Should I Run Google Ads Myself or Hire Someone?

An even-handed look at running your own Google Ads versus paying someone. What the job actually is, what the learning costs, and a test to decide.

By Gavin Sevastian · Updated August 11, 2026 · 6 min read

Start here: three things decide this, and budget is the weakest of them

Should I run Google Ads myself, or pay someone to run them? The honest answer is that it depends on three things, and the one most contractors lead with (the size of the budget) is the least useful of the three.

Your decision surface is the real driver

Count the moving parts before you count the dollars. Services you advertise, multiplied by the areas you serve, multiplied by the pages those clicks land on, multiplied by the ways a customer can actually reach you. That product is what decides how much judgment the account demands.

One trade, one city, one page, one phone number is a genuinely small problem. Four trades across a metro, each with its own season and its own job value, is a large one at any spend.

Budget matters, but mostly as evidence volume

A bigger budget does not make the work harder. It makes being wrong faster and more expensive. It also buys enough monthly data for a decision to be supportable, which is the part nobody mentions. At a small budget the opposite problem shows up: not enough happens in a month to tell a bad setting apart from an ordinary slow stretch. If you have not settled whether paid search is the right channel at all, work through that question first before this one.

Hours, counted in your busy season

Anyone can find two hours in February. The account needs attention in the months you are least available, and that is the whole difficulty. A plan built on an even two hours every week is a plan that quietly stops in July, which is exactly when the money is moving.

Both columns can be running the same monthly budget. The budget was never the thing that separated them.

What the job actually is, week to week

Most contractors make this decision without ever seeing the task list. Here is the shape of it.

The setup is decisions, not clicks

Building the campaign is the fast part. The slow part is the decisions underneath it: what counts as a conversion, where the clicks land, where you will and will not serve, and what a lead is worth to you. Get those four wrong and everything built on top of them is efficient in the wrong direction.

Take just the first one. A conversion could mean a form submitted, a phone call that lasted long enough to be a real conversation, a click on a phone number that nobody answered, or a page load. Those are four different definitions of success, and the platform will chase whichever one you hand it.

The recurring work never finishes

The account keeps expanding into searches you did not choose, which is by design. Google's own remedy for irrelevant traffic is manual and advertiser-initiated: you find the search term, you decide it is not you, you exclude it. Nobody does that for you. Doing nothing is therefore not neutral. It is a slow drift toward less relevant traffic that costs the same per click as the good stuff.

What automation actually changed

Automated bidding took away daily bid management. It did not take away the judgment: it moved the judgment upstream, to defining the goal, feeding back accurate conversion data, and bounding where the ads are allowed to run. That is fewer hours and a much higher consequence per hour. Choosing the bid strategy is now a larger decision than any individual bid you would once have typed in by hand.

The recurring job is to keep four things aligned: the searches being funded, the customer actions being counted, the services absorbing the budget, and the business priorities outside the platform. Automation does not know when one of those becomes untrue. The manager's value is noticing the mismatch, choosing whether it warrants a change, and then leaving the result readable long enough to learn from it.

The honest case for running it yourself

A page that will not argue this side properly has nothing to trade on when it argues the other one. So here it is, made as clearly as I can make it.

Nothing here is secret

The entire platform is publicly documented and the training is free. There is no locked room. If you want to know how to learn Google Ads, the answer is genuinely that you read Google's help pages and Skillshop courses, in that order, and then run money through the auction. What a manager sells is attention and having watched the same failure happen many times, not access to something you are not allowed to read.

Any page that hints at proprietary tricks is misleading you. The advantage is repetition, not information.

The tooling is better than it was

Bulk editing is free. Bidding no longer requires anyone to sit there adjusting keyword bids by hand. Pay-per-lead Local Services Ads exist for many eligible local service categories, which was not true a few years ago. The floor of competence has dropped, and anyone telling you otherwise is selling from an outdated script.

One service in a thin market is a small problem

Can I do Google Ads myself? If you run one trade in one city, send every click to one page built for that job, and answer one phone, then yes, and the honest recommendation is that you should. There is not enough decision surface there to justify paying someone a monthly fee to watch it. DIY Google Ads gets a bad name because it is usually attempted on an account that was never that simple.

