Why We Do Research Before Touching a Google Ads Account

In short

Most Google Ads problems are baked in before launch: in the offer, the unit economics, the choice of search queries, and what gets counted as a conversion.

Research answers three questions: where the demand is, whether we can win it profitably, and how we’ll measure it.

The output isn’t a box-ticking slide deck. It’s a set of concrete decisions: campaign structure, negative keywords, bidding targets, ad copy, and website requirements.

How deep the research goes depends on the niche: a narrow local service needs a light version, while e-commerce or B2B needs the full one.

Most accounts that come to us don’t have a bidding problem. They have a strategy problem. The previous contractor kept tweaking bid adjustments, adding negative keywords, and rewriting ads, while the foundation underneath was never validated. As a result, every optimization polishes something that was never going to work in the first place.

That’s why we don’t open the Google Ads interface until we understand the business. Below is how our research phase works, what exactly we check, and why it isn’t a delay but a cost saving.

The symptoms clients come to us with

Client requests usually sound something like this:

“We get lots of clicks but almost no calls.”
“We get leads, but our sales team says they’re not our customers.”
“Performance Max is spending the budget, and we have no idea what it’s doing.”
“Google Ads shows hundreds of conversions, but the CRM shows a handful of deals.”
“We paused our ads, and sales barely changed.”

These are all symptoms. Behind them there’s almost always one of five root causes: a wrongly defined conversion, unprofitable economics, a weak offer, a mismatch between the ad and the page, or ads running on queries with no commercial intent. None of them gets fixed by changing a bid.

Why “tweaking bids” doesn’t help

Smart Bidding optimizes for exactly what you’ve told it success looks like. If the primary conversion is any form submission, spam and duplicates included, the algorithm will learn to bring in cheap spam. If every lead carries the same value, it can’t tell a customer for a small job from a customer for a large project.

In other words, modern Google Ads is an amplifier. It amplifies the right strategy, and it amplifies a wrong one just as diligently. The more automation an account uses (broad match, AI Max, Performance Max), the more expensive a mistake at the input level becomes.

What we research: seven modules

We split the research into seven blocks. Each one answers a specific question and changes specific settings.

Module	Key question	What it changes in the account
Business and economics	How much can we pay per lead or sale and still stay profitable?	Target CPA or target ROAS, bid strategy choice, budget
Market and demand	Who searches for what and how, and when are the peaks and dips?	Keyword clusters, campaign structure, seasonal budget allocation
Competitors	Who do we sit next to in the search results, and what do they promise?	Positioning in ad copy, decisions on bidding for competitor brand terms
Audience	What is the customer’s job to be done, and what are their triggers, fears and decision criteria?	Headlines, assets, audience signals, segments
Offer and positioning	Why should someone choose you over the next result on the page?	Core message, the terms we feature in ads and above the fold
Website and customer journey	Where do people get lost between the click and the lead?	Requirements for landing pages, forms, speed and the mobile version
Measurement and budget	What counts as success, and how will we know a sale happened?	Conversion plan, conversion values, offline conversions, media plan

Business and economics

We start with numbers only the owner knows: margin by product line, average order value, share of repeat sales, and lead-to-deal conversion rate. From these, we calculate the allowable cost per lead and per sale. Without that calculation, “expensive” and “cheap” don’t exist. There’s only gut feeling.

This is often where it turns out that one line of business drives most of the profit, while another can’t pay back its ad spend at any realistic cost per click. That’s a decision better made before launch than after several months of spend.

Market and demand

We pull queries from Keyword Planner, look at seasonality, and, if the account has run before, analyze the historical search terms report. What matters is not just how many people search, but how they phrase it: by service name, by problem, by price, with a city or neighborhood modifier.

We also estimate how much of the demand has commercial intent and how much is informational. The query “how to choose an air conditioner” and the query “air conditioner installation cost” come from different people at different stages, and they need different campaigns or different channels altogether.

Competitors

We look at who shows up for key queries, what they promise in their ads, and what they offer on their pages. If the account has history, we analyze auction insights. A useful free tool is the Google Ads Transparency Center, where you can see the ads of any verified advertiser.

The goal isn’t to copy them, but to find where we can sound different: a more specific offer, more transparent pricing, a faster on-site visit, a clearer guarantee.

Audience

A customer profile for advertising isn’t “women 25–45.” We need the job the person is trying to get done, the event that triggers the search, their fears and objections, and their decision criteria. Our sources are conversations with the sales team, call recordings, reviews, and messenger chats. We cover this stage in detail in a separate article on audience research.

