On a small local budget, Search is the only one of Google's two campaign types you can read

What the two campaign types actually buy

A Search campaign and a Performance Max campaign are sold as alternatives and paid for out of the same small budget, but they are not the same kind of purchase. In a Search campaign you buy a list of queries. You write the keywords, you decide how loosely each may match what somebody types, and you can read the search terms that triggered your advert. In a Performance Max campaign you buy an outcome and let Google decide where to pursue it.

Google's own description is not coy about the breadth of that second bargain. Performance Max is "a goal-based campaign type that allows you to access all Google Ads inventory", the documentation says, "including YouTube, Display, Search, Discover, Gmail, and Maps, from a single campaign". The advertiser supplies text, images and videos, defines the conversion actions it cares about, and sets an optional target cost per action ^[1].

A Search campaign offers the opposite trade. Google describes exact match as the type that "gives you the most steering over who views your ad", while broad match, "the default match type that all your keywords are assigned", may show adverts on searches merely "related to your keyword" ^[2]. Control and reach are priced against each other, and the advertiser picks a point on that line.

I ran both for my own clinic on a small daily budget, a Performance Max campaign alongside a Search campaign built on high intent local terms, and read the click and conversion data myself rather than the summary on the front page of the interface. The useful question turned out not to be which type performs better in the abstract. It was how much money each one needs before its numbers mean anything.

Which campaign serves when both of them could

Because Performance Max can serve on Search, an account running both has two campaigns eligible for the same query. Google resolves this with one hard rule and one soft one, and the difference is where most of the confusion lives.

Search campaigns containing an exact match keyword are prioritized to serve over Performance Max

The documentation on campaign interaction puts the rule in operational terms, and its worked example is a local one: for the misspelled search term "plumbrs", with both an exact match keyword for plumbers and a broad match keyword in the account, "the exact match keyword is preferred" ^[3].

That is the hard rule, and it is narrow. Where no keyword matches the query exactly across any match type, "the campaign with the higher Ad Rank will serve" ^[3]. Google frames the mechanism as a protection, so campaigns in one account do not "compete against each other to drive up bids" ^[3]. Phrase match and broad match keywords hold no reserved place. Search themes, the query hints an advertiser adds inside Performance Max, "will have the same prioritization as your phrase match and broad match keywords", with up to 50 allowed per asset group ^[4].

Read together, those pages say something a small advertiser should take seriously. Exact match keywords are the only part of a Search campaign that Google's published rules protect from the automated campaign in the same account. The rest is settled by Ad Rank, which the advertiser cannot see.

What the overlap evidence shows, and what it does not

The best public attempt to measure that overlap comes from Optmyzr, which sells Google Ads software and so has an interest in the answer. Its study covered February 2025 and used accounts managed in its own platform ^[5].

The first part checked whether the text of a Search keyword matched the text of a Performance Max search term, regardless of match type. Overlap appeared in 91.45 per cent of 503 accounts and in 56.29 per cent of 5,768 Search campaigns ^[5]. A second part looked for search terms appearing in both reports without an exact keyword match and found overlap in 97.26 per cent of 511 accounts ^[5].

The performance comparison is more sober than the overlap figures suggest. On click-through rate, the Search campaign was the better performer in 28.37 per cent of cases, Performance Max in 15.98 per cent, and there was no significant difference in 55.65 per cent ^[5]. On conversion rate, Search was better in 18.91 per cent of cases, Performance Max in 6.17 per cent, and 74.92 per cent showed no significant difference ^[5].

Three caveats belong with those numbers. The study treated differences under ten per cent as insignificant, a generous band. It excluded cost per click and return on ad spend entirely, because Google did not report cost data for Performance Max search terms at the time ^[5]. And the authors are candid that an overlapping query might simply mean the Search campaign was ineligible to show, noting that "we did not analyze that possibility in this study" ^[5].

So the honest reading is not that Performance Max steals Search traffic. It is that the two campaigns were competing for the same queries in almost every account examined, that the exact match protection did not prevent it, and that where a difference was large enough to measure, the keyword campaign was more often the better one. For a large advertiser that is an optimisation problem. For one spending a few hundred pounds a month it is a sequencing problem, because the same money cannot fund both learning curves at once.

Why a few hundred pounds a month behaves differently from a few thousand

Automated bidding is a statistical process, and statistics need events. This is the part of the decision that budget size settles on its own, with no need to form a view about Google's intentions.

What Google asks for in time

For Performance Max, the documentation asks advertisers to "run new campaigns for at least 6 weeks" and to "avoid making frequent changes to budget, bid strategy, or campaign status within this initial period" ^[7]. The guidance for store goals campaigns, the variant closest to a local business, recommends a minimum duration of 30 days and advises advertisers to "wait for 1-2 complete conversion cycles before fully evaluating your campaign performance and adjusting targets" ^[8].

