Ad Spend Approval Thresholds: What Needs a Human (2026)

Key Takeaways

By Kris Newlin. Updated October 27, 2026.

Ad spend approval thresholds are the numeric limits that decide which budget changes software can make on its own and which ones wait for a person. A workable default: changes up to 20% per day inside the approved plan run automatically, anything above that or above a fixed dollar amount needs the account lead, and new spend or new accounts need the client or CFO.

• Gate by type of change, not by tool: Routine pacing, reallocations and new spend carry different risk, so they get different approvers.
• Use two limits at once: A percentage cap per day and per week catches fast jumps, a dollar cap per account catches large absolute moves on big budgets.
• Scale the numbers to the account: A $300 change is noise on a $200,000 monthly account and half a week of spend on a $5,000 one.
• Name one approver per tier: Media buyer, account lead, client or CFO. “Someone on the team” is not an approver.
• Every request carries the same three facts: What changes, why, and the expected impact, plus a log entry the client can read later.
• An AI employee proposes, a human approves: Viktor drafts the change, posts it in Slack, and only spends after the right person says yes.

The short answer

Set ad spend approval thresholds in three tiers. Tier 1: pacing changes within the approved monthly budget and under 20% per day run without sign-off but get logged. Tier 2: reallocations between campaigns or channels, or any change over 20% per day or 40% per week, need the account lead. Tier 3: new budget, new accounts, new channels and anything above your dollar cap need the client or finance. Write the numbers down once, put them where the person or software making the change can read them, and review them every quarter.

Why ad budgets need a human gate

Ad platforms are built to spend. On Google Ads, a campaign can spend up to two times its average daily budget on a single day, and the monthly charge is capped at roughly 30.4 times the daily budget, according to Google’s own budget documentation. Meta’s delivery system also moves money between ad sets automatically when you use campaign budget. That flexibility is useful when a person is watching. It becomes a problem when automated rules, scripts or AI tools start raising budgets on top of it.

Three things go wrong without a gate:

1. Small errors compound. A rule that raises budget 25% every time ROAS beats target can double a campaign’s spend in three days. If the conversion data behind it is wrong (a broken pixel, a duplicate purchase event), the budget follows bad data.
2. The plan stops matching the invoice. Agencies sign off on a monthly media plan with the client. When spend drifts past it, the agency either eats the overage or has an uncomfortable conversation.
3. Learning resets cost money. On Meta, large budget edits count as significant edits and can push an ad set back into the learning phase, which Meta says needs about 50 optimization events within seven days to exit (Meta Business Help Center). Frequent big jumps keep campaigns unstable.

A human gate does not slow down good changes. It makes sure the large ones get one look from the person accountable for the money. For approvals in general, see how human in the loop AI approvals work. This page is about the numbers.

The three kinds of ad spend changes

Not every budget edit carries the same risk. Sorting changes into three types is the first step to setting sensible thresholds.

Routine pacing inside the plan

Pacing changes keep spend on track against a budget the client already approved. Examples: nudging a daily budget up 10% because the month is running behind, or trimming it because a campaign is ahead of plan. The total for the month does not change. These are low risk and should run without waiting for anyone, as long as they stay under the percentage cap and are logged. If you want the method for checking whether spend is on track, budget pacing covers it.

Reallocations between campaigns or channels

Reallocations move money from one place to another: from a prospecting campaign to retargeting, from Meta to Google, from one product line to another. The monthly total may stay the same, but the plan changes. A reallocation is a judgment call about what is working, so it needs someone who knows the client’s goals. Whether to pause a campaign or shift its budget is its own decision rule, covered in when to pause ads vs reallocate budget.

New spend, new accounts and big jumps

This is money the client has not approved yet: raising the monthly total, launching a new channel, opening a new ad account, or a single change large enough to alter the month’s outcome. These always need the person who owns the budget, which is usually the client at an agency or the CFO or finance lead in-house. Whether the business should scale spend at all is a separate question, answered with marketing efficiency ratio (MER) and margin targets, not with approval thresholds.

Setting the thresholds

Good thresholds use two limits together: a percentage limit that catches fast relative jumps, and an absolute dollar limit that catches large moves on big accounts.

Percentage limits per day and per week

A common practitioner rule of thumb on Meta is to raise budgets by no more than about 20% at a time, a few days apart, to keep delivery stable. That makes 20% per day a natural cap for changes that run without approval.

The weekly limit matters just as much. Four daily changes of 15% each stay under a 20% daily cap but add up to a 75% increase in a week. A weekly cap of 40% cumulative closes that gap.

Suggested starting points:

• Up to 20% per day and 40% per week, inside the monthly budget: no approval, logged.
• Over 20% per day or over 40% per week: account lead approves.
• Any change that would push the month over the approved total: client or CFO approves.

Decreases deserve a lighter gate than increases, but not zero. Cutting a campaign by 80% overnight can starve a launch or break a client commitment. A reasonable rule is to let decreases up to 30% run freely and send larger cuts to the account lead.

Absolute dollar limits per account

Percentages break on large accounts. Twenty percent of a $10,000 daily budget is $2,000 a day, which is $60,000 over a month if left running. That is too much money to move without a person looking at it, even if the percentage looks routine.

Add a dollar cap per account per day, for example: no single change above $500 per day without the account lead, and no change above $2,500 per day without the client or CFO. Whichever limit is hit first, percentage or dollars, decides the approver.

Most ad platforms also offer hard spending limits you can set as a backstop. Meta has an account spending limit that stops delivery once a total is reached, and Google Ads supports account-level budgets for accounts on monthly invoicing. Thresholds in your policy decide who approves a change. Platform limits make sure a missed approval cannot burn through the account.