The honest case for paying someone

This case does not rest on expertise being mysterious. It rests on cost and time, which are both checkable.

The tuition is paid in ad spend, not study time

The course is free and short. The learning is neither, because it happens on live money in a live auction against competitors who are not waiting for you. The first months are also the most expensive ones to be wrong in, since that is when the structural errors are still in place and every click is buying you a lesson at full price.

There is a second cost that never appears on the invoice. While the account is being learned on, it is also the only lead source you were counting on that quarter.

The hours come out of the hours that earn

The comparison is never fee against zero. It is the fee against waste avoided, plus the hours you get back multiplied by what those hours are worth. And those hours do not come out of your evenings, at least not for long. They come out of quoting, selling, and running jobs.

Put a real number on your own time before you decide. For scale, the average hourly wage paid in specialty trade contracting in Canada was $37.05 in 2024, and an owner's hour bills for considerably more than the wage of an hour of labour.

Nobody is watching while you are on a roof

The account does not stop when your week gets busy. It keeps spending at full budget, and it does not tell you it needs you. If you are weighing whether to hire a Google Ads expert or consultant, the thing you are actually buying is that the account is somebody's job rather than somebody's evening, in the exact weeks when you have nothing left to give it.

What the learning actually costs, in published numbers

You can put a public price on being wrong, and you should, because the arithmetic is what makes this decision concrete rather than a matter of temperament.

What a click costs in this category

WordStream by LocaliQ publishes an annual search advertising benchmark study. Its 2026 edition, drawn from more than 13,000 US campaigns, puts Home and Home Improvement at an average cost per click of $8.33 USD and an average conversion rate of 8.05%. That makes this category the second most expensive of the 23 measured, behind legal services and ahead of dental.

At those numbers, a single mistake in your settings does not cost you a rounding error. A week of unfiltered search terms costs dozens of clicks at close to nine US dollars each. What that adds up to across a year is usually the number that changes people's minds.

How much evidence a month of spend actually buys

Run the division on your own budget. It is the most useful five seconds in this whole decision.

Ten conversions in a month is a small sample. On its own, it may not distinguish a broken setting from a slow month. That is precisely why wrong settings survive so long in self-managed accounts: the evidence needed to convict them arrives slowly, and in the meantime everything looks like normal variation.

Now put that next to the calibration clock from earlier. Thin monthly evidence and a recalibration measured in weeks are the same problem viewed from two directions, and together they explain why patience is a skill here rather than a personality trait.

Why the benchmark table does not reconcile with itself

Here is a useful thing to notice. That cost per click divided by that conversion rate gives a cost per lead of $103.48, while the same report states an average cost per lead of $90.92. The figures are not wrong. They are three separately computed medians over a mixed sample, not three numbers from one consistent account, so they were never going to agree.

The lesson is worth more than the numbers. Benchmarks are for sanity-checking your own arithmetic and for setting expectations. They are not goals, and an account that misses them is not necessarily a bad account.

The failure modes the dashboard will not show you

Everything above assumes the numbers on your screen mean what they look like they mean. That assumption is where most self-managed accounts quietly come apart, and it is the strongest argument on either side of this decision. Not because the work is difficult, but because the reporting is silent about one specific set of expensive problems.

A broken tag and no demand look identical

Zero conversions in a month has at least four separate causes. There was no demand that month. The tag stopped firing when somebody updated the website. The calls are arriving on a number that was never set up to be counted. Or a conversion action was never created at all, so nothing was ever going to be recorded in the first place.

The interface renders all four the same way. A zero.

That matters because the four have opposite responses. Genuinely soft demand argues for patience, or for putting the money somewhere else. A dead tag argues for stopping the spend today. Pick wrong and you either keep buying traffic you cannot see, or you shut off a campaign that was quietly working. The phone is where this bites hardest, because most contractors get more calls than form fills and calls are the harder path to measure.

You cannot see everything you paid for

The search terms report feels like a receipt for the month. It is not one. Google withholds low-volume terms from that report for privacy reasons and says so plainly in its own documentation.

Sit with what that means. The waste that is hardest to notice, dozens of odd one-off searches that each happened once or twice, is exactly the waste least likely to be shown to you. A report that looks clean is evidence about the terms with enough volume to appear in it, and nothing at all about the rest.