What Google asks for in conversions

Performance Max bids through Smart Bidding, which Google describes as "real-time auctions bid based on your conversion probability" ^[1], and Smart Bidding has thresholds of its own. Google states that "to use Google Ads Smart Bidding, you need to have conversion tracking enabled", that "some Smart Bidding strategies rely on a minimum volume of historical conversion data", and that performance should be measured "over longer time periods that have at least 30 conversions" ^[9]. Its lead generation guidance tells advertisers to pick the conversion goal closest to a closed sale, provided it "has at least 15 conversions in the last 30 days" ^[10].

What a small budget actually buys

Set those requirements against the price of a click. LocaliQ's published benchmarks, drawn from its own customers' Google and Microsoft Ads campaigns, give an all industry average cost per click of $5.42, an average conversion rate of 8.18 per cent and an average cost per lead of $66.69, rising to $9.87 per click and $131.63 per lead for legal services ^[11]. The figures are in dollars and LocaliQ does not state its sample size, period or countries, so they are an order of magnitude rather than a forecast.

Even as an order of magnitude the arithmetic is unforgiving. A campaign funded at £10 a day, which Google says "might spend up to twice your average daily budget" on any given day while never exceeding "30.4 times your average daily budget" over the month ^[6], buys a few dozen clicks in a sector priced near those benchmarks, and a conversion rate in single figures turns that into a handful of enquiries ^[11]. Reaching the 30 conversions Google suggests for a reliable read ^[9] can therefore take a quarter rather than a fortnight, and the six week learning period consumes most of the first half of it ^[7].

The market data points the same way. Optmyzr's study of 21,425 accounts across five quarters to the first quarter of 2026 found Performance Max campaign volume up 15.7 per cent year on year, with its cost per click rising 5.66 per cent against 1.36 per cent for Search ^[12]. Accounts spending under $10,000 a month returned 387.8 per cent return on ad spend, against 565.8 per cent for the $10,000 to $50,000 tier ^[12]. Return on ad spend is an ecommerce measure that most local service businesses never record, so the signal worth taking is the ordering rather than the percentages.

The measurement problem sitting underneath both options

Everything above assumes the conversion data is correct. That assumption fails quietly and often, and it punishes the automated campaign far more than the keyword one.

I have watched a broken conversion tag make a working campaign look like a failing one in my own account. The adverts kept serving, the phone kept ringing, and the reports went flat. What follows is inference rather than something Google publishes, though it follows from what it does: a bid strategy that needs conversion tracking enabled and leans on historical conversion volume ^[9] cannot tell a campaign that stopped producing enquiries from a tag that stopped recording them. A Search campaign bid manually on clicks keeps working through the same outage. A Performance Max campaign optimising towards a conversion signal has lost the only thing it was told to pursue. Practitioners report the mirror image of the problem in lead generation, where the recorded conversion is real but worthless: without data on which enquiries were any good, Google's systems are "prone to bringing in a lot of spam", since the platform "can only measure how many phone calls or form submissions you get" and not which ones became customers ^[13].

In the United Kingdom there is a second, legal reason the data is incomplete. The Information Commissioner's Office states that regulation 6 of the Privacy and Electronic Communications Regulations "applies whenever the use of scripts and tags accesses or stores information on a user's device", and that an organisation's "use of JavaScript tags for non-exempt purposes (eg advertising) requires consent under PECR" ^[14]. Advertising measurement is the regulator's own example of a purpose needing consent, so whatever share of visitors decline is a share the account cannot see.

The opacity is structural as well as technical. The Competition and Markets Authority found that Google "generated over 90% of UK search advertising revenues in 2019", in a market it valued at over £7 billion, and observed that "platforms have considerable discretion over a wide variety of parameters that affect the prices advertisers pay" while "market participants find it difficult to understand or challenge how decisions are made" ^[15]. That report is from July 2020, so read it as a description of the market's structure rather than of this month's auction.

What an advertiser can now see and control that it could not

Much of the case against Performance Max was built when it was close to opaque. Several of those objections have since been answered.

• Campaign level negative keywords were launched to all Performance Max campaigns ^[18]. They can be applied to every campaign or to one, and Google is explicit that they are "applicable to Search and Shopping inventory only" ^[16].
• Search terms reporting is available for all Performance Max campaigns and through the API, and the number of search themes allowed per asset group rose from 25 to 50 ^[18]. Google said in April 2025 that advertisers would get "the same granularity of Search reporting you do for Search and Standard Shopping campaigns" ^[17].
• Channel performance reporting arrived as a beta with a campaign level summary and a downloadable channel table ^[17], though Google's guidance still tells advertisers to "connect with your account manager to apply for the channel performance report beta" ^[20].
• Version 23 of the Google Ads API split the mixed network category for Performance Max into named channels including Search, YouTube, Display and Maps, with data only from 1 June 2025 and asset group channel reporting available through the API alone ^[19].

