Running Google Ads without an agency in 2026

In 2026, 65 to 75% of SMBs running their Google Ads in-house achieve a CPA equal to or lower than their former agency's, for a total budget 40-60% lower. In-housing has become the median pattern under $22,000/month — not the exception.

Every month, I get the same question from an SMB founder or a startup head of marketing: "Can we run Google Ads without an agency?" The honest answer, after the SteerAds 2025-2026 sample (continuously audited accounts): yes, in most cases under $22,000/month. About 36 to 46% of SMBs in this bracket today pay an agency they could bring in-house — not because the agency is bad, but because the delta it brings no longer justifies its cost in 2026.

This article is the complete guide I wish I had read before taking the plunge: skills to acquire (4 levels, 3 months to reach level 2-3), real time required (2-3 hours/week at cruising speed), tooling stack (steering tool + GA4 + Looker Studio), weekly/monthly/quarterly cadence, autonomous audit method in 30 min, and signals for knowing when to rehire an expert. Balanced tone — there are cases where the agency is truly worth it. But many others where it no longer is.

Why bring Google Ads in-house (or not)?

In-housing isn't a religion. It's an economic decision, tied to four variables: your media budget, the time available in the team, the PPC skill level already present, and the willingness to invest 90 days of learning. Here's the decision matrix I use in audits:

The 4 conditions of healthy in-housing:

Budget < $22,000/month — above that, the marginal gains of a full-time expert (in-house or freelance) systematically exceed their cost.
One person available 3-5 hrs/week — no less, or the account drifts in 6 weeks. No more either (above 8 hrs/week = hidden underperformance).
Ability to read a Google Ads account — understand CPA, CTR, QS, impression share, even intuitively. Otherwise, mandatory prior training (2-3 weeks).
Accept 90 days of learning — the first 30 days are frustrating, the next 60 comfortable, autonomy is earned around D75-90.

The case I see most often :

an SMB paying $1,320/month to an agency to manage $8,800/month of Google Ads budget. The agency spends 3 hrs/month on the account (the junior account manager, 10 accounts to manage). Fees/spend ratio: 15%. Median gain post-in-housing, fees reinvested in media: +22 to +34% conversions at constant budget by maturity. It's an inefficient market, not a mean agency — just an economic model that's misaligned.

Let's be clear on one point: agencies have a legitimate role. Above $55k/month of spend, with a complex B2B stack, multi-country or multi-account, their expertise and ability to hold a process across several people is worth the cost. The issue isn't the existence of agencies — it's that many SMBs pay Rolls-Royce prices to manage a scooter. The alignment is sometimes broken, and that disconnect is what justifies in-housing.

What skills do you learn in 3 months?

I've observed the in-house Google Ads learning curve across dozens of SMBs. It breaks into 4 clear levels. The good news: most SMBs only need levels 2-3 to steer correctly. Level 4 is for senior experts managing $100k+/month accounts.

Level 1 — Understand (2 weeks). Master the vocabulary: CPC (cost per click), CPA (cost per acquisition), CTR (click-through rate), Quality Score (Google's 1-10 rating), impression share (captured impression share), match types (exact, phrase, broad), Smart Bidding (automatic bidding). At this stage, you read a Google Ads report without getting lost. Resources: Google Ads Help (free, complete) and Google Skillshop (free certifications). Our ROAS / CPA / CPC guide covers the fundamentals in 15 min.

Level 2 — Operate (3-4 weeks). Create a Search campaign from A to Z: type choice, ad group structure, Responsive Search Ads writing, extension activation, Smart Bidding configuration (Maximize Conversions to start, Target CPA next), shared negative list setup. You can also adjust bids daily and keep the account in good hygiene. Sufficient level for 70% of SMBs under $16k/month.

Level 3 — Optimize (2 months). You add GA4 attribution (data-driven instead of last-click), cohort / LTV reading, Google Ads scripts to automate repetitive actions, fine dayparting, structured ad copy A/B tests, smart demographic exclusions. This is the level of a good junior PPC manager. Reachable in 4-6 months for a curious marketing profile.

Level 4 — Architect (6 months+). Senior expert level: incrementality tests via holdout, multi-campaign / multi-product budget allocation, scale long-tail strategy, setting up a Shopping CSS (to save 20% on Shopping), migration to Search Ads 360 above $110k/month of spend. Useless for an SMB, essential for a scale-up above $55k/month.

Allocated time: are 2 hours/week enough?

It's THE question I get. Honest answer: 2 hrs/week is the vital minimum, not comfort. On accounts we continuously observe, average time for healthy in-house steering on a $11-33k/month budget is 2.5 to 3.5 hrs/week (median by maturity). Under two conditions: (1) the initial setup was solid (30 one-shot hours), (2) a tool automates continuous monitoring (steering tool or custom scripts).