Nobody is hiding this from you. Those terms cannot be published without identifying individual people. But it does mean "I read my search terms and they looked fine" is a weaker statement than it sounds when you say it.

Nothing turns red for the expensive problems

Google will tell you when an ad is disapproved or a card is declined. Nothing tells you that your location settings are letting in clicks from two hours outside the area you would actually drive to. Nothing tells you the same caller got counted three times. Nothing tells you the leads arriving are people who were never going to buy at your price.

Visible metrics can be easier to diagnose than failures in measurement or lead quality. A clean dashboard is not proof that the expensive parts of the funnel are healthy.

A weak click-through rate is visible and correctable in an afternoon. A conversion definition that counts a tap on a phone number as a lead looks like success right up until you set it beside the jobs you actually booked. When the calls are not coming, the cause sits in this category far more often than it sits in the bidding.

If you already have an account running and something in the last three paragraphs made you uneasy, a second set of eyes on what you have built is a much smaller purchase than handing the whole account over. It is worth asking me to look before you decide anything bigger than that.

Where the defaults are set against you

This is the most immediately useful part of the page if you are running the account yourself. The out-of-the-box settings buy more clicks than a careful operator would, and a contractor who accepts them has not made a bad decision. He has made no decision, which is different and more common.

Location targeting includes people who are not here

By default, location targeting includes both physical locations and locations of interest. Google labels the broad setting Presence or Interest and recommends it, and to its credit it publishes the evidence behind that recommendation: advertisers who switched to it in the travel, real estate and education verticals saw about 5% more conversions on Search.

Look at those three categories. They are ones where interest in a place genuinely predicts a customer. Somebody in Toronto researching Collingwood is a real prospect for a Collingwood realtor. Somebody in Florida reading about Collingwood is not going to need their furnace fixed there next Tuesday.

Google's cited evidence says nothing about a trade where the customer has to physically be standing at the property. The default is not dishonest. It was just never built with your business in mind, and the setting is one click away from being tightened.

Negative keywords do not behave like the keywords you added

This is the single most counterintuitive mechanic in the platform, and almost nobody discovers it by using the interface.

Positive keywords match close variants. Plurals, misspellings, related forms. That is why one keyword can pull in traffic you never typed anywhere. Exclusions do not work that way, so blocking a word does not block its plural, and every exclusion you write has to be written more than once.

Where the clicks land

Sending paid traffic to your homepage is a structural error rather than a tactical one, and fixing it is worth more than any bid adjustment you will ever make.

A homepage answers the question "who are you". Somebody who typed a specific job in a specific city is asking a different question entirely, and every extra second spent hunting for the answer costs you a share of them. The gap between those two questions is where a lot of otherwise reasonable accounts lose their money.

The lower-effort entry points, and where they stop

Search is not the only door. Two other paths demand less skill up front, and both demand something else instead. Anyone weighing this decision deserves to know what the trade actually is.

Local Services Ads: pay per lead, but you still answer the phone

With Local Services Ads you are charged per valid lead rather than per click. Google says it assesses leads at initial contact, does not charge those it determines are invalid or low quality, and may automatically credit charged leads that are later reassessed as low quality. For a contractor that removes much of the keyword-auction work from the equation, which is an advantage and the reason this can be a useful first channel.

The catch is in Google's own Canadian documentation: if you regularly fail to answer calls or respond to messages, your ad ranking may be affected. So the hands-off channel is not hands-off. It trades campaign management for daily responsiveness, which is a fair trade for some businesses and an impossible one for others. There is setup work too, since the verification and badge process has to be completed before anything runs, and some lead management tooling is still US-only.

Smart campaigns: simpler, and Google is steering elsewhere

Smart campaigns use keyword themes instead of keywords, and Google recommends a maximum of seven to ten themes. Their negative keyword themes do not match close variants, and Google warns that adding too many can shrink reach. Simpler, then, but not free of the mechanics above.

Performance Max is not the easy option it looks like

It is the campaign type Google now points new advertisers toward, and it hands over more of the decision surface rather than less. That is only simpler if the goal you handed it is the right one. Give it a wrong definition of a lead and it will pursue that definition across more of Google than a Search campaign ever could.

