Analytics
Measure which atoms and images drive conversions. Watch the system learn from each one and improve automatically.
Conversion tracking
In Settings, add the URL path that identifies your confirmation page. order-received or /thank-you. When a visitor lands on a match, the conversion is recorded against every atom they saw and the image shown. No developer required.
For explicit control: call spectare('conversion') on a button click, form submit, or any custom event. It queues correctly even if called before the script loads.
Option 1: URL match (no code)
Option 2: explicit call
Spectare records:
✓ Which atoms were visible
✓ Which image was shown
✓ Visitor intent and segment
Co-conversion graph
87 co-conversions
74 co-conversions
61 co-conversions
Rebuilt nightly from 90 days of conversion data
Co-conversion graph
Every time a visitor converts, Spectare records the full set of atoms they saw. Over time it builds a weighted graph: which atoms consistently appear alongside each other in converting sessions.
When the next visitor arrives, Spectare picks supporting atoms using that graph rather than just semantic similarity. Atoms that have proven to work together get priority.
The graph is rebuilt nightly from 90 days of data and visualised in the admin. Isolated atoms that have never appeared in a converting session are flagged.
Learning loop
The co-conversion graph learns which atoms work together. The learning loop goes one level up: it records which arrangement of components each visitor saw, with its impressions and conversions.
Once an arrangement has enough data to clear your signal threshold, Spectare feeds it back into assembly. Claude is told which arrangements convert best for each audience and stage, and prefers them whenever your atoms support one.
You set the threshold, so nothing is promoted on a handful of clicks. The confidence report in the admin shows which arrangements have enough evidence to trust and which are still gathering data.
Arrangements: developer, decision stage
Promoted arrangements are preferred in future assemblies for this segment.
Image optimisation
Spectare selects images using a multi-armed bandit algorithm. Each image can be tagged with audience affinities and buying stage affinities, letting the engine match images to visitor segments.
90% of the time it serves the image with the highest conversion rate for that audience and stage combination. 10% of the time it explores a different image to gather data on alternatives. When a visitor converts, the image they saw earns a higher score.
For cases where a specific image must always travel with a specific atom, set the same Group ID on both. Locked pairs bypass the algorithm entirely.
Image selection logic
Each conversion updates the score. No manual analysis needed.
Pro features
Available on Pro and Enterprise.
Natural language summaries of your conversion and assembly performance, surfaced in the analytics dashboard.
Understand what is working and what to do next without configuring a reporting setup.
Spectare identifies intent patterns that qualify frequently but where no well-matched atom exists in your library.
Know exactly what to write next to improve coverage. No guesswork about where the gaps are.
Set a conversion URL in Settings. Everything else is automatic.