How to Run Keyword Discovery Case Studies: A Step-by-Step Guide for PPC Pros

TL;DR: A keyword discovery case study is a structured process where you document how you found, tested, and validated new keyword opportunities inside a PPC campaign, then package the results so they're repeatable and shareable. This guide walks you through all six steps, from setting your hypothesis to packaging your findings into a client-ready deliverable.

If you manage Google Ads accounts, you've probably done keyword research. But have you ever turned that work into a documented case study with a real before-and-after story? Most advertisers haven't, and that's a missed opportunity on two fronts: campaign performance and proof of expertise.

This guide is for marketers managing Google Ads accounts, freelancers pitching new clients, and agency owners who need to demonstrate ROI in concrete terms. Whether you're managing one account or fifty, the framework here gives you a repeatable process for running keyword discovery case studies that produce real data, not just gut-feel observations.

Here's the key distinction worth making upfront: keyword discovery is different from keyword research. Research usually means brainstorming from scratch, pulling ideas from tools, and guessing at intent. Discovery means finding what's already happening inside your account. Your search terms report is a live record of what real users typed before clicking your ad, and it's the single most underused tool in Google Ads. Mining it systematically is the core of what this guide covers.

The difference between casual keyword research and a structured case study approach comes down to documentation. A case study gives you a before snapshot, a clear hypothesis, a controlled test, and an after snapshot with real metrics. That structure is what makes your findings credible to clients, useful for team training, and valuable as a content marketing asset for your agency.

This guide focuses primarily on Google Ads PPC keyword discovery, though the framework transfers well to SEO and other paid channels. By the end, you'll have a six-step process you can run on any campaign, starting this week.

Step 1: Define Your Case Study Goal and Hypothesis

Before you touch a single keyword, you need to know exactly what question your case study is trying to answer. This sounds obvious, but in most accounts I audit, people skip this step entirely. They pull the search terms report, make a bunch of changes, and then wonder why they can't explain what actually moved the needle.

A good case study goal is specific and measurable. Not "improve keyword performance" but something like: "Can we find converting search queries in our search terms report that haven't been added as standalone keywords, and by adding them as exact match, can we lower CPA while maintaining volume?"

That's a question you can actually answer with data.

Once you have your question, frame it as a hypothesis using this structure: If we [action], then [expected outcome] because [reasoning].

A few examples that work well in practice:

Conversion mining hypothesis: "If we identify search terms with two or more conversions that aren't currently added as keywords, then adding them as exact match will lower our average CPA because we can bid more precisely on proven converters."

Negative keyword hypothesis: "If we identify and negate high-spend search terms with zero conversions over the past 60 days, then our overall campaign CPA will improve because we'll stop wasting budget on irrelevant queries."

Long-tail expansion hypothesis: "If we isolate three-word-plus queries that are converting at a lower CPC than our shorter head terms, then building ad groups around them will improve ROAS because long-tail intent is more specific."

Notice that each hypothesis has a clear action, a predicted outcome, and a reason why. That reasoning is what separates a case study from a lucky guess.

One common pitfall here: making the scope too broad. Don't try to run a case study on your entire account. Pick one campaign, one ad group cluster, or one product line. You need to be able to control variables, and that gets impossible when you're touching dozens of campaigns at once. Narrow scope produces cleaner data and a more compelling story. If you need help structuring experiments within Google Ads, check out how to manage keyword experiments in Google Ads for a deeper dive.

You'll know you're ready to move to Step 2 when you can explain your case study goal in a single sentence to a colleague or client without them needing to ask a follow-up question.

Step 2: Set Your Baseline Metrics and Timeframe

A case study without a baseline is just an anecdote. The before snapshot is what gives your after results meaning. This step is where a lot of well-intentioned PPC work falls apart, because people make changes first and try to reconstruct the baseline afterward. Don't do that.

Before you implement any changes, document the current performance of the campaign or ad group you're studying. The core metrics to capture are:

CTR: Gives you a signal on relevance and ad quality before and after your keyword changes.

Average CPC: Watch for shifts here when you add more specific exact match terms, which often compete less aggressively and come in cheaper. You can learn more about evaluating these shifts in our guide on how to benchmark keyword CPC vs industry average.

Conversion rate: The percentage of clicks that convert. If your new keywords are more intent-aligned, this should improve.

Cost per conversion (CPA): The headline metric for most accounts. This is usually the number clients care about most.

Impression share: Useful context for understanding whether your changes expanded or contracted your reach.

Total spend: Document this so you can confirm you're comparing apples to apples in your test period.

Export your search terms report for the baseline period too. You'll want to reference it later when you're showing which specific terms you acted on.

For your timeframe, the standard approach is to pick a lookback window of 30 to 60 days for your baseline, then run your test for an equal period. Matching the windows matters because you want comparable data volumes. If your baseline covers 30 days and your test only runs two weeks, the comparison won't hold up.