If you really only have 1 hr/week to dedicate to the account, only two options: rehire (agency or freelance), or consciously accept 15-25% underperformance. The in-between "I'll optimize at the last minute when I think about it" systematically produces a CPA that drifts 8% per month.

Typical breakdown of a 3-hour week:

30 min · Monday morning · KPI review — check dashboard, identify the 3 drifts of the week, prioritize.
30 min · Tuesday · Campaign adjustments — alert processing, Target CPA adjustments, budget pacing, negatives added.
30 min · Wednesday · Ad copy / creative — review underperforming RSAs, creative rotation, test a new angle.
60 min · Thursday · Projects — new campaign, A/B landing, deep cohort analysis, seasonality prep.
30 min · Friday morning · Wrap-up — write the weekly recap (1 shared team doc), note decisions, list the following week's to-do.

The secret: this time is blocked in the calendar. SMBs that successfully in-house without pain are those that treat Google Ads like a non-negotiable recurring meeting. Those that fail are those who "will do it when they have time." Spoiler: you never have time.

Which tooling stack to choose (SteerAds + GA4 + Looker)?

An effective in-house PPC manager needs 5-6 tools, no more. The classic in-housing trap: stacking tools instead of mastering 3 or 4 thoroughly. Here's the minimum recommended stack, with indicative budget.

1 · Google Ads UI — obviously. Free. The official interface remains the main cockpit. Learn to love Google Ads Editor (desktop app, allows offline bulk editing — 3x time saving on large accounts).
2 · Google Analytics 4 (GA4) — free. For data-driven attribution, event tracking, advanced remarketing audiences. GA4 isn't optional in 2026: without it, your Smart Bidding runs half-blind.
3 · Looker Studio (formerly Data Studio) — free. Unified dashboard that combines Google Ads + GA4 + your CRM / LTV data. One template is enough to start, refine afterward.
4 · SteerAds (or custom scripts) — $0 to $165/month. 24/7 monitoring, auto quarterly audit, anomaly alerts, missing-negative detection, /100 score. The tool that replaces the 10-15 hrs/week of manual monitoring by a senior PPC manager. Alternatives: Google Ads Scripts (free but requires JS skills), Google Ads API.
5 · Keyword research — Google Keyword Planner (free, built into Google Ads) is enough for 90% of SMBs.
6 · (Optional) Search Ads 360 — paid, justified only above $110k/month of multi-platform spend.

Total SMB stack budget: $0 to $220/month. Compare to $880-3,300/month of agency fees. The ROI of in-housing is essentially made there.

The role of the steering tool in this stack :

we're the tool that runs, continuously, what an experienced PPC manager does once a month during their audit. Automatic detection of 200+ checkpoints, real-time alerts, /100 score updated daily, recommendations prioritized by impact/effort. The idea: you keep control of strategic decisions, our tool handles tactical monitoring. It's not blind automation, it's a copilot.

To go further on tracking (critical axis 4 of any Google Ads audit), read our Conversion Tracking Google Ads guide covering GA4, Enhanced Conversions, Consent Mode v2, and clean integration with your CRM.

Cadence: weekly, monthly, quarterly

Running Google Ads is like piloting a motorboat: frequent micro-adjustments, a few major course corrections, and a big periodic technical inspection. The cadence across three horizons — weekly, monthly, quarterly — covers everything without redundancy.

The golden rule: never skip the quarterly cadence. It's the one SMBs sacrifice first under pressure ("no time this quarter, we'll do it next"). Observed result: accounts skipping 2 consecutive audit quarters see CPA drift 18 to 28% by vertical. The quarterly audit isn't negotiable — that's exactly why you have a tool that can do it automatically.

Weekly and monthly cadences can compress temporarily (vacations, product rush) without major damage for 2-3 weeks. Beyond that, the account drifts. The quarterly cadence, on the other hand, must fall exactly every 90 days, with the same rigor as your accounting close.

How do you do an autonomous quarterly audit in 30 minutes?

The quarterly audit is the pillar of healthy in-housing. Full version: 4-6 hrs with the complete 5-axis checklist. Express 30-minute version: 6 critical checks that capture 80% of drifts. If you only have 30 min, do these 6.

1 · Conversion tracking active? Tools → Measurement → Conversions. Every primary conversion must have "Active" status and a non-zero conversion rate. If "No recent activity" on a primary conversion: red alert.
2 · Smart Bidding optimizing on the right conversion? Verify each campaign points to the target primary conversion (not a micro-conversion by mistake).
3 · Lost search impression share > 30%? If yes: either insufficient budget (lost to budget), or bids too low (lost to rank). Requires arbitrage.
4 · Ad copy with CTR < 2%? List all RSAs under 2% CTR and rewrite them. Threshold below which performance clearly degrades on search.
5 · Budget pacing OK? No campaign should have spent 100% of its daily budget before 8 PM. If yes: increase budget or revise strategy.
6 · Shared negatives up to date? The shared negatives list should be reviewed monthly. Target: 150+ entries minimum, with regular cleanup of polluting search terms.

