Branded Searches in Google Ads: Measuring the Demand YouTube and Demand Gen Create

Every team that runs video has the same argument. Brand says the campaign worked and demand went up. Performance opens the report, sees three last-click conversions, and says it did not. Both are right inside their own frame, and the argument never ends because there is no shared number. Since 2026 there is one in Google Ads — branded searches — and it covers exactly the gap between “saw the ad” and “showed up ready to buy”.

This is a working guide: what the metric counts, which campaign types support it, how brand mapping gets configured, where the columns hide in the interface, why you cannot bid on it, and how to put it into reporting without turning it into one more pretty number nobody trusts.

What branded searches count

Branded searches is the number of searches for your brand on Google and YouTube made within 7 days of a user viewing your eligible ads. Not a click, not an on-site conversion — a search the person chose to perform after exposure.

• Eligible campaigns: YouTube and Demand Gen only. Search, Performance Max and Shopping do not have it, which makes sense: this measures demand creation, not demand capture.
• Attribution: last-click, view-through. The last ad the user saw takes full credit, and only one campaign is credited even when several contributed.
• Window: 7 days by default, editable anywhere from 1 to 30 days.
• Availability: always on. No Search Lift experiment is required to produce the number.
• Status: the conversion action is flagged Primary, but it stays reporting-only — it cannot be used for bidding.

Brand mapping comes first

For the system to know which queries count as yours, brand mapping has to be configured. It is requested through your Google sales representative, not toggled in the UI. After setup the data is not instant: branded searches can take up to 7 days to appear on the Conversions page and in campaign reports.

Things worth covering in that request:

• Spelling variants. Spaced and unspaced versions, common misspellings, “brand official site”, the app name if it differs from the company name, transliterations if you operate in non-Latin markets.
• Sub-brands and product lines. If a product line is searched more than the parent brand, it needs to be accounted for, otherwise the metric understates the real effect.
• Competitor terms. Comparison queries (“brand A vs brand B”) deserve an explicit decision, because how they are classified changes how you read the result.

Until mapping exists, the columns can legitimately read zero. That is a configuration finding, not evidence that video does not work.

Where the numbers hide

The most common complaint is “we turned it on and the report is empty”. Almost always it is a question of where you are looking:

• Branded searches land in All conv. and Results, not in the Conversions column.
• In the campaigns table you surface them by segmenting by conversion action.
• If All conversions shows zero, add View-through conversions and Results — the metric is view-through by nature.
• Breakdowns are available at campaign, ad group and asset level, and in the report editor, which is where you want them if you maintain a recurring export.

If you run custom column sets, build an “upper funnel” set right away: impressions, unique reach, frequency, views, view-through conversions, branded searches. The mechanics are in our guide to custom columns and the report editor.

Why reporting-only is the right call

The branded searches action is marked Primary but cannot drive bidding, and the Consideration goal cannot be added to a campaign for optimization. At first that reads as an arbitrary restriction. It is actually protection against an expensive failure mode.

Suppose bidding on branded searches were allowed. The system would quickly find the cheapest way to manufacture them: show ads to people who were going to search for you anyway — existing customers, subscribers, yesterday’s site visitors. The metric would soar, cost per branded search would collapse, and incremental demand would not move by a single unit. It is the same trap as retargeting optimized on ROAS: beautiful numbers, zero lift.

So treat the metric as diagnostic. It answers “is this campaign creating interest in the brand”, not “where should tomorrow’s budget go”.

Putting it into reporting

Step 1: normalize

An absolute count means nothing without a base. Derive two numbers:

• Cost per branded search = Campaign spend / Branded searches
• Search rate = Branded searches / Unique reach

The first compares campaigns and creatives against each other. The second shows how efficiently exposure turns into interest, which is as close to a creative quality score as this channel offers.

Step 2: compare the right things

Valid comparisons: creative A versus creative B on the same audience; the same cut across different lineups; different durations. Invalid: YouTube versus a Search campaign (Search has no such metric at all), peak season versus off season, a campaign running before brand mapping against one running after.