Be deliberate about avoiding seasonal anomalies unless seasonality is specifically what you're studying. If you set your baseline during a holiday sale period and run your test in a slow month, any decline in performance will look like your keyword changes caused it, when really it's just the calendar. Note any known external variables in your documentation, including budget changes, new competitors entering the auction, or major product changes on your landing pages. For accounts with heavy seasonal swings, our guide on how to optimize keywords for seasonal campaigns covers this in detail.

What usually happens here is that people underestimate how much time this documentation takes. Build a simple Google Sheet with one tab for baseline metrics and one tab for your change log. It doesn't need to be fancy. It just needs to exist before you start making changes.

Step 3: Mine Your Search Terms Report for Hidden Keyword Opportunities

This is where the actual discovery happens. Pull your search terms report for the baseline period and start sorting systematically. The goal is to categorize every meaningful query into one of three buckets.

Bucket 1: Winners to add. These are search terms that have already converted but aren't currently added as standalone keywords. They're triggering your ads through broad or phrase match, which means you're not bidding on them directly and likely paying more than you need to. Adding them as exact match keywords gives you precise control over bidding and messaging.

Bucket 2: Losers to negate. These are search terms eating budget with zero conversions, often over a meaningful spend threshold. The exact threshold depends on your account, but a common rule of thumb is any term that has spent more than your target CPA with no conversion is a candidate for negation. These are your budget leaks. If you're new to this process, our guide on how to use negative keywords in Google Ads walks through the fundamentals.

Bucket 3: Maybes to test. High-impression, low-data terms that haven't converted yet but show intent signals worth isolating. These might be emerging queries, long-tail variations, or terms with strong semantic relevance that just haven't had enough volume to prove themselves yet.

For the practical workflow, sort by conversions descending first. Your winners will surface immediately at the top. Then sort by cost descending and filter for zero conversions. Your losers become obvious. Finally, look for patterns in the long tail: question-based queries, location modifiers, comparison terms, and branded versus non-branded splits often reveal entire sub-themes worth building out. Our article on how to research long tail keywords for Google Ads covers these patterns in more depth.

In most accounts I audit, the Winners bucket alone contains enough material to justify the entire case study. Advertisers are often running broad or phrase match campaigns that are converting on very specific three-to-five word queries, but because those terms aren't added as keywords, the account is bidding on them inefficiently and missing the opportunity to write dedicated ad copy for them.

The mistake most agencies make at this stage is doing this work in a spreadsheet that lives outside Google Ads. You export the report, sort it in Excel, highlight rows, then have to manually re-enter everything back into the interface. It's slow, error-prone, and makes the whole process feel heavier than it needs to be.

This is exactly where a tool like Keywordme changes the workflow. The Chrome extension lets you work directly inside the search terms report in Google Ads, so you can add keywords, apply match types, and build negative keyword lists with single clicks, without ever leaving the native interface. For the categorization and implementation work in this step, that kind of in-interface speed makes a real difference, especially when you're managing multiple accounts.

You'll know this step is complete when you have a documented list of 10 to 50 keyword opportunities, each clearly categorized into one of your three buckets with a brief rationale for each decision.

Step 4: Implement Changes and Structure Your Test

Now you implement. The key discipline here is making only the changes directly related to your hypothesis, and documenting every single one.

Start with your Winners bucket. Add converting search terms as exact match keywords in the appropriate ad groups. Exact match is the right call here because these terms have already proven themselves. You want precise control, not broader reach. For a refresher on when to use each type, see our guide on how to compare keyword match types for effective PPC campaigns. For your Maybes bucket, phrase match is often the better starting point since you're still in discovery mode and want some flexibility in how those terms trigger.

For your Losers bucket, add negatives at the campaign level if the irrelevant terms are specific to that campaign, or at the account level if they're universally irrelevant across everything you manage. Use keyword clustering to group related negatives efficiently. For example, if you're negating terms related to a competitor you don't want to show for, add the root brand name as a negative rather than listing every variation individually. Our guide on how to organize negative keywords by theme covers this clustering approach in detail.

Your change log is your case study's methodology section. Every entry should include: the date, the keyword or negative added, the match type, the ad group or campaign it was added to, and a one-line rationale. This log is what separates a case study from a campaign update. When you're presenting results later, you can point to exactly what you did and when, which makes your findings credible and your process replicable.

The common pitfall at this stage is making too many changes at once. If you restructure your campaign architecture, update your ad copy, adjust your bidding strategy, and add new keywords all in the same week, you won't be able to attribute results to keyword discovery alone. Your case study loses its explanatory power. Discipline yourself to change only what your hypothesis calls for.

Once everything is implemented, record your test start date. This is your line in the sand. Everything before it is baseline; everything after it is test data. Set a calendar reminder for when your test window closes so you don't forget to pull results.

You're ready to move on when all changes are implemented, logged with dates and rationale, and you have a clear, documented start date for the test period.

Step 5: Monitor, Measure, and Collect Your Results

Let the test run. This step is mostly about patience and light monitoring, with a clear process for when to intervene and when to stay hands-off.

During the test period, do a weekly check-in rather than daily. You're looking for early signals, not final verdicts. Are your newly added keywords getting impressions? Are your negatives reducing spend on irrelevant queries? Is CPA trending in the right direction? These weekly checks are diagnostic, not decision-making moments.

