When the customer's request conflicts with the data

This week, I received an email from a client. She had three requests. One of them was to remove a specific keyword from all ad copy. The request sounded harmless. The data told a different story.

We’re sharing this case here because it illustrates a pattern we often see with small and medium-sized businesses. Gut instinct and data are drifting apart. Anyone who doesn’t speak up in moments like these is doing the client a disservice—at their own expense.

The Starting Point

We manage an association in the healthcare sector. Google Ads, landing pages, organic visibility. A revamped campaign had been running since February. The results were clear:

• Cost per registration reduced from 110 euros to 38 euros
• 35 percent more sign-ups per month
• With a significantly lower budget

In April, we recorded 20 sign-ups through the campaign. Then the email arrived: “Please remove the keyword [City] from all ad copy.”

Why the request made sense

The client didn’t want to appear regional. The association targets professionals throughout Germany. A city in the ad text signals at first glance: local, small, regional. Exactly the opposite of the desired public image.

From a marketing perspective: understandable. From a data perspective: an expensive misunderstanding.

What the data showed

We always check client requests against current conversion tracking. In this case, the analysis revealed:

In total: 6 out of 20 registrations. That’s 30 percent. Doctors specifically sought out combinations of city and specialty. The city was not a marketing detail. It was a sign of trust. The original versus generic competition.

If the request is tacitly implemented, these six registrations will most likely be lost in May. At 38 euros per registration, no one is talking about large sums. For an association with annual membership fees under three figures, every lead is cash.

How we responded

Three steps. No threats. No “I know better.” Just data.

Step 1: Acknowledge the request.

We wrote back that we take the request seriously. Period. No “but” in the same sentence. Anyone who starts with “yes, but” has lost the conversation.

Step 2: Present the data.

We provided the table above. With context: “That’s 30 percent of all April registrations that came about solely because doctors specifically searched for this location.”

Step 3: A/B test as a safety net.

Instead of a black-and-white decision, we proposed a four-week test:

– Variant A keeps the city keyword (60 percent of the budget)

– Variant B replaces it with “nationally recognized” (40 percent of the budget)

– Evaluation based on CTR, CPC, Quality Score, and Cost per Lead–

Effort: 1.5 hours

This gave the client two options. Cut it out immediately with documented risk. Or test it and decide based on the new numbers.

Why this approach is important

This isn’t diplomacy. It’s taking responsibility for the client’s budget.

Anyone who implements client requests without checking the data is a service provider on call. Anyone who blocks every request is a consultant without a mandate. In between lies the point where collaboration begins. We listen, show the data, and propose a test.

The client opted for the A/B test. If the variant without the city performs comparably after four weeks, we’ll remove the keyword without risk. If the numbers are worse, the anchor stays. In any case, we’ll end up with numbers, not opinions.

The process takes us 1.5 hours to set up and four weeks of patience. In return, we’ll retain 30 percent of April sign-ups if the data confirms the anchor. At a cost of 38 euros per sign-up, the A/B test pays for itself starting with the second additional secured lead.

The Broader Trend

We see this pattern more often. AI tools deliver “marketing recommendations” at the push of a button. ChatGPT writes ad copy. Tools suggest keyword lists. Mid-sized companies get the impression that a data check is unnecessary.

That doesn’t work. An AI has no access to the conversion tracking for the specific campaign. It doesn’t know the last 20 sign-ups. It doesn’t know whether a specific city keyword brings in 30 percent of the leads or none at all.

Data-driven marketing decisions require three things:

1. Accurate tracking. Which sign-up comes from which campaign, which keyword, which ad?
2. Transparent analysis. A spreadsheet that everyone can understand. Not a reporting dashboard that looks like a NASA cockpit.
3. The courage to test an assumption instead of implementing it.

Point three is the hardest. It takes patience. It takes 1.5 hours to set up. In return, it provides a basis for decision-making instead of just an opinion.

What we do with every new campaign

From the very first hour with a new client, we build the tracking foundation. Specifically:

• Clearly define conversion goals in Google Ads (sign-up, inquiry, download)
• Link Search Console to the account
• Regularly analyze Auction Insights (visibility compared to competitors)
• Use the Ad Preview Tool instead of searching manually (otherwise Google personalizes the results)

The last point surprises many clients. If a business owner regularly searches for their own keyword and never clicks on the ad, Google learns: “This person isn’t interested.” The ad appears less frequently to them. But it continues to appear to other users. This explains 90 percent of all “I can’t find my ad anymore” complaints.

That’s why we set up a live dashboard in Google Looker Studio for every client. This gives management and marketing teams up-to-date figures at any time, without having to search Google themselves. Conversion tracking, cost per sign-up, top keywords, and competitor comparisons—all on one page. In the association example, the dashboard will launch in June and replace the old PDF reporting.

Three Lessons for Every Small Business

Three points that aren’t covered in any marketing training but save money every month. We call them internally “the three questions before clicking Save”:

1. Check the tracking before making any gut decisions. A quick look at the conversion data takes five minutes. A wrong decision costs weeks.

2. Test before changing strategy. Even if your gut is 90 percent right: A/B testing is cheaper than a full overhaul. Four weeks with 40 percent of the budget is enough for a reliable conclusion.

3. Ask the service provider for the numbers, not their opinion. Anyone who answers the question “What do the data say?” with “I think…” has the wrong tool at the table.

Conclusion

A city name in an ad text may seem like a minor detail. In this case, it accounted for 30 percent of monthly sign-ups. Without conversion tracking, this figure remains hidden. Without the courage to ask follow-up questions, the report at the end of the month will show weaker results.

For us, service means: listening, reviewing, and presenting options. Not: fulfilling every request. Not: rejecting every request. But rather: working with the client to reach a sound decision.

Frequently Asked Questions

How long does a meaningful A/B test take on Google Ads?

Four weeks is a reliable guideline for moderate budgets. For very small campaigns with few clicks, it takes longer to arrive at statistically reliable conclusions. For very large campaigns, two weeks are often sufficient.

Why don’t I see my own Google ad anymore?

Google heavily personalizes search results. If you repeatedly search for a term and don’t click on your own ad, it signals low interest. The system then displays the ad less frequently. For objective visibility checks, there’s the Google Ad Preview Tool and Auction Insights.

Flat-rate or pay-as-you-go—which is better?

Both models have advantages. A fixed retainer helps at the beginning when there are many tasks to be done. Pay-as-you-go is suitable once the basics are in place and only ongoing optimization is required. Transparency is key: a monthly breakdown of services rendered.

How many registrations per month are realistic for a B2B campaign?

That depends on the market, search volume, and competition. For an association or training campaign with a moderate budget, 15 to 25 registrations per month is a good range. More important than the absolute number is the trend in costs per registration over time.

Do we need our own dashboard for our campaigns?

A live dashboard in Google Looker Studio makes sense as soon as multiple people need data. It requires a one-time setup effort and then provides up-to-date figures 24/7. For very small setups, a monthly PDF report is sufficient.