For the full 5-axis version (200+ checkpoints), follow the detailed checklist in our Google Ads audit guide. It's the same methodology as $2,000-5,000/audit consultants, documented step by step.

Automated audit shortcut :

these 6 critical checks (and the other 194) run continuously on your account. You receive alerts by email, and an exportable quarterly PDF report generates automatically. Time spent: 3 minutes to read the report, instead of 30 minutes to produce it manually. Launch your free audit to try it.

What are the typical mistakes of a beginner PPC manager?

I've seen these mistakes so many times that I rank them in order of impact. The most expensive beginner SMB in-house mistake: letting Smart Bidding run without verifying tracking. Median observed impact: -22 to -34% ROAS over 60 days depending on context. It's mistake #1 on its own. Here are the 8 classics.

1 · Touching too many things at the same time. Modifying structure + Smart Bidding + ad copy the same week = impossible to attribute effects. Rule: 2 changes max per week, separated by 7-14 days minimum.
2 · Judging a campaign in less than 14 days. The minimum statistical threshold before drawing conclusions. Below that: you're reacting to noise.
3 · Copying the structure of an old account without adapting it. Each market, each product, each season demands its own architecture.
4 · Over-segmentation. 50 ad groups with 10 conversions each = blind Smart Bidding (below the 30 conv/30d threshold). Consolidate: 10-15 ad groups with 30+ conv each > 50 ad groups with 10 conv.
5 · Forgetting negatives. Broad match without robust negatives = 20-30% waste in 3 months. Our Smart Bidding guide details the negatives × Smart Bidding interaction.
6 · Leaving old paused campaigns with active ad extensions. Basic hygiene cleanup but forgotten 9 times out of 10.
7 · Ignoring the mobile / desktop difference. Conversion rates can vary 2-3× between mobile and desktop. No bid modifier = money left on the table.
8 · Changing bid strategy every 2 weeks. Smart Bidding needs 14-30 days to calibrate. Changing too often = you're permanently in "learning phase," never in performance phase.

Cumulative median of these mistakes with a non-accompanied in-house PPC: -20 to -30% ROAS over the first 6 months, before the learning curve stabilizes. The good news: each mistake has a clear corrective action, and most resolve in 30 minutes once identified. Our article 10 Google Ads mistakes covers them with concrete examples.

Budgets ≥ $22,000/month: when to rehire an expert?

Let's be honest: in-housing has its limits. Above $22,000/month of spend, some signals indicate that an external expert (or senior in-house) will generate more value than their cost. Don't ignore them out of autonomy pride.

The 4 signals that it's time:

More time managing than analyzing. If you spend > 5 hrs/week on pure operations (click, export, adjust), there's no time for strategy. Delegate.
Multi-country or multi-brand. Running 3 countries with 3 languages and 3 exchange rates exceeds the capacity of a part-time person.
Complex B2B attribution. 60-180 day sales cycle, multi-touch, lead scoring — it's a profession, not a side project.
Need for advanced strategies. CSS Shopping, Performance Max multi-feed, Search Ads 360, incrementality tests — level 4 required, rarely found in-house at SMBs.

My advice in the $22-55k/month gray zone: the hybrid model wins almost always. Motivated in-house (level 2-3) + steering tool for continuous operations + senior freelance 4-8 hrs/month for strategic arbitrages = total cost $1,650-2,750/month, quality equivalent to an agency at $3,850/month. The team keeps control, the expert intervenes where they add the most value.

When and how do you fire your agency?

If the decision is made — and it must be made cleanly — the "how" matters as much as the "when." A poorly managed transition costs on average 4 to 8 weeks of post-departure underperformance, sometimes more. Here's the diplomatic method that works.

The 4 steps of a clean handover:

1 · Recover all access. Not just Google Ads — also GA4, Google Merchant Center, Google Tag Manager, Search Console, Looker Studio, possibly Google Business Profile. Often the agency created the accounts with their own email: ask to be Owner (not just editor) on each platform. Timeline: 2 weeks before contract end, not the last day.
2 · Ask for setup documentation. Conversion tracking (which conversions, which tags, which values), audience lists (remarketing, Customer Match), custom scripts, Zapier / Make automations, campaign naming rules. If the agency refuses or drags feet: escalate politely, remind them it's in the contract.
3 · Respect legal notice period. 1-3 months depending on contract. Don't leave slamming the door — it's your leverage to get documentation and access. Use the time to set up the shadow transition.
4 · Progressive transition. Month 1 — shadow: you observe in real time, don't touch anything. Month 2 — joint operation: you take over routine tasks, agency validates. Month 3 — takeover: you steer, agency as support only when blocked. End of month 3 — full autonomy.

