One of the great things about services like AdWords, Wufoo, Mailchimp, Flowtown, Performable, and so on is that they allow the marketing side of a startup to work independently of the engineering side of the startup. Whip out your credit card and you can have a source of traffic, lead generation forms, e-mail marketing campaigns, social media insights, A/B testing of your landing page, and all that great stuff.
The problem with this is that as these tools have begun to play better together (thanks to the magic of Webhooks), your data becomes a bit more trapped inside each of these programs. It can be hard to audit your data and even harder to integrate this data into your core application. You end up going to each service independently to answer basic questions such as "when did this user give us their e-mail address?" and "just where did we get this e-mail address from?" And if you want to answer more complex queries such as "just what is our overall (blended) customer acquisition cost (CAC)?", you have to export your data into CSVs and manually massage them together, which can be a very painful process indeed.
The bigger issue is when you try to integrate this information with your main app so you can do cohort analysis. Cohort analysis is the magic analysis that tells you how you're doing on your key retention and engagement metrics over time: "did our June signups come back to the site more often than our April signups did?" Trying to import user acquisition dates into your database of existing users (once you've sent them an invite to your app) is very time-consuming as you want to make sure that you don't break anything - there are a large number of edge cases you have to deal with: you'll have users who gave you their address more than once (for example), you'll have users that have changed their e-mail address inside the app from the one they gave you on your lead gen form, and you'll have users that have invited other users.
When you're a pre-launch or early startup, you don't have a lot of data points to deal with. This is good because it means that you can, with effort, use Excel or some other tool to tie together all these different repositories of data. But it's also bad because you may not think to have systems in place to more elegantly handle measuring your AARRR metrics post-launch. You'll be so busy putting out fires and doing PR that you may end up neglecting instrumentation.
While you may not have to do everything in-house, it's critical to have an engineer on your team who's involved in business/marketing support from time to time. Defining and implementing an integration and migration paths (a lot of these services *do* have APIs for example, but those aren't something the mouthbreathing "business guys" can do anything with) before you launch is critical to make sure that you can calculate and measure the things you'll need to calculate and measure as you scale your business.