Thresholds by account size (small, mid, large budgets)

The same numbers do not fit every account. Size them to monthly spend:

Small accounts get a looser percentage because the dollars are small, but they have little room for error, so the dollar caps stay tight. Large accounts get a tighter percentage because each point is real money. Treat these as starting values and adjust after a quarter of real data.

Who approves what

An approval threshold only works if each tier has one named person behind it.

Media buyer, account lead, client or CFO

• Media buyer (or automation): owns tier 1. Pacing inside the plan, small daily adjustments, pausing ads that break brand or policy rules. No approval, full logging.
• Account lead: owns tier 2. Reallocations between campaigns and channels, changes over the percentage or dollar cap, larger decreases. Usually responds within a few hours during working days.
• Client: owns tier 3 at an agency. Any increase to the monthly total, new channels, new accounts. Get this in writing, even if writing means a reply in a shared Slack channel.
• CFO or finance lead: owns tier 3 in-house, and often co-signs at agencies on large retainers. Finance cares about the monthly total and cash timing more than campaign detail, so their request should lead with the money.

Two practical rules help. First, give every tier a backup approver for holidays and sick days. Second, set a time limit: if a tier 2 request gets no answer within one business day, it escalates to the next approver rather than sitting idle while a good campaign stays capped.

A sample threshold policy (table)

Copy this into your agency playbook or client onboarding doc and change the numbers to fit.

The last row matters. Stopping spend should never wait for approval. Spending more should.

How the approval works in practice

The policy is only as good as the request format. If approving a change takes ten minutes of digging, people either rubber-stamp it or ignore it.

The request: what changed, why, expected impact

Every approval request should answer three questions in a few lines:

1. What changes: account, campaign, current budget, proposed budget, the difference in percent and dollars, start date.
2. Why: the data behind it, such as “CPA $38 vs $52 target over the last 7 days, 140 conversions, spend 18% behind plan.”
3. Expected impact: what happens to the month. “Adds about $4,200 to this month, keeps total within the $60,000 plan” or “takes total to $64,200, $4,200 over plan.”

A request that includes the threshold it crosses (“tier 2: 32% daily increase”) tells the approver why they are seeing it.

Approve, edit or reject in Slack

The approval should happen where the team already works. For most agencies and in-house teams that is Slack or Microsoft Teams, not a separate dashboard.

This is where Viktor fits. Viktor is an AI employee who lives in Slack and Microsoft Teams and connects to Meta Ads, Google Ads, LinkedIn Ads and 9,800+ other tools, most through one-click OAuth. He reads the threshold policy you give him, checks pacing and performance, and when a change crosses a limit he posts the request in the client’s channel with the three facts above and tags the right approver. The account lead can approve it, edit the number (“make it 20%, not 32%”) or reject it with a reason. Viktor only makes the change after a yes, and for tier 1 changes he still posts a short note so nothing moves silently. If you want to see it with your own accounts, try Viktor in your Slack.

If you are new to the idea, here is what an AI employee is and how that differs from rule-based automation: instead of a fixed if-then script, he explains each proposed change in plain language and waits for a person when the policy says so.

Logging every change for the client

Log every budget change, approved or not, in one place the client can read. Each entry needs:

• date and time
• account and campaign
• old and new budget
• who requested it and who approved it
• the reason given

A weekly summary of the log turns budget changes into a client conversation instead of a surprise on the invoice. It also gives you the data to tune thresholds: if account leads approve 95% of tier 2 requests without edits, the cap is probably too low. Viktor can keep this log in a Google Sheet or Notion page and post the weekly summary to the client channel. Set up your first approval workflow and run it alongside your current process for a couple of weeks before relying on it.

Mistakes that make thresholds useless

• Only a percentage cap. On large accounts a 20% change can be thousands of dollars a day. Always pair it with a dollar cap.
• No weekly limit. Several small daily changes add up. Track the cumulative change over seven days.
• Vague approvers. “The team” never approves anything. Name a person and a backup per tier.
• Requests without context. “Raise budget?” with no numbers forces the approver to research it or guess.
• Thresholds set once and forgotten. Accounts grow. Review the numbers every quarter and after any big change in monthly spend.
• Automation that bypasses the policy. Platform automated rules and scripts often sit outside the approval flow. Put every source of budget change under the same thresholds, or switch off the ones you cannot gate.
• No log. Without a record, nobody can tell a client why spend moved, and nobody can tell whether the thresholds are working.

FAQ

What is a good approval threshold for ad spend changes?

A good default is 20% per day and 40% per week for changes inside the approved monthly budget, with a dollar cap on top: around $500 per day for mid-sized accounts. Anything above those limits goes to the account lead, and anything that raises the monthly total goes to the client or CFO. Large accounts should use a tighter percentage, around 10%, because each point is more money.

Who should approve budget changes at an agency?

The media buyer handles pacing inside the plan without approval. The account lead approves reallocations and changes over the cap. The client approves any increase to the monthly budget, new channels and new accounts. On large retainers, the client’s finance lead or CFO often co-signs tier 3 changes. Each tier needs one named approver and a backup.

Can an AI employee change budgets without approval?

Only inside the limits you set. A well-configured AI employee like Viktor can make small pacing changes within the plan on his own and log them, but he proposes anything above the threshold in Slack or Teams and waits for the named approver before spending. Emergency pauses are the exception: stopping spend can happen right away, with a notification. If you want an AI employee who works this way in your own channels, start with Viktor here.