How do you project that paid campaigns will stay profitable when you scale them?

You can try setting up a ACE (adwords campaign experiment) where you take your 'tried and true' exact/phrase match terms and then triple the bid.

See what happens to your ROI. https://support.google.com/adwords/editor/answer/1399246?hl=en

Honestly though, if the campaigns were running well and ROI only improved by cutting...you're projection is really an imaginative narrative.

Also - I'd try to find your cost per new customer, provided you're single channel and have a good working customer definition.

best,

Jeff

Couple of things:
Any good investor knows that a projection is just that. The exercise is more to help articulate what the investor needs to know is possible: That their infusion of capital can truly accelerate your growth. So it's your responsibility to help create that confidence.

The biggest concern I have for you is that you speak about not knowing what's working. Understanding the mechanics of each channel of spend, and segmenting the users' behavior (churn, upsell, etc) into those channels will give you a much better read into what's really working and IMO, are required before you can really answer the question credibly.

Also, I assume you're doing a Series A or beyond raise, in which case, a potential lead or significant investor asking for projections are entirely reasonable. But if you're raising initial seed and/or this ask is coming from a non potential lead investor, then I'd raise a flag of caution.

Happy to talk through this in more detail if you'd like. Best of luck with the funding round!

Projections for scaling can be difficult.

For Facebook ads, part of the question will be whether you haven't captured all of your core audience because of budget. If you have narrowed down what's working and have been hamstrung by the spend, then Facebook projections can be pretty reliable as long as you're maintaining the core demographics.

For AdWords and paid search, if you're losing impression share due to budget then projecting based on that will be pretty reasonable. If you have to expand into new keywords to make the budget expansion happen, I would be conservative with the estimates and would not assume a 1 to 1 based on past performance.

Hope this helps!
best,
Paolo