Budget Decisions

At a glance

Every budget decision is one of three: increase, hold or stop. Decide on the downstream result (activated or paid trialsqualified demos and pipeline), not on clicks or raw form fills. Wait long enough to see the signal: judge after enough clicks and conversions for your budget band, not after a bad day. Change budgets in steps of about 20 to 30% (smaller for big accounts), then wait one to two weeks before the next step. Lost impression share (budget) says more money would buy more of the same; lost impression share (rank) says money isn’t the constraint. Since August 2026 a Target CPA or ROAS on a budget-limited campaign is a setpoint Google aims to hit, not a ceiling. Set targets to what you actually want to pay. As of October 2026

What counts as a result

Google Ads counts what you told it to count. Your budget decisions need one step further down the funnel, from your CRM:

Level Metric Use it for

Early Trial sign-ups, cost per trial Week-one sanity checks; what Google optimises at first.

Decision Activated trials (users who reached your key action), cost per activated trial Increase / hold / stop decisions after the waiting period.

Final Trial-to-paid conversions, new revenue per ad dollar Quarterly budget allocation between campaigns and regions.

Level Metric Use it for

Early Demo requests, cost per request Week-one sanity checks; what Google optimises at first.

Decision Sales-qualified demos (SQLs), cost per SQL Increase / hold / stop decisions after the waiting period.

Final Pipeline created and closed-won revenue per ad dollar Quarterly budget allocation between campaigns and regions.

Work out your target cost per decision metric before launch: what you can afford to pay for one activated trialSQL given your win rate and deal value. A common starting rule is to spend no more than a third of the expected value of that lead, so a target return of 3×. The geo bid calculator does the arithmetic per region.

Increase, hold or stop: the signals

Thresholds below are starting rules of thumb for your budget band. Adjust them once you know your own sales cycle.

Starter ($5,000/month or less) Wait before judging: three to four weeks, or about 300 clicks, whichever comes later. With only a few conversions a month, one deal more or less swings everything.

Increase (+20 to 30%)

Cost per activated trialSQL at or below target over the last 30 days. Search lost IS (budget) above about 20%. Search terms are clean and on-topic.

Stop or restructure

After about 300 clicks: no qualified result at all, and search terms and landing page have already been fixed. Cost per qualified result above 2× target for four straight weeks. Leads are mostly junk (students, wrong company size, wrong countries) after negatives and exclusions.

Hold in every other case. Holding is the right answer most weeks on a small budget. Use the time to tighten negatives and test one ad or page change. Before stopping, try one structural fix: narrower keywords (exact match on the highest-intent terms), a clearer landing page, or a different offer. If that doesn’t work in another three to four weeks, stop and put the money elsewhere.

Intermediate (about $5,000 to $50,000/month) Wait before judging: two to three weeks and at least 30 conversions per campaign. After any big change (bid strategy, target, budget step over 30%) give Smart Bidding a week before reading results.

Increase (+15 to 25% per step)

Cost per activated trialSQL at or under target for two consecutive review periods. Search lost IS (budget) above 10 to 15%. Lead quality steady: share of leads qualifying hasn’t dropped.

Cut or stop

Cost per qualified result above 1.5× target for four weeks: cut budget 30 to 50% or tighten the target. Above 2× target for six weeks after fixes: stop and reallocate. Qualification rate falling while lead volume rises (a sign of broadening into poor traffic).

Hold when results are within about 20% of target, or when lost IS is mostly from rank. Reallocate monthly: move budget from the campaign with the worst cost per qualified result to the one with the best, as long as the receiving one has lost IS (budget) to absorb it.

Advanced (above $50,000/month) Wait before judging: one to two weeks for volume and CPA; four to six weeks (or your typical lead-to-opportunity time) for pipeline. Use campaign experiments for anything bigger than a budget step.

Increase (+10 to 20% per step)

Pipeline (or value) per ad dollar at or above target for four or more weeks. Search lost IS (budget) above 10%, or a value-based campaign is consistently limited by budget. Marginal returns hold: after each step, cost per SQL rose less than about 15%.

Cut or stop

Marginal cost per SQL after the last step more than 1.5× the campaign average: step back. Pipeline per dollar under target for a full sales-cycle window after fixes. Experiment shows no incremental lift (for example, a brand campaign cannibalising organic clicks).

Hold while an experiment is running, during the week after a big change, and when the CRM data for the period isn’t mature yet. At this spend, plan budget against marginal returns, not averages: the last $10,000 usually buys less than the first.

Reading impression share

Add three columns to your campaign view: Search impr. share, Search lost IS (budget) and Search lost IS (rank). Together they tell you whether more money would help.

What you see What it means What to do

High lost IS (budget), results on target Demand you’re not buying because the budget runs out. Increase budget in steps.

High lost IS (budget), results off target You’re spending all you have on traffic that doesn’t convert well. Don’t add budget. Narrow keywords, fix pages, then reconsider.

