Audience Research That Actually Changes Your Ad Copy

In short

A persona like “women 25–45, interested in fashion” doesn’t change a single word in an ad. Google already factors contextual signals about each person into the auction.

Advertising needs different data: the customer’s job to be done, the event that triggers the search, their fears, objections and decision criteria, and the words they use to describe their problem.

The best sources are the sales team, call recordings, reviews (yours and your competitors’), messenger chats, and the search terms report.

Every insight has to land in a specific place: a headline, a description, an asset, the ad group structure, an audience signal, or the landing page above the fold.

We validate personas with data, not gut feeling: the assets report, experiments, and lead quality.

Almost every brief we receive has a “Target audience” section. And it almost always says something like “men and women 25–55, middle to upper-middle income, interested in home renovation.” That’s not wrong, but it isn’t a working tool either. With a description like that, you can’t write a single headline that stands apart from your competitors’ headlines.

Why demographics don’t write ads

In search advertising, people tell you what they want themselves: they type in a query. Age and gender influence the decision far less than the situation the person is in. Two people of different ages searching for “apartment window replacement cost” have the same job to be done and similar fears.

On top of that, Smart Bidding already accounts for countless signals about the person in every auction: device, location, time of day, language, and past interactions with your site. When you describe your audience only in demographic terms, you give the algorithm nothing new, and you give the copywriter nothing to work with.

What a working persona should include

Element	Question we answer	Where it goes in the ads
Job to be done	What is the person trying to do or change?	Core message, page headline
Trigger	What event made them start searching right now?	Seasonal headlines, ad schedule, urgency
Alternatives	What else are they considering, including doing nothing?	Comparison arguments, positioning
Decision criteria	What do they compare options on?	Headlines with specific terms, callouts
Fears and objections	What worries them, and what stops them from submitting a request?	Descriptions, callouts, FAQ on the page
Customer language	What words do they use to describe the problem?	Keywords, headline wording
Decision-makers	Is it one person or several? Who pays, and who uses it?	Separate messages, landing page structure, B2B campaigns
Timing and channel	When and on what device do they search, and how do they want to get in touch?	Call assets and messengers, device bid adjustments

Where to get the data

The most valuable customer information usually already exists inside the company. Nobody has collected it yet.

The sales team and front desk staff. We ask: What are the five questions customers ask most often? What do they say right before they decide not to buy? What closes the deal? This is the fastest source of objections.
Call recordings. If you have call tracking or VoIP, listening to a few conversations tells you more than any survey. We write down customers’ exact phrases.
Reviews. Both yours and your competitors’: on Google Business Profiles, marketplaces, and industry directories. Competitors’ negative reviews are a ready-made list of your audience’s fears.
Messenger and chat conversations. These show what people ask before they work up to a phone call.
The search terms report. Real phrasing, modifiers, slang, typos, and query language. For the Ukrainian market, it’s important to look at how Ukrainian- and Russian-language queries mix and make a deliberate decision about how to handle them.
Customer interviews. Even a few conversations with recent buyers help you understand the path from trigger to decision. The best question to ask: “What was going on when you decided to take care of this?”
On-site behavior. Microsoft Clarity and GA4 show where people pause, which sections they open before submitting a request, and where they leave the page.

How an insight becomes ad copy

Let’s take a hypothetical example: a company that installs uPVC windows. After talking to the sales reps, listening to calls, and analyzing reviews, we see the following picture.

Triggers: the arrival of cold weather and drafts, renovating an apartment in a new building, and replacing old wooden windows before selling an apartment.

Fears:

“They’ll quote one price at the measurement, and the bill will be a completely different amount.”
“After installation, I’ll have to clean up the dust and redo the window reveals.”
“They’ll install whenever it suits the crew, not when we agreed.”

Decision criteria: a transparent final price, a clean installation, meeting deadlines, and a warranty on the installation itself, not just on the window frames.