Step 3: do not stack measurements

Brand Lift, Search Lift, branded searches and MMM answer different questions and overlap. Add their effects together and you produce a number no CFO will accept. Keep the roles separate:

Instrument	Question it answers	Needs an experiment
Branded searches	How many people searched for the brand after exposure	No, always on
Brand Lift	Did brand perception shift against a control group	Yes
Search Lift	Did search activity rise against a control group	Yes
Geo experiment	What is the incremental effect on sales	Yes
MMM	How channels interact over months	No, modelled

The last two are covered in incrementality and geo experiments and marketing mix modelling with Meridian; the classic studies in Brand Lift and Search Lift on YouTube.

Decisions the metric actually changes

• Creative selection. Two cuts with identical CPV and identical view rate can differ several times over on branded searches. The difference is nearly always whether the brand is named in the first seconds and whether it is memorable. This is the rare metric that points directly at an edit.
• Audience sizing. Broad audiences often produce more branded searches per dollar than a narrow “core” segment, simply because the core already knows you. If you see the opposite, suspect frequency and reach problems rather than targeting.
• Defending upper-funnel budget. When cuts come, video goes first because it has no legible number. Branded searches give you one: this many people started looking for us, and this is what it cost. That matters most for expensive guaranteed buys such as YouTube reservation campaigns, which cannot be judged on CPA at all.

One more useful cross-check: line the branded search trend up against brand query volume in your Search campaigns and in Search Console. If video is generating brand demand, brand impressions should rise one to three weeks later. If they do not, either the attribution window is too short or the effect is smaller than it looked. The paid-and-organic view is covered in our piece on the paid and organic report.

What a usable flight report looks like

The most common upper-funnel reporting failure is mixing incomparable quantities in one table. Separate the blocks: delivery first, then response, then impact. A monthly flight, illustrative numbers:

Block	Metric	Creative A	Creative B
Delivery	Spend	$12,000	$12,000
	Unique reach	1.4M	1.35M
	Average frequency	3.1	3.2
Response	Branded searches	9,800	5,400
	Cost per branded search	$1.22	$2.22
	Search rate (of reach)	0.70%	0.40%
Impact	Brand query impressions, +2 weeks	+18%	+6%
	Lower-funnel CPA during flight	−9%	−2%

That table settles the argument without anyone raising their voice: at equal delivery, creative A produces a branded search at roughly half the cost, and an independent signal — brand query impressions two weeks later — moves in the same direction. One metric could be noise. Two agreeing metrics is an argument.

Note why the comparison is fair: delivery is nearly identical. Had creative B reached half as many people, the raw branded search counts would be meaningless and only the derived rates could be compared.

When brand mapping is not available

Mapping goes through a sales representative, and not every account has one. Until the metric exists, you can approximate from open signals — less precise, better than nothing:

• Search Console. Filter queries containing the brand name and track impressions (not clicks — impressions reflect interest more directly) by day against the flight dates.
• Your brand search campaign. Exact-match brand impression volume is a proxy for the same signal. Watch impression share, or a competitor’s budget increase will read as demand growth.
• Direct traffic and on-site search. A jump in direct sessions during the flight, with other sources flat, is corroborating evidence.
• A geo split. The most reliable option available without Google’s help: run video in half your regions and compare brand search behaviour between exposed and held-out markets. That is a real experiment, not a proxy.

When mapping does arrive, keep the old series. Comparing before and after tells you how good your proxies were, and ends the perennial argument about whether upper-funnel work was doing anything in the months before you could measure it.

How to misread it

• Treating a branded search as a sale. It is intent. Everything between intent and purchase can still break.
• Putting it in the same table as conversions. Different event type, different attribution, different window. Give it its own block in the report.
• Forgetting single-campaign credit. In a three-campaign flight, credit goes to whichever ad was seen last, not to whichever did the work. Compare campaigns inside one flight with that in mind.
• Stretching the window to 30 days to get a bigger number. Bigger is not more accurate. The wider the window, the more coincidence you absorb — people who would have searched anyway.
• Ignoring PR and seasonality. A press mention, a collaboration or a viral post lifts branded search regardless of media. Without controlling for the background, the metric hands your advertising credit it did not earn.