That is a real narrowing of the gap, and it is not equality. The controls sit at campaign level while the decisions are taken at auction level, and a reporting beta that requires an account manager to apply for it ^[20] is out of reach of a business spending a few hundred pounds a month, because that business does not have one. The advertiser most dependent on automation has the least visibility into it.

The compliance question a regulated advertiser should ask first

One setting in Performance Max matters more to a clinic, a dental practice or a solicitor than any bidding decision, and it is on unless somebody turns it off.

Google describes text customisation as a feature that "allows Google to curate text for headlines and descriptions from your landing page(s)", and says that Google "may automatically add a dynamic headline, description, or additional text assets". Such text "will only serve if they're predicted to perform better than the headlines and description lines you uploaded". Crucially, "text customization is turned on by default", and an advertiser who wants it off must uncheck the box during setup or in the campaign's settings ^[21].

Advertising rules do not soften because a machine wrote the wording. The CAP Code applies to "paid-for search listings" ^[23], and its first section is unambiguous about where the duty sits: "Marketing communications must comply with the Code. Primary responsibility for observing the Code falls on marketers." ^[22] Agencies and publishers also accept an obligation to abide by it, but the advertiser's responsibility is not transferable ^[22].

The practical consequence is straightforward. If text customisation is left on, the campaign may serve headlines assembled from the landing page, and the business remains accountable for every claim in them. Two responses are defensible: turn the setting off and write the headlines, or leave it on and treat the landing page as advertising copy already checked line by line, since Google says that page is the source it draws from ^[21]. Leaving it on while assuming somebody else owns the risk is not.

Whether there is a third option for some businesses

For part of the local market the choice is not between two campaign types but among three. Google says the Local Services unit "is shown prominently in Google Search results", and that advertisers "pay only for leads related to your business and the services that you offer" ^[24].

Eligibility is the limiting factor. Google states that "at present, Local Services Ads are available only for certain service categories", and the United Kingdom list is weighted towards home service trades including electricians, plumbers, roofers and carpenters, while lawyer, estate agent and personal injury services are listed as currently available in Greater London only ^[24]. The Google Verified badge goes to businesses that have "passed Google's proprietary screening process" ^[24].

So a Bradford electrician has three options to compare and a Bradford clinic has two, because its category is not on the list ^[24].

What it means for a local business

The question as usually posed, Performance Max or Search, assumes the advertiser gets to choose. At a few hundred pounds a month the budget has largely chosen, because only one of the two produces a readable result inside a month. A defensible order of work looks like this.

1. Fix measurement before comparing anything. Confirm that calls, form submissions and booking clicks each record a conversion, and recheck after every website change, because a bid strategy that requires conversion tracking to be enabled ^[9] is only as good as the tag.
2. Start with Search on the terms you would be embarrassed to miss, as exact match. That is the only placement Google's published rules give you first claim on while an automated campaign runs in the same account ^[1][3].
3. Treat broad match as spending rather than reach. It competes with Performance Max on Ad Rank with no reserved priority ^[3], and the search terms report is where you find out what it bought.
4. Add Performance Max as a funded decision, not a leftover. Google asks for at least six weeks without changes ^[7] and a conversion goal with recent volume ^[10], so give it that runway or leave it until enquiries support it.
5. Turn text customisation off unless the landing page copy has been checked as advertising ^[21], and keep the negative keyword list maintained at both account and campaign level ^[16].
6. Judge both on enquiries that became customers rather than conversions recorded. Google's lead generation guidance says "your inputs control the output" and advises optimising to deeper funnel actions where possible ^[10].

None of this promises a position on a results page or a volume of enquiries, and no honest account of the subject can. It says which of two purchases a small budget can actually evaluate, and in what order.

What remains uncertain

Several things here are genuinely unsettled, and a reader making a decision should know which.

The overlap study measured a single month in early 2025, before campaign level negative keywords and search terms reporting reached every Performance Max campaign ^[5][18]. Those controls could plausibly reduce the overlap it found, and no repeat has been published at the budgets a UK local business runs.

Google publishes no minimum budget or conversion volume for Performance Max. Its guidance is directional, telling advertisers to ensure a campaign is "not limited by budget" and that "the more conversion data that you have, the faster Google AI is able to deliver the results" ^[20]. The thresholds used above are inferred from its learning period and measurement guidance ^[7][9], not stated as requirements.

The public cost benchmarks are vendor data from vendors' own customers, in dollars ^[11][12]. No authoritative figure exists for what a new enquiry costs a UK local business by sector and region, so every business is measuring its own.

Transparency is still partial. Channel level performance remains part allowlisted beta and part API only, with no asset group channel view in the interface ^[19][20], so the plain question of how much of a Performance Max budget went to Search cannot be answered by everyone paying it.

Finally, the measurement gap created by consent has no published size. The regulator's position that advertising tags require consent ^[14] guarantees a difference between what happened and what the account recorded, and that difference is not knowable from inside the account.