The one exception: if a newly added keyword is spending heavily with zero conversions after accumulating data equivalent to two to three times your typical CPA, pause it. Document this in your change log. Not every discovery is a winner, and that's genuinely valuable data. A case study that shows you added ten keywords, eight improved CPA, and two were paused for poor performance is more credible than one claiming everything worked perfectly. For a systematic approach to pausing underperformers, our guide on how to refresh and prune underperforming keywords is a useful companion resource.

When your test window closes, pull the same metrics you documented in Step 2. Build a simple comparison table in your Google Sheet: baseline period on the left, test period on the right, and percentage change in a third column for each metric. Calculate the changes for CTR, CPC, conversion rate, CPA, and total spend.

Look for the story in the data. Did CPA drop while conversion rate held steady? That suggests your new exact match keywords are bidding more efficiently on the same quality of intent. Did impression share decline slightly while CPA improved? That's often a sign your negatives are working, cutting volume on junk queries while keeping the profitable ones.

The goal of this step is a clear, data-backed comparison that shows what improved, what stayed flat, and what didn't work. That honest picture is your raw material for Step 6.

Step 6: Package Your Case Study for Maximum Impact

Raw data is not a case study. Turning your comparison table into a narrative that's useful to clients, your team, or your portfolio is the final step, and it's where most PPC pros leave value on the table.

Structure your case study using this five-part framework:

Background: What was the account situation before you started? What problem were you trying to solve? Keep this brief, two to three sentences.

Hypothesis: The exact hypothesis you defined in Step 1. This shows your work was structured, not reactive.

Method: What you did and when. Reference your change log. This is your methodology section, and it's what makes the case study repeatable by someone else.

Results: Your comparison table with the key metrics. Include before/after screenshots from Google Ads where possible. Visuals add credibility that numbers alone don't always convey.

Takeaways: What you learned, what you'd do differently, and what you're doing next. This is the most underwritten section in most case studies, but it's the one that demonstrates genuine expertise.

For agency owners, this document becomes a client deliverable that proves your value and justifies your management fee. Clients don't always understand what keyword optimization involves, but they understand a clear before-and-after story with their own account data. This format makes your work visible in a way that a dashboard alone never does.

For freelancers, the same document becomes a portfolio piece. Anonymize the client details if needed, keep the methodology and results, and publish it as a blog post or case study on your site. It demonstrates PPC expertise far more convincingly than a generic "I manage Google Ads campaigns" bio.

For in-house marketers, present it to stakeholders as a repeatable framework: here's how we found improvements this quarter, here's the process, and here's how we'll run it again next quarter. That framing positions you as someone who builds systems, not just someone who runs ads. Once you've validated your findings, you can use them to plan keyword expansion strategies for the next growth phase.

One important note: don't cherry-pick only the positive results. Including what didn't work builds credibility and, more practically, helps you refine your process for the next round. A case study that says "eight of our ten new keywords improved CPA, two were paused for poor performance, and here's what we learned from those two" is more trustworthy and more useful than one that only shows the wins.

Your case study is complete when you have a shareable document or presentation that tells a clear story with real data, structured so that someone unfamiliar with the account could understand what you did and why it worked.

Your Quick-Reference Checklist and Next Steps

Here's the full process condensed into a checklist you can reference before your next campaign audit:

1. Define a specific, measurable hypothesis using the "If we / then / because" format and narrow your scope to one campaign or ad group cluster.

2. Document baseline metrics (CTR, CPC, conversion rate, CPA, impression share, spend) before making any changes. Export your search terms report for the baseline period.

3. Mine your search terms report and categorize opportunities into three buckets: Winners to add, Losers to negate, and Maybes to test.

4. Implement only the changes your hypothesis calls for. Log every change with date, keyword, match type, ad group, and rationale.

5. Run your test for a window matching your baseline period. Do weekly check-ins, intervene only when clearly necessary, and document everything.

6. Pull your results, build your comparison table, and package everything into a Background / Hypothesis / Method / Results / Takeaways narrative.

Running keyword discovery case studies isn't a one-and-done activity. It's a recurring practice that compounds over time. Every case study you complete adds to a library of documented wins and lessons that makes you faster and more confident in the next round. Over a year, that library becomes a genuine competitive advantage, whether you're pitching new clients, training junior team members, or making the case for a bigger budget.

The process also gets significantly faster when you're working with the right tools. Bouncing between Google Ads, spreadsheets, and a separate keyword tool adds friction at every step. The more of this work you can do directly inside the Google Ads interface, the more time you spend on analysis and strategy rather than data wrangling.

Start with one campaign this week. Pick a clear hypothesis, document your baseline, spend 30 minutes in your search terms report, and make your first round of changes. You don't need a perfect system on day one. You just need to start documenting.

If you want to speed up the search terms mining and implementation work, Start your free 7-day trial of Keywordme and see how much faster keyword discovery gets when you can add keywords, apply match types, and build negative lists directly inside Google Ads without touching a spreadsheet. After the trial, it's $12/month per user, which is a straightforward trade-off when the alternative is hours of manual work per account.