Explaining the metric to a client or a board

Being technically right about the metric does not help if the other person hears “another number that does not reach the bank account”. Three moves that defuse that:

• Call it what it is. Not a conversion — an intent signal. “After seeing the campaign, 9,800 people went and searched for us by name” needs no attribution lecture and promises no revenue that did not happen.
• Lead with price, not volume. An absolute count invites the obvious objection (“how many would have searched anyway?”). Cost per branded search does not: it is a comparison inside your own budget, between creatives and audiences.
• Connect it to the next funnel step. A branded search is the entry point to your brand search campaign, which already has a measurable conversion rate and conversion value. Show the chain end to end — exposure, search, click, conversion — with the real step-through rates from your own account.

What not to do: assign a dollar value to branded searches and add it to campaign revenue. That turns a sensible indicator into a contested number you will defend in every subsequent review. Value estimates belong in internal budget allocation decisions, not in the P&L line.

Agree on the time horizon separately. Upper-funnel work does not produce results inside a week, and that is a property of the channel rather than an excuse: in most categories the gap between first exposure and purchase runs from days to months. Settle on the horizon you will both judge it over, or the conversation restarts from zero every month.

Implementation checklist

1. Request brand mapping; supply every spelling variant, sub-brand and the app name.
2. Wait the full 7 days after setup before concluding anything from zeros.
3. Build an upper-funnel column set with All conv., Results and view-through conversions.
4. Enable segmentation by conversion action in your main campaign view.
5. Fix the window (7 days by default) and leave it alone mid-flight, or your series stop being comparable.
6. Add the two derived metrics — cost per branded search and search rate — to the weekly report.
7. Cross-check against brand query volume in Search Console with a one-to-three-week lag.
8. Run Brand Lift alongside it on major flights; they complement each other.
9. Validate with a geo experiment once a quarter — that is what proves the effect is real.
10. Never make it an optimization target, even if that ever becomes technically possible.

If upper-funnel work is a permanent part of your strategy rather than a one-off test, the infrastructure matters too: accounts separated by job, consistent reporting, clean access management. That is what agency Google Ads accounts and the rest of the PPC Rebels service stack are built around.

Branded searches is the best fast upper-funnel metric available: free, always on, and based on what people actually did rather than what they told a survey. But it is an indicator, not a goal. The moment you start optimizing campaigns toward it, it stops measuring the thing you wanted it for.

FAQ

What exactly does the metric count?

Searches for your brand on Google and YouTube made within 7 days of viewing your eligible ads. The 7-day window is the default and can be set anywhere from 1 to 30 days.

Which campaign types support it?

YouTube and Demand Gen only. It does not exist for Search, Performance Max or Shopping.

Do I need to run a Search Lift study to get it?

No. The metric is always on. Search Lift answers a different question — whether there is a statistically significant lift against a control group.

Why is my report showing zeros?

Three usual causes: brand mapping is not configured; fewer than 7 days have passed since it was; or you are reading the wrong columns, since the metric lands in All conv. and Results rather than Conversions. Also confirm the campaign type is supported.

How is brand mapping set up?

Through your Google representative — there is no self-serve switch. Provide every spelling variant, sub-brands, and the app name if it differs from the company name.

Can I bid on branded searches?

No. The action is flagged Primary but remains reporting-only, and the Consideration goal cannot be added to a campaign for optimization. That limit is useful: otherwise the system would buy searches from people who were going to search anyway.

What window should I use?

Start with the default 7 days. Extending to 14–30 days is justified only in long-consideration categories, and remember that a wider window absorbs more coincidence.

How does it relate to Brand Lift?

Brand Lift measures perception shift against a control group and requires an experiment. Branded searches record actual search behaviour and are always available. On large campaigns, run both.

Why does only one campaign get credit?

The model is fixed: last-click view-through, full credit to the last ad seen. Inside a multi-campaign flight that biases credit toward whichever campaign served most often and most recently.

What if branded searches are very low?

Check the obvious first: is the brand named in the first five seconds and reinforced at the end, is the audience too narrow, is frequency sufficient. Google’s own documentation points to weak creative, overly narrow audiences, low category search volume and unsupported campaign types as the likely causes.

Can I compare branded searches across brands or clients?

Across campaigns for one brand, yes. Across brands, no — volume depends on fame and category size, so absolute values are not comparable. Compare derived metrics, inside one brand.

How do I convert it to money?

Through the funnel: branded search → brand query click → conversion. Take the historical conversion rate of brand search traffic and your average conversion value to estimate what one branded search is worth. It is an estimate, and a geo experiment is what tells you whether it holds.