What hiring someone actually costs, and why nobody can tell you

The three structures, and what each one rewards

Fees come in three shapes. A percentage of ad spend, a flat monthly fee, or a hybrid with a floor underneath it. Percentage models almost always carry a minimum, because a small account costs roughly as much to run as a medium one. The settings, the reviews and the reporting are the same work either way.

Each shape creates a different incentive, and none of them is dishonest. A percentage ties the fee to spend growth, which is fine when growth is warranted and awkward when you want to cut back for a slow season. A flat fee is predictable and rewards nothing in particular, which is mostly a virtue. Asking a candidate which structure they use and why is a fair question, and the answer tells you more than any case study will.

Every published fee range comes from someone selling the service

Search for how much Google Ads management costs and you will find confident ranges everywhere. Follow them back. They trace to agency marketing pages with no stated sample, no methodology and no geography, citing each other in a loop until the number acquires the feel of a fact.

That does not make any particular figure wrong. It makes all of them uncheckable, and a number you cannot check is not evidence, whether it comes from an agency page or from me.

The arithmetic that actually decides it

The comparison that matters is the fee against two things: waste avoided, plus hours reclaimed multiplied by what an hour of your attention actually earns. Run it over a year, not a month, because the fee recurs and so does the waste.

For some readers that arithmetic genuinely favours doing it yourself. At a small budget a management fee can exceed the waste it prevents, and no amount of skill on the other side of the invoice changes that. Anyone who will not say so plainly is not worth listening to when they argue the other side of it.

An audit and ongoing management solve different problems

A one-time audit is useful when the question is structural. Is tracking recording the intended actions? Are locations constrained to the service area? Do the searches and landing pages describe the same job? Those are conditions that can be inspected at a point in time, documented, and handed back to the owner.

Ongoing management addresses changes after that snapshot. Search phrasing moves, competitors enter or leave auctions, the mix of services changes with the season, pages are edited, and phone handling changes when the office gets busy. An audit cannot own those future decisions. A monthly arrangement is paying for repeated judgment and accountability, not for the initial list of settings.

That distinction creates a useful middle ground. A contractor with a simple account and protected time may need a careful setup review and then keep control. A contractor with several services, locations, and seasonal peaks may have no structural emergency at all but still need somebody watching the moving parts. The correct purchase follows the problem. Buying ongoing management for a bounded setup issue can be wasteful; buying a one-time audit when nobody will monitor the account afterwards can leave the original operating problem untouched.

Whichever route you choose, the client account, direct access, and conversion history should remain usable without the provider. That preserves the option to move between self-management, periodic review, and ongoing help without throwing away the evidence already accumulated.

What a workable handoff needs from both sides

Hiring someone does not remove the business owner's role. The manager can see searches, clicks, configured conversions, and the account's change history. The contractor still knows which callers were qualified, which quotes were serious, which jobs booked, and which work was profitable. Without that feedback, the account can become efficient at producing the easiest conversion rather than the best work.

A workable arrangement makes that boundary explicit. The manager owns the advertising decisions and explains material changes. The owner supplies lead outcomes and flags changes in capacity, service area, pricing, or the services worth promoting. Neither side has to pretend the platform contains information that only exists in the estimating system or job calendar.

Access matters for the same reason. The contractor should be able to enter the client account directly, see the Google cost separately from any management fee, and retain the history if the relationship ends. A provider may use a manager account to do the work, but that convenience should not turn into practical dependence on one login or one company.

The reporting conversation should also use business outcomes rather than activity as its finish line. Search terms reviewed, ads written, and settings changed describe labour. Leads and booked jobs describe the result of that labour. A useful review needs both: enough detail to show what was done and enough outcome data to decide whether the work is earning its place.

That still leaves uncertainty. A low-volume account may not produce a stable answer every month, and one large job can distort a short window. A competent manager should be able to say when the evidence is thin instead of turning normal variation into a confident story. Paying for judgment includes paying for restraint when the account has not produced enough information to support a change.

It is not a permanent fork

The question is almost always asked as a fork. Do it yourself, or hand it over. That framing is tidy and it is wrong. Most of the useful ground in this decision sits in the middle, and hardly anything written about it says so.

