Why good copy appears to underperform and how to diagnose it

Executive overview

A client opens Google Analytics and asks why your copy is tanking their conversion rate. That moment is high-pressure, but the copy is rarely the only explanation.

Four common culprits account for most apparent failures — and most have nothing to do with the quality of the writing. The fix is asking the right questions before the project starts, not defending yourself after the fact.

Good copy can look bad when it's measured badly.

Keeping your cool when the client pushes back

• You followed a process. Take comfort in that, not in the outcome.
• The client signed off on the copy too — they believed in it as much as you did.
• Avoid confirmation bias; resist scanning data only for evidence that you were right.
• A loss with a solid hypothesis is still useful signal.
• Work with clients who are oriented toward iterative improvement, not a single conversion-rate win.

The four culprits behind underperforming copy

1. Your copy isn't working. Acknowledge this is possible. You may have optimised for the wrong segment or misread the data going in.

2. Seasonality. Month-over-month comparisons can be misleading. A 30% → 20% drop from June to July may be normal for that business. Always ask for year-over-year comparisons alongside month-over-month — one without the other tells an incomplete story.

3. Attribution and funnel blind spots. Clients often zoom in on the single page you wrote and ignore everything else. Conversions happen across devices and multiple touchpoints. If the attribution model only captures the last click, earlier nurture (email, ads) is invisible. Make sure no one is drawing conclusions about one page when the full funnel is unmeasured.

4. Too little data, too soon. Statistical confidence matters. 300 visitors at a 2% conversion rate is not enough to make a call. If a client is checking results a week after launch, or drawing conclusions from a tiny sample, the data cannot support any conclusion — positive or negative.

Upfront questions to protect yourself and your client

• Ask the client to walk you through Google Analytics at kickoff — for the pages you're optimising, or similar pages if you're writing something new.
• Document everything during that walkthrough: traffic sources, funnel stages, historical numbers, anomalies.
• Ask about potential sources of variation: seasonality, PR spikes, events, audience pay-cycle patterns (e.g. a low-income audience converts more around the 1st and 15th of the month).
• Agree upfront on what measuring success looks like — and what it doesn't. If there won't be enough traffic for statistical significance, agree on what you will and won't conclude from the data.
• Set a timeline for when you'll check in on performance and what metrics you'll review together.

Structuring the data conversation with the client

• Never let a client look only at month-over-month — always ask to see year-over-year as well.
• If traffic was unusually low in either period, acknowledge the sample size problem directly.
• If the conversion path spans multiple pages, ask to see the full funnel — not just the page you wrote.
• Segment out anomalous traffic (event spikes, PR bursts) that would distort the baseline.
• Frame losses as answers: a drop tells you the hypothesis was wrong, which is still progress if you're both committed to optimising over time.