TopClick SEO Expert

Changed Prompts

Keep comparison groups separate

Dmytro Nasyrov PhD

The comparison question

A report can use the same query ID after its wording changes. That stable label does not make the new observation answer the old question.

Before comparing visibility, record whether prompt text, intended audience, market or language changed. This proposed register separates revisions rather than hiding them inside one trend.

Q-18: a fictional revision

Version 1 asks about a supplier for a ten-person team. Version 2 asks about a supplier for a hundred-person team. This is a fictional change in the decision context.

Retain both exact prompts. Start a separate comparison group for version 2 unless the report explicitly explains a different comparison method. No AI answers were collected for this example.

The prompt version register

FieldPurpose
Query ID / versionKeep family identity and exact revision
Exact promptPreserve wording, not a paraphrase
Audience / taskRecord who the query is for
Market / languageKeep observation scope visible
Comparison groupSpecify which observations may be compared

Two groups under one family

RecordFictional comparison scope
Q-18 v1Ten-person buyer; group G-18A
Q-18 v2Hundred-person buyer; group G-18B
Market / languageSame in both illustrative records
Observed answersNONE COLLECTED
Reporting decisionDo not label the group change as a measured gain

An unchanged prompt is not a guarantee

The same prompt can still produce different answers. Preserving wording supports a comparison protocol; it does not make engines deterministic.

Record the capture date and engine alongside the agreed conditions. Treat changes in the observation setup as separate metadata, even when the prompt stayed the same.

Replacing a query in the portfolio

Keep a replacement log linking the retired prompt to its successor. Record the reason and date without rewriting earlier captures.

A report may explain the coverage of the current query set separately from the trend for unchanged queries. The chosen method must say which group and period each conclusion describes.

Receiving the next report

Ask the receiver to locate the exact prompt behind one reported comparison. They should be able to identify its revision and comparison group without guessing.

TopClick agrees query quantities and scope per platform in the proposal. A changed query set should remain visible in the evidence delivered under that agreed scope.

Sources and limits

Google Search Central explains Google AI feature eligibility and reporting; it does not validate this proposed cross-engine comparison register. The versioning decision is an editorial measurement proposal.

Q-18 and its groups are fictional. No engine collection, client visibility result or statistically controlled experiment was performed. A preserved prompt alone does not establish causal improvement.

About the author

A useful trend states which question stayed comparable and which question was replaced.

Portrait of Dmytro Nasyrov wearing a dark suit and light blue shirt against a dark background.

Written by Dmytro Nasyrov PhD, software architect with 24 years of production experience. Dmytro is the CEO of TopClick.

Changed Prompts: Keep Comparison Groups Separate

By TopClick SEO Expert

Changed Prompts: Keep Comparison Groups Separate

Q-18 changes its intended audience. A version register preserves the old prompt, records the revision and starts a separate comparison group.

  • 68