Run it yourself and buy a set of eyes

A one-off review of an account you built is a much smaller purchase than a monthly management relationship, and it aims at exactly the class of problem described further up this page. Structural errors are expensive because nothing surfaces them, and you cannot find a problem you do not know to suspect.

Someone who has seen the same failures repeatedly can review tracking, targeting, and destinations quickly because they know where structural errors tend to hide. You keep the account and login, without taking on a monthly management relationship. A periodic review can be useful, especially after significant website or account changes, but there is no universal cadence.

Split it by season

Run it yourself through the quiet months and hand it over for the stretch where the account matters most and you have the least attention left. The hours you cannot supply are seasonal, so the help can be seasonal too.

Two cautions make this work better. Plan the handover before you need it, because an emergency handover in week two of your busy season starts with somebody learning your account instead of running it. And remember the calibration clock: arriving in the middle of peak and immediately changing several things means the account spends your best weeks recalibrating.

Start on pay per lead and graduate

Local Services Ads ask less of you up front, and running them first teaches you something a Search campaign will charge you to learn: what a lead in your trade is actually worth to you, and how many of them you can handle in a week. That number is the input everything else depends on, and you cannot look it up.

The limits are real. Categories are restricted, your volume is capped by however many people search near you, and the badge process has to be finished first. So comparing the two channels honestly is worth doing before you assume one replaces the other. If your real question is which channel deserves the next hour of your attention rather than which one you should manage, that is a different comparison again.

The one input you cannot hand to anyone

Whichever way you decide, one job stays with you. It is free, it takes minutes a week, and it moves an account further than any setting in it.

Conversions are not jobs

A conversion is somebody who contacted you. A job is somebody who paid you. Those two numbers move independently, and the gap between them is where most disappointment with paid advertising actually lives.

An account optimized toward conversions with no idea which ones became work will get very efficient at buying whichever leads are cheapest to produce. That usually means price shoppers, tyre kickers, and people well outside the kind of work you want. Everything on the screen improves while the business does not.

What to write down

Six columns. Date, how they found you, what they wanted, whether you quoted, whether it booked, and what it was worth.

It does not require specialist software or a CRM. A consistent record beside the phone can be more useful than a spreadsheet nobody opens. The amount of history needed depends on lead volume and how much one unusually large or small job can move the result.

Why this changes the account more than any setting

Nobody outside your business knows which leads were profitable. No platform can infer it, because the information never leaves your side of the transaction. Feed it back and the same budget starts buying different work. Do not feed it back and every adjustment anyone makes, you or somebody you pay, is aimed at a proxy for the thing you actually care about.

This is also the honest answer to the fear that a manager will not care about lead quality. Nobody can care about it usefully without that record, including you. If you want the same loop written out for a single trade, the plumbing lead guide walks through it end to end.

The decision: complexity, evidence, and available time

The decision rests on how much judgment the account creates, how quickly the budget produces evidence, whether time remains during the busy season, and whether lead quality is recorded outside the ad platform.

Count your decision surface, not your budget

Complexity rises with each added service, area, landing page, and contact path. A contractor advertising one service in one place faces a different management job from one spreading demand across a metro and several job types.

A small product is a small problem at any spend. A large one is a large problem at any spend, and no budget figure rescues it.

Estimate how quickly evidence can accumulate

The same budget produces different amounts of evidence in different trades. Published category medians can bound a scenario, but they are US medians in US dollars and they do not predict a Canadian account.

If the answer is a handful of conversions a month, that is not a failure. It just means your evidence arrives slowly, so decisions have to be made carefully and changes have to be given room to resolve.

Count your hours in July, not in February

Every DIY plan gets built in the off-season, which is the one time of year the plan is guaranteed to look easy. Count the hours you will have in your busiest month, and be honest about where they would come from. If the answer is quoting and selling, you have found the real price of doing it yourself.

If the account is complex and the busy season leaves no time for recurring review, management needs to be somebody's job rather than somebody's evening. If the scope is narrow, evidence is interpretable, and you can protect the time, running it yourself can be the better call. In either case, verify measurement, search intent, geography, destination quality, and booked-job outcomes before treating dashboard performance as business performance.

If the conditions point toward paid help, I take management on as a standing job rather than a setup project, with measurement and account structure addressed first. You can see how I work with contractors or read the Greater Toronto Area overview if you want the local version first.