Now each insight gets its own place in a responsive search ad:

Insight	Headline (up to 30 characters)	Where else we use it
Fear of extra charges	”No Extra Fees After Measuring”	Callouts, a “What’s included in the price” block on the page
Fear of extra charges	”Contract Price Won’t Change”	Ad description, FAQ
Fear of mess	”Dust-Free, Mess-Free Install”	Photos of the process on the page, callouts
Fear of mess	”We Clean Up After Install”	Scope of work on the page
Deadlines	”Installed on the Contract Date”	Description, contract terms on the page
Seasonal trigger	”Warm Home by This Winter”	Seasonal ads, campaign scheduling

Compare these headlines with the typical “Quality Windows at Low Prices” and “Best Prices in Town.” The first set answers a specific fear. The second answers nothing.

One important caveat: every promise in an ad has to be true for the business. If the company can’t guarantee the installation date, that headline is off the table. Audience research tells you what matters to customers, and the business decides what it can actually promise.

Structure campaigns by motive, not just by product

Audience research changes not just the copy but the account structure, too. If some people search for something “urgent” and others are “planning a renovation,” those are different intents with different readiness to act. They should be split into separate ad groups, and sometimes into separate campaigns with different budgets and schedules.

Typical axes we split ad groups along after research:

urgency: “today,” “24/7,” and “fast” versus planned purchases;
price: queries with “price,” “cost,” or “how much” get ads and a landing page with transparent pricing;
problem versus solution: “leaking” or “drafty window” versus “window installation”;
customer type: homeowners versus businesses or property developers.

Personas for B2B: several people, one decision

In B2B, a persona almost never describes a single person. Several people are involved in the decision to buy a service or equipment, and each of them is looking for something different:

the user: the person who’ll work with the product every day and cares whether their work will get easier;
the department head: accountable for results and focused on implementation timelines, risks, and predictability;
the finance lead or owner: compares cost, payment terms, and payback period;
the technical specialist: checks compatibility, security, and integration requirements.

For advertising, this means several things. First, different roles search with different queries: the user by feature, the manager by business task, the technical specialist by standards and integrations. Second, the landing page needs answers for every role, because the link gets forwarded around the company. Third, the conversion is often not a purchase request but a request for a demo, a quote, or a proposal, and those are exactly the actions we measure and send from the CRM as funnel stages.

A common B2B mistake is writing ads only for the owner, when the person who actually searches and clicks is usually the specialist tasked with finding a solution. Their language is more technical and their questions more specific.

Signals for Performance Max and Demand Gen

For campaigns without keywords, audience research turns into signals and segments:

custom segments based on search terms: we phrase them in the customer language gathered during research;
custom segments based on websites and apps: the resources the audience uses while choosing;
customer lists: to find similar people and to exclude those who’ve already bought;
creatives: each asset group addresses one motive, not all of them at once.

Keep in mind that in Performance Max, audience signals are starting hints, not hard targeting. That’s why creatives play the decisive role: they’re what “selects” the people who respond to a specific motive.

How to check that a persona works

A persona is a hypothesis. Here’s how we test it:

The assets report. We look at which headlines and descriptions the system serves most often and which ones are associated with conversions.
Experiments. For major changes, we test campaign variants in the Experiments section to compare results on split traffic.
Lead quality. We ask the sales team whether customers’ questions have changed since the new copy went live. If people arrive already understanding the price and terms, conversations get shorter.
Search terms. New phrasing in the report is a signal to update the persona and the copy.

Common mistakes

The persona is created once “for the presentation” and never updated.
Only the marketer does the research, without the people who talk to customers every day.
Insights are written in generic terms, like “customers value quality.” That’s not an insight, because everyone says it.
Promises the business doesn’t keep make it into the copy. Leads go up, but so do cancellations and negative reviews.
The research changes the ad copy but not the landing page. Then the promise in the ad isn’t confirmed above the fold.

What to do next

Start simple: ask your sales reps to write down the five most common questions and the five most common reasons customers said no over the past month, then compare them with your ads. If your copy doesn’t answer a single one of those questions, you’ve found the fastest way to improve your ads without changing the budget.

If you want to run audience research systematically, the AdsConcept team builds customer personas from conversations with your team, calls, reviews, and search data, then turns them into campaign structure, ad copy, and landing page requirements.