In my opinion, scaling up should not be Baskettt's immediate concern - they should concentrate on an initial beta group of 50-100 users and make that group very happy, and only try to scale up further once they have a clear understanding of user motives and perhaps some market segments to target. This applies also to algorithm selection - at this stage, they are likely to get good feedback from beta users even if they use a fairly dumb algorithm, and they can refine it to add intelligence much later when a few percentage points improvement is of significant value.
It seems to me that J.D. and Arnaud need to get to the stage where at least one of them could quit and focus full-time on Baskettt. I can see two obvious paths to this goal: 1. Find opportunities to monetise, perhaps through user subscriptions or partnerships with small suppliers - for instance, offer to put Squirrel's Biscuits in the basket for every user whose behaviour suggests they would like biscuits. They wouldn't necessarily need to aim for profitability, just a repeatable sales cycle with enough certainty to attract angel investment. 2. Ignore revenue and focus on growth exclusively, aiming to build a big user base fast and attract investment based on trajectory. Option 2. sees much less likely to work, given the founder's limited time for the startup and the caution of European investors, so I think they'd do best to pursue option 1. Whatever strategy they follow, they should consider applying for a startup accelerator as a means to get a few months off to build the company.
As with most startups, the founders lack some of the skills needed to make the company successful, and they will need to learn these quickly. Of particular note is that they lack sales and marketing experience - selling will be needed if they pursue early partnerships (option 1 above), while they will need very good marketing, particularly over email, to keep users coming back often enough to build a meaningful profile. I suggested they meet with suppliers to understand their problems, and that they look at Graze as their snack service has a similar need to drive repeat usage and their emails are particularly effective at doing so.
Three months after giving the above advice, I got back in touch with Baskettt and they said:
Since we last met, we learned tons of stuff from our users. It allowed us to focus on the most essential features of the product. We have already generate a few hundreds baskets. Pretty cool! On the business part, we found an simple way to monetise our product and bootstrap, we’ll keep you posted on that. Thank you very much for your advice. Talking with our users helped us to see where we should focus our efforts. We’re currently looking at accelerators in London. I think we are at the perfect stage to be part of one right now. We’re trying to pick the one that fits our needs and industry the best.
Sounds like avoiding premature scaling and talking to users paid off for them so far!