Two: (a) Top-line growth and (b) Cancellations.
I like businesses where "growth" means "revenue," otherwise to me it's only an indication that people are mildly interested rather than proof that it's a *business* that is turning into a validated business model. But of course with consumer often you have to be content with "active users" or somesuch.
True revenue growth I measure (early on) in $/mo of new recurring revenue, not in percentages (because those will be all over the place and large, just because the denominator is small). It's nice to see that number steadily growing, usually linearly (no, not exponential!). It's OK for it to bounce around early on, e.g. you get a big pop from good PR or a dip because it's December.
Soon I like to see you follow that with increasing ARPU, because almost always your prices are too low and you need higher ARPU to drive a real, profitable business, and to allow for higher CAC which means the ability to get varied sources of new signups. But that's not for very early on -- you can do that next. Just having people sign up at all for any amount of non-zero dollars is a wonderful sign.
The other is cancellation. I've seen companies where their cancellation rate is 25%/mo. That means people turn over completely in 4 months, and that means it's NOT a SaaS business!
More importantly, it means that although you've gotten them in the door, and even paying, the fact is you're not delivering perceived value, and that means you don't actually have a product people want, nor a business.
Note I said "perceived." Sometimes you ARE delivering value but they don't understand that, so education or a better UX or follow-up is actually what's needed. Usually you are in fact not delivering much value, and that's what needs to be addressed.
If cancellations are above 5%/mo, you don't have this "fit," and there's no sense in spending time/money growing fast when the bucket is so leaky. You're just force-feeding something rotten.
Actually a SaaS business needs to be more like 2%/mo in the long run, else it cannot grow large enough and without tremendous marketing/sales expense. But to me, <5%/mo *early* on is good enough that you can address that over time.
This depends entirely on the stage you're at with the business. Although some people have mentioned churn, no one has mentioned any measurements around engagement -- i.e. "Is anyone using this product at all, and in ways that mean they'll keep using it?"
In the early stages that's what I would focus on -- a measure of engagement. If you prove engagement (or what I describe as "Stickiness") you can move to the next stage, and now you're interested in seeing if engaged users stick around long enough. That's where LTV comes in. But there's no point measuring LTV or CAC at the beginning--you just started, and you don't know if your product is doing what it needs to in the first place.
If you get solid engagement with a small group, you might then say, "Now I want to bring more people in the top of the funnel and see how they respond" and care about sign-up conversions, while also looking at free to paid conversion from the early adopters (if you're going freemium).
If conversion looks good, and you can map out a scalable business based on that simple number, then you look at churn and LTV. "People are engaged and using the product, converting to paid and then dropping off..."
P.S. This answer is heavily influenced by Ben Yoskovitz's writing :)
1) Site visitors (they can't try if they don't come)
2) Trials as a percentage of visitors (they can't buy if they don't try)
3) Purchases as a percentage of trials (if they don't buy, nothing matters)
4) Other engagement metrics like growth in positive social proof, repeat visits, newsletter signups etc. are fine but the first three are what really matter in early days.
Retention (aka churn). This is the most important because it drives all other decisions around sales, marketing and product. If churn is high it's because: a) sales people are not setting expectations properly (e.g. selling too much sizzle and not enough steak); b) your marketing is bringing in the wrong type of customer; c) your product doesn't solve a big enough problem/pain point for your customer.
It's also the key to forecasting; if you can accurately predict churn, you know how much you need to sell, how many leads you need to generate, etc. etc.
Don't wait, track this asap. And don't fake it (like Salesforce.com used to say, oh, we have less than 1% churn per year, when they were actually tracking the % of customers that don't renew contracts). That's not churn; people cancel or stop paying ALL THE TIME. As soon as you know you're not going to get money again from that customer, they've churned, track that end date right then and there.
Engagement and Referrals.
Engagement answers, "Do people like this and how much?".
Referrals answer, "Is this good enough to refer to other people like me?"
At an early stage company, you're still not quite sure of the best way to monetize but you cannot monetize without seeing engagement.
For an early B2B SaaS company, getting customers is more difficult than building a product. If your early users or customers are not referring the product, the limited people, time and focus you have still needs to be put into improving the product.