Happy Summer!
A productive month, even though our CTO spent most of it hiking the Scottish hills. Nine releases went out, and a handful of features are worth flagging. Here's the short version, with links if you want to go deeper.
Conjoint and MaxDiff analytics without the wait
Hierarchical Bayes now runs natively in the platform. The moment a MaxDiff or choice-based conjoint study closes, utility scores and simulated share-of-preference are estimated and dropped into your cross-tab reports with the same weighting, same filters, same cuts as the rest of your data. No more "export to Sawtooth, wait three days, reconcile back." This step is gone.
Simulated share of preference for QSR brands, with confidence whiskers and significance highlighting, in a standard cross-tab.
Read: Advanced Analytics Without the Wait
Survey people who actually saw your ad
We've added Airtory as a respondent source. Airtory is the rich-media ad-serving and creative platform a lot of brands and agencies already use to run digital campaigns. Wiring it directly into MX8 means you can now field a survey against people who were genuinely exposed to your creative in the wild, with the exposure data already matched. Brand lift, message recall, and creative-effectiveness studies without the panel-stitching exercise.
Learn more: Setting up an Airtory respondent source
Derived questions
You can now define a new question as a combination of existing ones (recoded buckets, composite scores, conditional flags) and use it in any report exactly like a regular question. Configurable base sizes per question, re-orderable rows, and the same question can appear multiple times in one report with different aggregations.
Editing a derived question, combining responses from multiple source questions into a single reportable variable.
Synthetic copy testing for images and video
Synthetic respondents already worked well for text stimuli. They now work for creative too: detailed AI-generated descriptions of images and video clips are fed to synthetic respondents so they can give you a directional read on a piece of creative before you spend on real audiences. Useful for screening concepts down before you field for real.
Creative testing results across concepts, scored by a synthetic audience.
Also new this month
- Multi-stage weighting with a diagnostic report, plus the ability to specify which respondent statuses to include in the weighting.
- Reports while you're still in field: you can now run reports on surveys with no complete respondents yet, useful for live monitoring of incoming data.
- Read-only dashboard views for sharing live progress with stakeholders who shouldn't be editing anything.
From the blog
- From 12 Weeks to 12 Hours: The Dollar Value of Speed in Research — Tom, on what speed is actually worth.
- Cutting Your Agency Spend by 50%: What AI-Assisted Research Actually Replaces — Megan, on which agency services still justify their fees and which don't.
- The $400,000 Researcher: Rethinking What Your Talent Costs (and What It's Worth) — Tom, on flipping the ratio of low-value to high-value work.
Please let me know if you have any questions or suggestions.