Marketing an AI tool in a pre-ChatGPT era
Challenge
French higher-education institutions are legally required to prove graduate outcomes to maintain their accreditations: employment rate, industry, salary bands, etc. For most institutions, these surveys are costly, slow, and unreliable. Alumni are hard to reach, data decays quickly, and response rates depend heavily on follow-up effort.
To respond to this need, Millionroads launched Enquêtes, a new service combining pre-filled regulatory surveys with a fully white-labelled campaign execution. At the same time, the company was migrating its data infrastructure to ESCO, the European taxonomy for jobs and skills.
As a Content Manager with a strong product marketing remit, my role was to help take this to market: clarify the problem, package the offer, and build trust in a context where AI was not yet a familiar reference point.
What I did
I worked at the intersection of product, data, and go-to-market.
On the product side
I partnered with data and engineering teams to understand how AI was used between taxonomies to:
Clean and normalise historical alumni data
Map inconsistent job titles and skills to ESCO
Ensure that only reliable, explainable data was injected into pre-filled surveys
This AI system became the invisible layer that made our upcoming pre-filled questionnaires viable, reducing cognitive load for alumni to simple confirmation or correction.
On the go-to-market side
I worked with the Customer Success team to help shape what Enquêtes should look like in the market:
Pre-filled, conditional Typeform surveys connected to the institution's CRM
White-labelled reporting to preserve institutional ownership
A full communication playbook: email sequences, SMS copy, social posts, call-centre scripts, newsletters, and website banners
Before launch, we ran a pilot campaign with a large private education group to generate social proof, validate response rates, and model cost per completed record.
To support the launch, I co-wrote the product page for the website as well as a three-part editorial series designed to shift how institutions understood their problem.
Outcomes
Enquêtes launched as a repeatable, market-ready service.
For the business, the pilot delivered:
A 77% response rate across nearly 10,000 alumni
Pre-filled surveys that significantly reduced completion friction
A validated cost model per completed record, used in commercial discussions
Social proof with the institution's testimonials, as well as some from their alumni
The editorial series became the most-read content on the company blog for three consecutive months and was reused across newsletters, sales decks and client onboarding.
For me:
End-to-end experience launching an AI-enabled product before AI had cultural shorthand
Deep exposure to data pipelines, taxonomy-driven systems, and ETL logic
A durable PMM framework: design for adoption first, validate with pilots, refine the positioning with customer language