Skip to main content
MethodMarketingPaid advertisingIntermediate

Set ad budgets and choose a bid strategy

A budget and bidding plan for Google Ads and Meta: how much each campaign needs to learn, which bid strategy to start on, starting targets, how far daily spend can swing, and when to switch or change them.

Use this task 5 quick questions, then open it in ChatGPT, Claude or Gemini.

This is a practical, copy-paste process for you are launching campaigns and do not know what budget or bidding to use. Use it when you want a clear structure, then add your own details.

Verified on 2026-10-227 sources cited

Quick start

1. Count how many sales or leads each campaign got in the last 30 days, leaving out the last few days if results arrive late.
2. Divide your weekly budget by your cost per result. That is how many results you can buy a week.
3. On Google, start with Maximize conversions. Add a CPA target only when you have a baseline.
4. Base any target on what you really paid recently, not on what you wish.
5. On Meta, fewer ad sets learn faster. About 50 results a week per ad set is the goal.
6. Daily budgets can overspend: Google up to 2x on a day, Meta up to 75% more. Judge by the week or month.
7. Change one thing at a time, then wait 2 weeks or 2 conversion cycles, whichever is longer.

Before you start

Full method

How to use this method

  1. 1. Copy or downloadCopy the prompt below, or download the method file and attach it to your chat.
  2. 2. Paste into any AIChatGPT, Claude, Gemini or Copilot. Replace each [PLACEHOLDER] with your details, or let the AI ask you.
  3. 3. Add your materialAttach your own files, data or brief so the AI works from your facts, not guesses.

The prompt

Paste it as-is. The AI will ask for anything still in [BRACKETS].

Download .md
# Set ad budgets and choose a bid strategy
Method by Deepstory Research: deepstoryresearch.com/method/ad-budget-and-bid-strategy

You are a performance-marketing lead who sets budgets and bid strategies for Google Ads and Meta accounts. Follow the method below step by step. Goal: A budget and bidding plan for Google Ads and Meta: how much each campaign needs to learn, which bid strategy to start on, starting targets, how far daily spend can swing, and when to switch or change them.

## My inputs
- Platforms and campaigns: [PLATFORMS_AND_CAMPAIGNS]  (example: Google Search and Meta Advantage+ leads)
- Monthly budget: [MONTHLY_BUDGET]  (example: AED 6,000)
- Conversions in the last 30 days: [CONVERSIONS_IN_THE_LAST_30_DAYS]  (example: 22 bookings on Google, 9 on Meta)
- Current cost per result or ROAS: [CURRENT_COST_PER_RESULT_OR_ROAS]  (example: about AED 180 on Google, AED 270 on Meta)
- Target: [TARGET]  (example: under AED 250 per booking)

Before you start: if any input above is still in [BRACKETS] or missing, ask me for it in one short numbered list and wait. Do not guess. If I attached files (data, briefs, Deepstory evidence files), treat them as the primary source.

## Steps to follow
1. Check each campaign has one tracked primary goal, its conversion delay, and whether it is limited by budget.
2. Work out how many results each campaign can buy per week at the current cost.
3. Choose the starting bid strategy per campaign and say why.
4. Set starting targets from real history, excluding the latest conversion-delay days.
5. Split the budget so each campaign can learn, and state the maximum possible spend.
6. Explain how far daily budgets can overspend on each platform.
7. Write the rules for when to switch strategy or change a target or budget, and by how much.
8. Set the review dates and what to compare.

## Deliver the output in this structure
- Readiness check per campaign
- Weekly volume math
- Bid strategy choice and reasons
- Starting targets
- Budget split with maximum possible spend
- Switch and change rules
- Review calendar

## Quality rules
- Show the math for every budget figure.
- Label practitioner thresholds as rules of thumb.
- Never set a starting target tighter than recent actual results without saying volume will drop.
- Judge leads on cost per qualified lead or booking, not cost per form fill.
- Label every number as FACT (with source and date), ESTIMATE (show the calculation) or ASSUMPTION.
- If a Deepstory evidence file is attached, use its figures as the reference values and cite its source references; tag claims that are not from it as [UNVERIFIED].
- End with: the 3 assumptions that matter most, what would change the answer, and the single best next step.

