_cadence": {
      "check": "from change history, count budget edits per campaign or ad set over the last 30 days and the size of each edit as a percent of the prior budget",
      "flag": "more than one google budget edit on the same day, more than one meta budget edit per ad set per week, any meta or tiktok edit above the brand's step limit, or tiktok edits more often than every two days",
      "recommend": "consolidate edits to one per period at the documented step size and note that each large edit on meta or tiktok restarts learning"
    },
    "audit_budget_level_vs_pacing_unit": {
      "check": "for each campaign, identify whether the budget sits at the campaign or the ad set or ad group level and whether the brand's monthly plan is defined per campaign or per audience",
      "flag": "a campaign group reported as one plan line whose budgets are split across ad sets, or a mix of campaign and ad set budgets under one plan line",
      "recommend": "move the budget to the level that matches the plan line, or split the plan line to match the budget structure, before running pacing alerts on it"
    },
    "audit_budget_capped_efficient_campaigns": {
      "check": "list campaigns flagged limited by budget or whose spend equals the budget on 5 of the last 7 days, together with their cpa or roas versus the brand goal",
      "flag": "a budget-capped campaign beating its efficiency goal by more than 10% while the platform is under pace, or an uncapped campaign missing goal by more than 20% while the platform is over pace",
      "recommend": "propose a one-step reallocation within the brand's step limit from the inefficient campaign to the capped efficient one, and re-check after one period"
    },
    "monitor_month_end_projection": {
      "check": "from day 25, recompute the projection daily and list any budget edits made in the last 5 days of the month",
      "flag": "projection outside 5% of plan, or any campaign reduced to zero or paused in the final 3 days",
      "recommend": "make at most one final adjustment sized by the remaining-days formula, and never zero out a campaign to hit the number because it re-enters learning on the first of the month"
    }
  }
}

---

the bottom line

daily budgets are averages. google will spend 2x on a day and 30.4x in a month, meta will spend 75% over on a day and 7x in a week, tiktok will spend a lifetime budget unevenly. build the monthly model on those rules, alert on the projection instead of the day, change budgets once per platform period in documented steps, and read spend before you read conversions.

the platforms already pace. your job is to set the number they pace to and stop fighting them in the middle of their period.

---

built by orbit - the performance marketing intelligence layer. questions? jose@orbitllm.com