High lost IS (rank), low lost IS (budget) Budget isn’t the limit. You’re losing auctions on bids or quality. Improve ads and landing pages; loosen the target or raise bids if economics allow.

Both low, impression share high You’re close to the ceiling for these keywords. Growth needs new keywords, regions or campaign types, not more budget here.

Budget pacing and seasonality

How daily budgets actually spend

The budget you set is an average daily budget. On busy days Google can spend up to twice that amount; over a calendar month you are charged at most 30.4 times the daily budget. So a campaign that spends 180% of its daily budget on a Tuesday is working as designed. Judge spend by week and month, not by day. Since 1 June 2026 this also holds for campaigns with an ad schedule that switches off on some days of the week: Google now aims to spend the full 30.4 times the daily budget over the days the campaign runs. As of November 2026

Seasonality in B2B

B2B demand follows budgets and calendars. Expect conversion rate and search volume to move around these points, and don't judge a campaign on a week that contains one:

End of quarter and fiscal year: buyers with budget to spend move faster; others freeze purchases until the new year. Late December to early January: volume and conversion rate drop in most markets. Industry events and your own launches: a spike in brand searches is not a sign the generic campaign got better.

EU/UK August is quiet across much of Europe, especially France, Italy and Spain. Lower volume then is normal, not a failing campaign.

India Diwali week and the March fiscal year-end change B2B buying activity. Many companies decide purchases in February and March.

Rest of APAC Lunar New Year slows activity in China-facing and Southeast Asian markets for one to two weeks. Japan's fiscal year ends in March, and Golden Week in early May is quiet.

Latin America Carnival (February or March) and the December to January summer holidays in the southern countries are slow periods.

Middle East/Africa During Ramadan working hours are shorter in many Gulf countries, and the weeks around Eid are slow. South Africa's December holidays are long.

US/Canada The weeks of US Thanksgiving and the end of December are slow; so is the July 4 week. Canadian and US holidays differ, which matters if one campaign covers both.

Two things to do about it:

Compare like with like: compare a holiday week with the same week last year, or skip it, rather than with the week before. Smart Bidding and short events: Smart Bidding adjusts to normal seasonality on its own. For a short, predictable change of a few days (a launch-day promotion, a big event), Google offers seasonality adjustments in the bid strategy tools to tell it the conversion rate will jump. Don't use them for whole seasons or holidays. As of November 2026

How to change a budget

Step size: 20 to 30% at a time on Starter, 15 to 25% on Intermediate, 10 to 20% on Advanced. Bigger jumps can make Smart Bidding chase expensive auctions while it recalibrates. Spacing: one to two weeks between steps, so each step has its own read. One change at a time: don’t change budget, target and keywords in the same week. Daily vs monthly: Google can spend up to twice the daily budget on a given day but won’t charge more than about 30.4 × daily budget in a month. Set the daily budget from the monthly plan and don’t panic about single days. Shared budgets are fine for small, similar campaigns but make it impossible to protect a high-value campaign. Give your best campaign its own budget. Pausing: pausing for a few days is fine. Pausing for weeks resets some of what Smart Bidding learned; prefer cutting budget.

Targets are now setpoints

As of October 2026

Since 17 August 2026, Google treats Target CPA and Target ROAS (including the optional targets in Maximize conversions and Maximize conversion value) as a level to deliver close to, not just a limit. In budget-limited campaigns that used to beat their target, CPA tends to rise toward the target and volume rises with it. Google also relabelled the strategies in June 2026 (“Maximize conversions with a target CPA” is now shown as “Target CPA”). What this means for you:

Set the target to the CPA you actually want to pay, not an aspirational ceiling. If you see CPA drift up toward target after August 2026 with no other change, that’s this update, not a performance problem. Lower the target if the old CPA was better for you. If you want the most conversions for a fixed budget and no target, use Maximize conversions without one.

Go deeper Why marginal returns fall as you spend more

Your first dollars go to the best auctions: the most specific searches, at times and places where you convert best. As budget rises, Google has to buy auctions it previously skipped, which convert worse or cost more. So your average CPA can look fine while the last slice of spend is poor. To see it, compare two periods: if spend rose 25% and conversions rose 10%, the extra spend bought conversions at 2.5× the old average. That marginal CPA is the number to compare with your target when deciding the next step.

Do this now

Pacing check run mid-month; daily budget set to finish the month on target. Slow weeks for your region marked in the calendar so you don't judge campaigns on them.

Write down your decision metric and its target (cost per activated trialSQL). Add Search impr. share, Search lost IS (budget) and Search lost IS (rank) columns to your campaign view. Put the first judging checkpoint in the calendar, based on your band’s waiting period. Check every Target CPA/ROAS is set to what you actually want to pay (post-August 2026 behaviour). Run the monthly increase / hold / stop prompt and record the decisions.

Next: Monitoring With an LLM