What you need to give the AI

InputPlaceholderExample
Platforms and campaigns[PLATFORMS_AND_CAMPAIGNS]Google Search and Meta Advantage+ leads
Monthly budget[MONTHLY_BUDGET]AED 6,000
Conversions in the last 30 days[CONVERSIONS_IN_THE_LAST_30_DAYS]22 bookings on Google, 9 on Meta
Current cost per result or ROAS[CURRENT_COST_PER_RESULT_OR_ROAS]about AED 180 on Google, AED 270 on Meta
Target[TARGET]under AED 250 per booking

Use it when

  • You are launching campaigns and do not know what budget or bidding to use.
  • Results swing every week and you suspect bidding or budget.
  • You want to move from Maximize conversions to a CPA or ROAS target.

Best with: Any chat AI; attach the last 30 to 90 days of campaign data with customer details removed.

What you get

  • Readiness check per campaign
  • Weekly volume math
  • Bid strategy choice and reasons
  • Starting targets
  • Budget split with maximum possible spend
  • Switch and change rules
  • Review calendar

The AI will ask you first

Answer these before the AI advises you; each one changes what it recommends.

  1. 1

    How many conversions did each campaign get in the last 30 days, and how long after the click are they usually recorded?

    Conversion volume and delay decide which strategies work and when results can be judged.

    If unknown: Assume under 30 a month and a 1 to 7 day delay.

  2. 2

    What is the actual cost per conversion or ROAS over the last 30 days, and do conversions measure qualified leads or sales?

    Google suggests starting a Target CPA from the last 30 days' average adjusted for delay; raw form fills can hide junk leads.

    If unknown: Start without a target until there is a baseline.

  3. 3

    What can you afford per sale or lead, or what ROAS do you need to break even?

    That is where targets should move toward over time, not the starting point.

    If unknown: Calculate break-even from price and margin if the user gives them.

  4. 4

    Is any campaign "Limited by budget" today, and is auto-apply for recommendations on?

    Budget-limited targets fight each other, and auto-apply can change budgets or targets without anyone deciding.

    If unknown: Ask the user to check both before changing anything.

  5. 5

    Are conversions counted once (for example Pixel and Conversions API deduplicated, one primary goal on Google)?

    Double counting makes every volume and cost figure in this plan wrong.

    If unknown: Check deduplication before using the numbers.

  6. 6

    Do conversions have values that differ a lot, such as order sizes?

    Value-based bidding needs reliable values.

    If unknown: Bid for conversions, not value.

  7. 7

    Are there sales, seasons or promotions coming up?

    Short spikes confuse Smart Bidding unless handled.

    If unknown: Ask about the next 60 days, including seasonal peaks such as Black Friday, Diwali, Eid or year-end sales.

The procedure, step by step

Read this before you run it, so you can check the AI actually followed the method.

  1. 1

    Check readiness

    For each campaign: one primary goal, last 30 days of conversions (excluding the latest delay days), cost per result, conversion delay, "Limited by budget" status, auto-apply status.

  2. 2

    Do the volume math

    Weekly budget / cost per result = results a week. Compare with Google (2 conversion cycles; about 30 a month for target evaluation) and Meta (about 50 a week).

  3. 3

    Choose strategies

    Google: Maximize conversions or value first; Target CPA or ROAS once there is a baseline. Meta: highest volume or value first; cost or ROAS goals later.

  4. 4

    Set starting targets

    From recent actual results, not the goal. Say what volume you expect to lose if the target is tighter.

  5. 5

    Split the budget

    Fund proven demand first. If Meta cannot get near learning volume, use one ad set and accept "Learning limited", or move budget to the platform that converts.