Absolutely avoid :

cutting the relationship suddenly without transition, without recovering documentation, without having access. Average cost of a failed handover I observe: 4 to 8 weeks during which the account runs on the agency's config (which you don't master), with CPA drifting 15-25%. Always prefer slow transition to brutal rupture.

Last point: don't leave in resentment. An agency that did average work for 2 years isn't an enemy — maybe it's just an agency poorly suited to your stage. Keep them in your network. It happens that an in-housed SMB decides 18 months later, when moving to $77k/month of spend, to return to an agency. It's easier if you left cleanly.

What are the 7 KPIs to track in a steering dashboard?

An in-house PPC manager needs a single dashboard that synthesizes account health in 30 seconds. Too many KPIs = paralysis. These 7 suffice for 95% of daily decisions. The other metrics, you go to on demand.

1 · Daily spend vs monthly budget. Pacing. If you've spent 40% of budget at day 15 (of 30), you're exactly on. 60% = alert, 30% = you're underutilizing.
2 · CPA vs target per campaign. Per campaign, not account average. The average masks catastrophic campaigns offset by cash cows.
3 · Account average CTR. Early warning signal: if global CTR drops 20% in 2 weeks without an identifiable reason, an axis is degrading (creative, Quality Score, audience saturation).
4 · Impression share (absolute + lost to budget + lost to rank). The triple breakdown. If lost to budget > 20%: increase budget OR revise bids. If lost to rank > 30%: your Quality Score or bids are too low.
5 · Conversion rate per landing page. Not natively in Google Ads — measured via GA4. If one landing converts 3× less than another while its CPA is close: you're paying clicks at the same price but conversion is more expensive. Investigation signal.
6 · Top 10 missing negatives detected auto. Polluting search terms consuming without converting. A monitoring tool detects this continuously; otherwise, weekly export of the Search Terms Report to do yourself.
7 · Anomaly alerts. Spend spike (abnormal overrun), CPA spike, broken tracking, conversion rate drop, new search term volume. The element that wakes you at 7 AM instead of 6 PM — precious.

CTA — Try SteerAds for free :

the 7 KPIs above are unified in our dashboard, with email alerts for anomalies and action recommendations prioritized by impact/effort. Start with a free account audit (3 minutes, no credit card, exportable PDF report). Then, if you like the tool, activate continuous monitoring to replace your 10 hrs/month of manual surveillance — or automation via auto-optimization.

If you prefer to build your own dashboard: Looker Studio connected to Google Ads + GA4 + (optionally) your CRM via BigQuery. Expect 4-6 hrs of initial setup for a template covering these 7 KPIs, then monthly iterations. It's free, but anomaly alerts remain to code yourself (Google Ads Scripts or Apps Script). Need personalized support? Contact us, we'll look together.

Sources

Official sources consulted for this guide:

FAQ

Do you need to be an expert to run Google Ads in-house?

No. A motivated SMB owner reaches 'autonomous' level on a basic account in 90 days at 3-5 hours/week. Minimum skills: read a report, create a simple Search campaign, understand Quality Score, add negatives, interpret a CPA. A steering tool like SteerAds compensates for blind spots (continuous monitoring, alerts, auto audit). An expert remains necessary if your budget exceeds $22,000/month or if multi-country / long B2B complexity is added — not before.

My agency says it's too complex to bring in-house — true?

Partially. Google Ads 2026 is more automated than in 2018: Smart Bidding, Performance Max, Responsive Search Ads do 70% of optimization work on their own. What remains complex: tracking (GA4 + Enhanced Conversions + Consent Mode v2), account structure, data interpretation. But 'complex' doesn't mean 'inaccessible': 41% of SMBs under $33k/month of budget pay an agency they could bring in-house, according to our data. The real question is available time, not pure complexity.

What's the minimum budget where in-housing is worth it?

No low threshold: even at $2,200/month, in-housing makes sense if you have the time. The real threshold is in the other direction: above $22,000/month, an expert's gains exceed their cost. Between 0 and $22k/month: in-housing saves 15-30% of budget that was going to agency fees ($880-3,300/month), reinvestable in media. Between $22k and $55k/month: gray zone, to arbitrate based on in-house skills. Above $55k/month: senior dedicated expert (in-house or freelance) almost mandatory.

Can you in-house without a tool like SteerAds?

Yes, it's doable. You can use only Google Ads UI + GA4 + free Looker Studio. The cost is then your time: expect 5-8 hours/week instead of 2-3, because you have to manually do what a tool automates (anomaly monitoring, quarterly audit, detection of missing negatives). SteerAds is an accelerator, not a prerequisite. The choice: either you have time and no tool budget, or the inverse. Both work — but above $11k/month of spend, the ROI of a steering tool is generally positive.