Offer and positioning

Even the best-built campaign won’t save an offer that’s indistinguishable from the ten ads next to it. At this stage, we work with the client to define exactly what goes into the headlines: terms, guarantees, timelines, how you work, and proven advantages.

Website and customer journey

We walk through the customer journey ourselves: on mobile, on a slow connection, in different browsers. We submit test leads and check where they end up and how long it takes to get a response. We look at page speed and, if Microsoft Clarity is installed on the site, session recordings.

Typical findings: the form doesn’t work in one of the browsers, leads go to an inbox nobody reads, or the cookie banner covers the call button on mobile.

Measurement and budget

We build a measurement plan: which actions are primary conversions and which are secondary, what value they carry, and how to send real sales data from the CRM back into Google Ads. In parallel, we prepare a media plan: how much budget each campaign needs to get enough data to learn.

How research changes the setup

Here are a few typical situations we run into.

A clinic with several service lines. Research shows that some people search by service name and others by symptom. These are different intents with different readiness to book. We split them into separate ad groups or campaigns, write separate copy, and word our promises carefully, keeping Google’s healthcare advertising policies in mind.

A B2B company with a long sales cycle. Leads are few, and each deal takes weeks. Starting with a Target ROAS strategy makes no sense here: there won’t be enough data. Instead, we set up imports of qualified leads and deals from the CRM so the algorithm gradually learns from real sales, not from submitted forms.

An online store with varying margins. The site’s catalog is organized by product type, but margins within a single type vary widely. We structure campaigns around economics rather than the catalog tree: high-margin products in one place, and products sold to drive traffic in another.

A local service business. Research shows that most inquiries are phone calls from mobile during business hours. So the priorities are call and location assets, a Google Business Profile, an ad schedule, and accurate call tracking as conversions.

If the account has already been running

Account history is a valuable source, and we don’t throw it away. During the research phase, we run an audit:

which conversions are set as primary, and whether there’s any double counting;
which search terms actually drove leads and which just burned budget;
what changes were made and how metrics behaved afterward (change history);
how budget was split between brand and non-brand queries;
how impression share and position relative to competitors changed over time.

Often the audit shows that the account doesn’t need to be rebuilt from scratch. Fixing a few fundamentals is enough, and only then do we start optimizing.

How long it takes, and why it isn’t a delay

The timeline depends on the niche, the number of service lines, and how quickly we get access and data. For a small local business, it’s a short piece of work. For e-commerce with a large catalog or B2B across several markets, it takes a few weeks.

Importantly, some of the work runs in parallel. While we analyze the market and competitors, a developer can already be fixing forms and we can be setting up measurement. So research rarely delays the launch as much as it seems at first.

Most importantly, the alternative to research isn’t a “quick start.” It’s months of spend on campaigns built on assumptions, after which you’ll have to do the research anyway.

What you get at the end

The output of the research is a working document you can launch from:

Strategy: goals, priority lines of business, and the allowable cost per lead and per sale.
Account structure: campaigns, ad groups, campaign types, and the logic behind budget allocation.
Keywords: keywords grouped by intent and a starter negative keyword list.
Messaging: the core offer, headlines, assets, and answers to objections.
Measurement plan: primary and secondary conversions, conversion values, and offline conversions.
Website requirements: what to fix on the pages before launch and what to fix in the first month.
Checkpoints: which metrics we review after one week, one month, and one quarter.

When deep research isn’t needed

We don’t do research for research’s sake. In some situations, a light version is enough:

a narrow local service with obvious demand and a small set of queries;
a brand campaign for a company that already has a steady flow of inquiries;
a short hypothesis test with a limited budget and a clearly defined success criterion.

But even in these cases, we never skip two steps: calculating the allowable cost per lead and verifying measurement. Without them, there’s no way to tell whether the ads are working at all.

What to do next

If you’re planning a launch, or you’re already running Google Ads but aren’t confident in the foundation, start by asking yourself three questions: How much can you afford to pay per lead? What exactly counts as a conversion in your account? Can you see which of those conversions turned into sales? If you don’t have a clear answer to even one of them, it’s too early to optimize bids.

If you’d like to go through this phase with specialists, the AdsConcept team runs pre-launch research and audits existing accounts. We’ll show you what’s working in your account, what isn’t, and in what order to fix it.