  6. 6

    State maximum possible spend

    Google: up to 2x a day, 30.4x a month. Meta: up to 175% a day, 7x a week (210% a day with budget sharing). Recommend account spending limits as a backstop.

  7. 7

    Write change rules

    One target change per conversion cycle; budget steps of about 20%; no Meta edits during learning; data exclusions in Google for tracking outages.

  8. 8

    Set review dates

    Two weeks or 2 conversion cycles after each change, whichever is longer, comparing like-for-like periods.

Platform settings(show)
PlatformPathSettingNote
Google AdsCampaign settings > BiddingMaximize conversions / Target CPA; Maximize conversion value / Target ROAS"Maximize conversions with a Target CPA" shows as "Target CPA" from June 2026; behaviour is the same.
Google AdsCampaign settings > BudgetAverage daily budgetUp to 2x on a day; at most 30.4x a month.
Google AdsCampaigns > Bid strategy status"Learning" statusAt most 7 days.
Meta Ads ManagerCampaign or Ad set > BudgetDaily budget (weekly pacing), ad set budget sharingUp to 175% a day; at most 7x a week.
Meta Ads ManagerCampaign or Ad set > Bid strategyHighest volume or value, Cost per result goal, Bid cap, ROAS goal
Meta Ads ManagerAd set > Columns > Last significant edit, ResultsLearning phase progress

Worked example

The figures below are illustrative. Use your own numbers and facts.

Scenario: A Dubai dental clinic with AED 6,000 a month on Google Search and Meta.

Input: Average treatment value about AED 1,200, gross margin about 50%, about half of bookings turn up and pay. Google Search: 22 bookings in 30 days at about AED 180 each (about AED 3,960), Maximize conversions. Meta Advantage+ leads: 9 bookings at about AED 270 (about AED 2,430). Target: under AED 250 per booking overall. Bookings are recorded within 3 days.

Output excerpt: Break-even: an average treatment is worth about AED 1,200 at a 50% gross margin, so break-even is AED 600 per paying patient; about 50% of bookings turn up and pay, so break-even cost per booking is AED 300, and the AED 250 target leaves a margin. How much to spend: judging a Google Target CPA needs about 30 bookings a month, which is about AED 5,400 at AED 180, more than the AED 4,000 Google budget, so Google stays on Maximize conversions. On Meta, about 50 bookings a week at AED 270 would need about AED 58,000 a month, so Meta cannot leave learning on bookings at this budget. Google: 22 a month is below the 30 Google uses to judge a Target CPA, so any target would be provisional. Keep about AED 4,000, which is roughly flat; check impression share lost to budget before adding more. Meta: keep about AED 2,000 a month (about AED 66 a day, so Google plus Meta stay within AED 6,000), which buys about 2 bookings a week, far from about 50 results a week; use one ad set and accept "Learning limited" while cost per booking is near target, and optimise for leads only if leads still turn into bookings. Maximum possible spend: Google up to about AED 263 on one day (2x of AED 131.6), Meta up to about AED 115 on one day (175% of AED 66). Review on 5 and 19 November.

Check your result

Before you trust the AI's output, confirm each of these.

  • ☐ Every campaign has one primary goal and a stated conversion volume and delay.
  • ☐ Each strategy choice meets the platform's minimums.
  • ☐ Starting targets come from recent actual results, excluding the delay window.
  • ☐ Budget math and maximum possible spend are shown for each campaign.
  • ☐ Change rules and review dates are written down.
What you'll get(show template)
## Budget and bidding plan: [BUSINESS]

Total monthly budget: [amount] | Data used: [dates, excluding last [n] delay days]

### Readiness
Campaign | Platform | Primary goal | Conversions (30 days) | Cost per result / ROAS | Delay | Limited by budget? | Auto-apply?

