Why warm-up exists for Google Ads
Google's smart bidding system is fundamentally a conversion-prediction engine. It watches what users do after seeing your ad, learns the patterns that lead to conversions, and bids accordingly. A fresh account has no learning data. Smart bidding on a fresh account behaves like an inexperienced trader on the first day of a job: it makes confident-looking decisions on insufficient information.
The empirical pattern from our internal test fleet is consistent. Cold-launched fresh accounts on smart bidding strategies pass the first ninety-six hours with roughly forty percent of expected conversion volume. Smart bidding compensates by raising bids, the cost-per-click runs forty to sixty percent above market, and the campaign loses money. Warm-up exists to feed the algorithm enough data so this period does not happen on your budget.
The protocol below is the twenty-one-day version we run on every fresh Google Ads account that enters the test fleet. Operators under quota pressure typically compress it to fourteen days. The principles do not change. Slow start, predictable conversion-signal, signal-watching at every step.
The first seventy-two hours
Day zero is delivery day. The credentials arrive in your dashboard, you import the cookies into your antidetect browser, you set a residential proxy that matches the account's registration country, and you log in. That is the entire activity for day zero. No campaign creation, no payment-method addition, no Ads Manager visit. Let Google record the login as a normal session.
Day one is Google ecosystem validation. Open Gmail, browse the inbox briefly. Open Google Drive if there is anything there. Visit YouTube, watch one or two short videos that match the kind of content the account would naturally consume. Do not yet add a payment method to Ads. Do not yet create campaigns. The point is to look like a returning user, not a buyer rushing to deploy budget.
Day two repeats day one with slightly elevated activity. Send one email if the account has natural Gmail activity history. Browse YouTube subscriptions if any. Search a few queries on Google itself — the natural kind, not the kind your campaign targets.
Day three is the first contact with Ads Manager. Open it, look around, do not create anything. Read the policy reminders Google surfaces if any. Close the tab. Visiting and leaving is normal user behaviour; visiting and creating is the automated-buyer pattern Google's review pipeline reads as suspicious.
Week one: conversion signal seeding
Days four through seven are the conversion-seeding phase. The goal is to get the first ten to fifteen conversions on the account before introducing smart bidding. We launch a deliberately simple manual-CPC campaign on day four with a small budget — typically ten dollars per day — and let it run for seventy-two hours untouched. The keywords should be high-intent and low-competition; the goal is conversion volume, not volume volume.
Conversion tracking should be configured before the campaign launches. Without tracking, smart bidding cannot learn even from organic conversions. We configure both Google Ads conversion actions and Google Analytics 4 events; the redundancy ensures Google has at least one signal source to learn from.
By day seven the campaign should have produced at least eight to twelve conversions. If it has not, slow the protocol — extend week one by three days. Smart bidding cannot operate cleanly on fewer than ten learning conversions, and forcing it onto a thin signal degrades performance for two to three weeks afterward.
Weeks two and three: smart bidding handoff
Week two opens the budget envelope to thirty dollars per day and switches the campaign to a target-CPA bidding strategy with a conservative target. The conservative number gives smart bidding room to learn without immediately overshooting. By the end of week two the algorithm should have absorbed enough signal to bid efficiently within the target range.
Week three crosses into target-ROAS or maximize-conversions territory if your campaign needs aggressive optimisation. By day twenty-one the account has shown Google a pattern: this is an active advertiser running optimised campaigns, with conversion data, with disciplined budget pacing. That pattern is what Google's review pipeline rewards. Cost-per-click should drop noticeably, conversion volume should stabilise, and the account is now in operational state.
The protocol deliberately keeps to one campaign through week three. Multi-campaign launches before the account has built signal divide that signal across campaigns and slow learning. Single-campaign warm-up first, then expand.
Three signals that say slow down
Cost-per-click well above your target during week one is the earliest warning that smart bidding is operating without sufficient signal. If your CPC sits forty percent or more above market for two consecutive days, slow the protocol. Do not advance to week two.
Conversion volume below ten by day seven is the second signal. Without ten conversions in the learning bucket, target-CPA bidding cannot operate. Extend week one until you have the volume.
Policy warnings during the warm-up window are the third signal. Soft warnings indicate Google's review pipeline has flagged something. Resolve the warning, retreat to a safer creative for forty-eight hours, then resume schedule.
Common mistakes we still see
Cold-loading the account on day one with a target-CPA campaign and an aggressive budget. This is the single most common reason fresh accounts underperform. Smart bidding without conversion signal is worse than manual bidding.
Skipping the conversion-tracking configuration. Without it, smart bidding has nothing to optimise against, and the account never builds a useful signal profile.
Editing the campaign multiple times during the first week. Each edit re-triggers the smart bidding learning phase. Once you have set the day-four campaign, leave it alone for seventy-two hours.
Mixing residential proxy GEO with payment-method GEO mismatch. Google reads payment-method country as a stronger trust signal than IP, and a US payment method on a Vietnamese IP is a red flag.
Downloadable checklist
We maintain a one-page version of the protocol as a printable PDF for operators who want the schedule on a wall. The link rotates every quarter to track Google's bidding-strategy updates. The current version timestamp shows when it was last revised.
Comments on this article are open to checked buyers. If your warm-up failed at a specific step, write the day number and the symptom in a comment and we will respond.
From the comments (38 total)
The day-three Ads Manager visit advice is gold. We were creating campaigns the moment we logged in. Survival rate doubled after we changed.
We compressed this protocol to fourteen days for one campaign and saw a noticeable conversion-signal drop. Going back to twenty-one for the next batch.