Your cheapest Google Ads leads may be people who already knew your name

TL;DR

• Google Ads brand vs nonbrand reporting needs two separate checks: whether the search named your business, and whether the customer had hired you before. One does not establish the other.
• A first-time homeowner can search your name after a referral. A returning customer can search “AC repair near me.”
• Keep unreported queries and customers with missing history in an unknown group. Assigning them to nonbrand or new makes the report look more complete than it is.
• Reconcile completed, collected jobs by job ID before calling lower lead cost an acquisition improvement.

What does a search for your name actually prove?

A homeowner gets your name from a neighbor. She searches it, clicks your ad, and books a water-heater replacement. She is a new customer. The search is branded. Advertising may have helped her reach you, but the search itself does not tell you who created the demand.

Now take a homeowner whose furnace you repaired last winter. He searches “furnace repair near me,” recognizes your name in the results, and calls again. The search is nonbrand. The customer is returning.

Those two calls disappear into the same conversion total unless someone keeps the distinctions. An agency can correctly report that the ads received both calls and still be unable to say that both are new business created by advertising.

There are three questions here: what the homeowner searched, whether you had served that customer, and whether the advertising caused a job that would otherwise not have happened. Your records can help answer the first two. Answering the third requires a comparison beyond the attribution report.

The Google Ads and targeting hub covers the settings. This article is a reporting worksheet, so you can ask for the split before changing a campaign.

Build four groups before comparing Google Ads brand vs nonbrand

Use this table with one row per acquired customer inquiry, then connect each inquiry to its eventual job. Count repeat calls about one appointment together. Keep a separate count of contacts if answering workload matters.

Add unknown search type and unknown customer history as explicit categories. A shop with records only from its current CRM cannot promise that a person absent from that CRM has never used the business. Mark the history as “new in available records,” and state when those records begin.

Classify a search as your brand only when it names your business or an identifiable spelling variant. “Carrier repair” names a manufacturer. “Best plumber” names no business. A word that appears in your company name and in ordinary service searches needs a manual check; the word alone is not sufficient.

The search terms report documentation, checked October 8, 2026, distinguishes the actual search from the keyword that matched it. It also says some low-activity queries are withheld for privacy. A campaign named “Nonbrand” is therefore not evidence that every search inside it was nonbrand.

When you cannot connect a reported query to an individual inquiry, keep the search split at the aggregate level. Do not manufacture a row-level join from matching dates or approximate call counts. You can still report known new customers by campaign; just state that query and customer history are not connected at that level.

How can cheaper leads hide fewer new jobs?

Here is an invented example, with two equal-length periods and the same $6,000 ad spend. “Inquiries” means deduplicated customer inquiries, not every phone call. All jobs shown have completed and been collected by the same follow-up cutoff.

The headline lead cost falls by a third. Total jobs rise. Yet fewer first-time customers complete work, and the ad spend divided by those jobs rises from $240 to $400.

That last calculation assigns all ad spend to the new-customer outcome. It is useful as an account-wide acquisition measure, but it is not the cost of a particular nonbrand campaign. Returning customers also bought work. Their contribution belongs in the business result even when they do not count as acquisition.

You still cannot conclude that the extra branded inquiries caused the decline in new customers. The table identifies a change worth investigating, not its cause. Check the search mix, customer mix, answered hours, service area, job type and collection cutoff before making a bid decision.

For why a lead-only target can misread the work itself, see cost per lead versus cost per sold job. For the campaign question, see advertising to existing customers.

Which records belong in the monthly review?

Start with three exports: advertising cost, customer inquiries, and CRM jobs. Keep customer-level records in your own systems. The public worksheet contains no customer data and asks for internal IDs rather than names, phone numbers or addresses.

Advertising: period, campaign ID, spend, visible brand-query spend, visible nonbrand-query spend, and spend whose query is not reported. These groups must reconcile to the campaign total for the same filters. If the export cannot support that reconciliation, state the gap.

Inquiries: inquiry ID, first contact date, tracked campaign where available, search classification where genuinely connected, customer ID, and customer-history classification. Use unknown when tracking is missing. One caller who phones three times about the same repair should not become three acquisition wins.

Jobs: job ID, linked inquiry ID where available, job type, completion status, collection status and collected amount. A replacement sold this month but installed next month is a different stage from a collected repair. Compare the same stage in both periods.

Google's conversion measurement guide, checked October 8, 2026, describes conversions as actions an advertiser chooses, including calls and purchases. That is why the label “conversion” alone cannot establish that a technician completed work. Ask which action the report counted.

Choose a follow-up cutoff that fits your sales cycle and write it beside the comparison. Use inquiries first received in each period and follow both groups for the same number of days. Do not compare a mature group from March with last week's leads while half their estimates are still open.

Should you stop buying your own name?

This worksheet does not make that decision for you. Brand ads may help a customer find the right phone number or reach a specific service page. They may also receive credit for a job the customer would have booked through another route. A branded conversion total cannot measure the difference.

Google documents brand inclusions for Search and brand exclusions for Search and Performance Max in its brand settings guide, checked October 8, 2026. Those are controls for traffic, not proof of incremental sales. Google also cautions that brand restrictions can limit conversion opportunities. Do the reporting split first; then decide whether a controlled test is warranted.

If the proposal is to move money into generic service searches, check dispatch capacity and the HVAC campaign structure as well. A first-time customer forty miles away is not automatically better work than a returning customer around the corner.

What this doesn't cover

• Incrementality. A tagged call shows an advertising interaction. It does not tell you what would have happened without the ad.
• Complete customer history. Merged companies, older paper records and changed household details can defeat a new-versus-returning label.
• Every search. Queries withheld by the platform remain unknown. The visible split is not a complete census.
• A universal brand budget. This article contains no recommended percentage and no instruction to turn a campaign off.

Do this in your account

1. Download the brand and customer-history worksheet. It is plain text: copy it into your reporting document, with no signup.
2. Write the review period, CRM history start date, job stage and equal follow-up cutoff at the top.
3. Ask for advertising costs split into visible brand, visible nonbrand and unreported query spend. Reconcile the totals.
4. Count deduplicated inquiries, then completed and collected jobs for new, returning and unknown-history customers.
5. Cross the two classifications only where the records support a real connection. Mark aggregate-only reporting clearly.
6. Compare account-wide spend per new-customer job alongside total collected contribution. Keep both; they answer different questions.

The first review can be a sample of twenty inquiries to check whether the records join. Label it a sample. Twenty rows can expose a tracking problem; they cannot establish the customer mix of an entire account.

FAQ

Is a branded lead always an existing customer? No. Someone referred by a neighbor can search your name before hiring you for the first time. Use customer history to classify the customer and the actual query to classify the search.

Can a nonbrand campaign include branded searches? Check the actual searches rather than relying on the campaign name. Google's search terms report documents the distinction between a keyword and the query that triggered it. Queries missing from the report stay unclassified.

Does nonbrand traffic prove the ads created new business? No. A returning customer can use a generic service search, and attribution does not establish what would have happened without advertising. Keep customer history and causal claims separate.

What if the CRM and ad report do not join? Report each at the level the records support, and name the unmatched count. Fix the connection before presenting an individual search-to-job story.

Sources

Checked October 8, 2026. The four-group worksheet and numerical example are our reporting framework and invented arithmetic, not a Google recommendation or client result.

• Google Ads Help: About the search terms report.
• Google Ads Help: About brand settings for Search and Performance Max.
• Google Ads Help: About conversion measurement.