### Volume math
Campaign | Weekly budget | Results a week | Enough to learn? (Google: 2 cycles, 30 a month to judge targets; Meta: about 50 a week)

### Bid strategies
Campaign | Strategy now | Starting target | Why | Switch trigger

### Budget split and maximum possible spend
Campaign | Monthly | Daily | Max in a day | Max in a week or month | Reason

### Change rules
- One target change per conversion cycle; budget steps about 20% (rule of thumb)
- No Meta edits during learning; batch changes

### Review dates
[date 1], [date 2]: compare [period] vs [period]

Quality checks built into the prompt

  • • Show the math for every budget figure.
  • • Label practitioner thresholds as rules of thumb.
  • • Never set a starting target tighter than recent actual results without saying volume will drop.
  • • Judge leads on cost per qualified lead or booking, not cost per form fill.
  • • Every number labelled FACT, ESTIMATE or ASSUMPTION; claims outside attached evidence tagged [UNVERIFIED].

Decision rules

If/then rules the AI should apply. Heuristic rules are a practitioner's rule of thumb, not a platform rule.

IfThenType
You need a break-even target.Break-even ROAS = 1 / gross margin (40% margin means 2.5). Break-even cost per sale = order value x gross margin. For leads: break-even cost per lead = value of a customer x gross margin x lead-to-customer rate. Set ROAS targets above break-even ROAS and cost targets below break-even cost; use lifetime gross profit only if customers reliably reorder.Threshold: arithmetic, not a platform ruleHeuristic
Someone asks how much to spend on Google or Facebook ads per month.Work backwards: a Google campaign needs about 30 conversions a month to judge a target (so about 30 x expected cost per conversion), and a Meta ad set needs about 50 results a week to leave learning (about 50 x cost per result x 4.3 a month). If that is more than you can afford to lose while testing for 4 to 6 weeks, run one campaign or ad set, optimise for an earlier event only if it still predicts sales, or use one platform only.Threshold: about 30 a month (Google), about 50 a week (Meta); rules of thumb based on Google and Meta guidanceHeuristic
You are choosing between Performance Max and Search for a new or low-volume account.Start with Search (plus Standard Shopping if you sell from a product feed), and add Performance Max once the account records about 30 conversions a month and has enough images and videos.Threshold: about 30 conversions a month (rule of thumb)Heuristic
There is no conversion tracking at all.Stop: set up conversion tracking first, because conversion-based bidding (Maximize conversions, Target CPA, Target ROAS) needs it. Use only real numbers from the account; never invent conversion history.Platform rule
Leads are tracked but many are not real customers.Judge budgets on qualified leads, and import qualified leads from the CRM (see the conversion tracking task) before raising budgets.Heuristic
A Google campaign has conversion tracking but little or no history.Start on Maximize conversions (or conversion value) with no target. Target CPA can start without history, but a target without a baseline is a guess.Platform rule
You set or add a Google Target CPA.Start from the last 30 days' average CPA adjusted for conversion delay; a target set too low loses conversions. Google recommends judging Target CPA on 30 days with at least 30 conversions, so below that treat any target as provisional.Platform rule
You want Target ROAS.Meet the minimums for the campaign type (past 30 days unless stated: Search and Shopping 15 conversions; Demand Gen 50 in 35 days; Video 30), report values for about 4 weeks or 1 to 2 conversion cycles while bidding to Target CPA, then start from recent actual ROAS.Platform rule
You change a Google bid target.Avoid more than one change within a conversion cycle, and judge over 2 conversion cycles or 2 weeks, whichever is longer.Platform rule
A Google bid strategy shows "Learning".It lasts at most 7 days. New or reactivated strategies, composition changes and goal changes trigger it; budget and target changes do not.Platform rule
You are tempted to pause a Smart Bidding campaign for a few days to save budget.Lower the budget instead; reactivating a strategy can put it back into learning.Heuristic
You set a Google daily budget.Expect up to 2 times the daily budget on some days, but no more than 30.4 times the daily budget in a month.Platform rule
A Google campaign uses Target CPA or Target ROAS.Google updated bidding from 17 August 2026 and warned of temporary swings. Check targets with the Bid Target Adjustment Tool, especially if limited by budget, and note the change when comparing with earlier periods.Platform rule
You set a Meta daily budget.Expect up to 75% more than the daily budget on some days (175% of it), and no more than 7 times the daily budget in a Sunday to Saturday week; with ad set budget sharing, up to 210% a day and 8.4 times a week. Judge Meta spend by the week.Platform rule
A Meta ad set will not reach about 50 results in a week.Merge similar ad sets and set a realistic budget; ad sets usually exit learning after about 50 results in the week after the last significant edit.Platform rule
You change a Meta budget or bid strategy.Avoid frequent budget changes, which can send an ad set back into learning. Batch changes, then leave the ad set alone.Platform rule
You need cost control on Meta.Start with the highest volume or highest value strategy; add a cost per result goal or ROAS goal only once you have a stable baseline.Platform rule
Merging still cannot reach about 50 Meta results a week.Consider optimising for an earlier, higher-volume event only if it still predicts revenue, and track cost per qualified lead or booking. "Learning limited" is acceptable if cost per qualified result is on target.Heuristic
You scale or tighten.Raise budgets by about 20% at a time every few days, and move targets by no more than about 10% to 20% per change, then wait the review window.Threshold: 20% budget steps; 10% to 20% target steps (rules of thumb)Heuristic

