agentleFS
mardab96mardab96/landing-pages-claude-skills/traffic-temperature-match-review/SKILL.md

traffic-temperature-match-review

Skill2 starsChanged 4 months agoMITView on GitHub

Checks whether one landing page is being asked to serve cold prospecting, warm retargeting and branded search at the same time, and what that costs. Use when a page performs on one source and collapses on another, or when a new channel was pointed at an existing page.

The file

Show its rules and text

We split this skill into the rules it gives an agent

14 rules. About 0 are common, 14 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈14

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If conversions are only available in aggregate, say so.

    line 23
  2. Distinctive:

    If your conversion is not a GA4 key event: use the platform's own numbers per channel (Meta Ads Manager and Google Ads report their own …

    line 30
  3. Distinctive:

    Nothing at all: do not run this skill. Its entire value is the split, and a blended number is what it exists to refuse.

    line 32
  4. Distinctive:

    Read the page's opening as each of those visitors.

    line 38
  5. Distinctive:

    Check the ask against each temperature.

    line 39
  6. Distinctive:

    If splitting the page is the answer, name the cost honestly: another page to maintain, another test to run, and traffic thin enough that …

    line 42
  7. Distinctive:

    Never report a single conversion rate for a page served by multiple temperatures without also reporting it by source.

    line 46
  8. Distinctive:

    If a source has too few sessions to have a stable rate, say so and do not rank it.

    line 49
  9. Distinctive:

    Do not recommend personalisation tooling as the fix.

    line 51
  10. Distinctive:

    If conversions are defined differently per source, stop and fix the definition first.

    line 52
  11. Distinctive:

    Do not average across temperatures without also reporting the split.

    line 80
  12. Distinctive:

    Do not rank a source with too few sessions to be stable.

    line 81
  13. Distinctive:

    Do not recommend personalisation software as the fix.

    line 82
  14. Distinctive:

    Do not change traffic routing, campaigns or the page.

    line 83

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Traffic Temperature Match Review

Use this skill when

  • one page receives several sources and the conversion rate between them differs by more than the page can explain.
  • a new channel was launched and pointed at whichever page already existed.
  • branded search and cold prospecting share a page, and someone is averaging their conversion rate into one number.
  • a page was built for one audience and quietly became the default destination for everything.

paid-traffic-message-match-audit compares one ad to one page. This skill compares the page against the intent temperature of every source hitting it.

Required input

  • traffic by source with sessions and conversions per source, not just the total.
  • what each source knows on arrival: has the visitor heard of the brand, seen the product, been to the site before, searched for a solution or been interrupted mid-scroll.
  • the page as it stands: what it assumes the visitor already knows, what it explains, and what it asks for.
  • the conversion definition, and whether it is the same event for every source.
  • If conversions are only available in aggregate, say so. An averaged rate across temperatures hides both the working source and the broken one.

If the data does not exist

Conversions by source is the one input this skill genuinely cannot do without.

  • In GA4: Reports, then Acquisition, then Traffic acquisition. Set the primary dimension to Session source / medium and add your key event as the metric. That gives sessions and conversions per source in one table, which is exactly the input.
  • If your conversion is not a GA4 key event: use the platform's own numbers per channel (Meta Ads Manager and Google Ads report their own conversions) and say explicitly that the two platforms count differently and may double-count the same person.
  • If conversion happens off-site, in a CRM or a booking tool: export the last 100 records and tally by the source field, if there is one. If there is not, that is the finding, and the fix is a hidden source field on the form.
  • Nothing at all: do not run this skill. Its entire value is the split, and a blended number is what it exists to refuse. Say that plainly and route to a skill that works from the page itself.

Analysis workflow

  1. Tabulate every source with sessions, conversions and rate, and sort by volume. Note where a single source dominates the average.
  2. Assign each source a temperature: cold, warm, branded or existing-relationship, and write what that visitor knows at the moment they land.
  3. Read the page's opening as each of those visitors. Note what a cold visitor needs that is missing and what a branded visitor is forced to read that they already know.
  4. Check the ask against each temperature. The same commitment is reasonable from a branded searcher and unreasonable from a cold interrupted scroller.
  5. Look for the averaging trap: recompute the rate with the dominant source removed, and say which kind of dominant you mean, because the answer differs. Here branded search dominates conversions while cold traffic dominates sessions, and removing one gives 1.70 percent where removing the other gives 5.03 percent. State the basis every time.
  6. Decide per source: the page fits, the page needs a section the source requires, or the source needs a different page.
  7. If splitting the page is the answer, name the cost honestly: another page to maintain, another test to run, and traffic thin enough that neither version may reach significance.

Decision rules

  • Never report a single conversion rate for a page served by multiple temperatures without also reporting it by source. The average is the least useful number available.
  • Recompute the overall rate from the per-source counts before anything else. If it does not match the number the user quoted, that discrepancy is the first finding and everything downstream waits for it.
  • A branded searcher forced through a cold-traffic argument is a real cost, not a harmless one. It shows up as time on page and a lower rate than branded traffic should produce.
  • If a source has too few sessions to have a stable rate, say so and do not rank it. Small-sample rates invite confident wrong decisions.
  • Splitting a page is a maintenance commitment. Recommend it only when the volume per variant can support a decision, and say what volume would be needed.
  • Do not recommend personalisation tooling as the fix. Recommend the page change, and mention tooling only if the team already runs it.
  • If conversions are defined differently per source, stop and fix the definition first. Everything else is measuring two things.

Output format

SourceSessionsConversionsRateTemperatureWhat the visitor knowsPage fit
Named sourceNumberNumberPercent, or "small sample"Cold / warm / branded / existingOne lineFits / needs a section / needs its own page

Then:

  • Rate without the dominant source: the number, the basis used (sessions or conversions), and what it changes
  • What a cold visitor is missing: the list
  • What a branded visitor is forced to read: the list

End with:

  • Decision: keep one page / add a section / build a second page / fix the conversion definition / ask for data
  • Volume needed to support a split: the honest number, or unknown
  • Missing data: what would change the verdict

Practical example

User: "Landing page does 2.8 percent overall. Sources: branded search 1200 sessions 61 conversions, Meta cold prospecting 3400 sessions 42 conversions, retargeting 400 sessions 18 conversions, newsletter 90 sessions 6 conversions. The page opens with a long explanation of what the category is, then asks to book a call."

Assistant should start with the reconciliation, not the diagnosis: the supplied sources total 5090 sessions and 127 conversions, which is 2.50 percent, not the 2.8 percent stated, so either a source is missing from the list or the two figures come from different date ranges. That gets resolved before any per-source finding is trusted, because this is the exact discrepancy the skill exists to catch. Then the honest split: 5.08 percent branded, 1.24 percent cold, 4.50 percent retargeting, and the newsletter's 90 sessions is a small sample that should not be ranked. Rate with the dominant source by conversions (branded search) removed: 1.70 percent, 66 conversions of 3890 sessions. Cold traffic is 66.8 percent of sessions and 33.1 percent of conversions, so it carries the volume and the problem: an interrupted scroller is asked to book a call after a category explanation, with no lower-commitment step, while branded searchers read a category explanation they do not need. Then: add a lower-commitment path for cold traffic and move the specific offer above the category explanation, with a second page named as the alternative and the volume it would need to be decidable.

Guardrails

  • Do not average across temperatures without also reporting the split.
  • Do not rank a source with too few sessions to be stable.
  • Do not recommend personalisation software as the fix.
  • Do not change traffic routing, campaigns or the page.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.