Failure modes

Common mistakes with this method, how to spot them, and how to fix them.

MistakeSymptomFix
Setting a Target CPA at the dream number instead of the real one.Spend and conversions collapse.Start near the last 30 days' CPA and tighten in small steps.
Changing targets or budgets every few days.Performance never settles; Meta keeps re-entering learning.One change per conversion cycle; batch Meta changes.
Panicking at a day that spent far more than the daily budget.Budget cuts that starve good days.Judge Google by the month (30.4x) and Meta by the week (7x).
Optimising Meta for an earlier event without a quality check.Cheap junk leads; bookings fall.Track cost per qualified lead or booking.
A Target CPA or ROAS campaign limited by budget.It cannot spend or hit the target.Raise the budget, loosen the target, or return to Maximize conversions.
A tracking outage or a promotion spike left in the data.Smart Bidding over- or under-bids afterwards.Use data exclusions for outages and seasonality adjustments for short promotions.
Auto-apply recommendations left on.Budgets or targets changed without anyone noticing.Check Recommendations > Auto-apply.
A Meta cost per result goal set below recent costs.The ad set stops spending.Remove the goal or set it near recent actual cost.
A shared budget or portfolio mixing campaigns with very different costs.Cheap campaigns starve the ones that matter.Separate budgets for campaigns with different goals.
Switching to Target ROAS with unreliable values.Bidding chases noisy values and volume falls.Fix values first and meet the minimums.

Agent mode: if an AI agent runs this unattended

May do without asking

  • • Calculate volumes, budgets, maximum possible spend and starting targets from data the user attaches.
  • • Recommend bid strategies and change rules.
  • • Draft a change log template.

Must stop before (the user decides and does these)

  • • Changing any live budget, bid, target or bid strategy, which changes spend; state the maximum possible spend of the change first.
  • • Creating or applying Google experiments, which split live traffic and spend.
  • • Turning on or accepting auto-apply recommendations.
  • • Changing account spending limits or turning on Meta ad set budget sharing.
  • • Publishing, enabling or duplicating any campaign or ad set.
  • • Uploading data to the ad accounts.
  • • Accepting platform terms or changing billing or payment settings.

Evidence for this method

This method runs on your own material: your files, brief or code. If your task needs market or economic numbers, attach a sourced context file from the evidence library alongside the prompt.

Browse evidence

Run next

Methods that chain well after this one.

Deepstory methods are process guidance for AI tools. Review AI output before relying on it; legal, tax, medical and investment decisions need a qualified professional.