If you're a cake business, don't outsource cake making, but do get someone else to build your shop.
But so many web companies spend time and energy building CMS, db astraction layers, and tonnes and tonnes of infrastructure that is nothing to do with their business.
The most common problem I see is technical outsourcing, there are a lot of MBA based teams who cannot convince a great technical person to join.
As a result, outsourcing to get the product done becomes a necessity and also easier than going around to find a good technical cofounder. It is also tempting since outsourcing seems cheaper and promises to get the product built.
The product just ends up not being good enough.
As a technical person in the startup industry, this is the #1 reason why I think startups fails.
A few years ago I did some pro bono work(not intentionally :-) for a couple of startups with domain experience but no software development experience. They had offshored their work to groups in India. The delivered products were marginally functional and not satisfactory. The had compounded the domain to developer communications problems with remote and cultural issues.
I tend to agree with this advice, but OXO (maker of well-designed houseware products) is a famous counter-example. Not only do they outsource manufacturing (as nearly every product company does these days), but they also outsource product design, which is absolutely their core competency.
I've never understood how they made that work, but obviously it is a strategy that has worked well for them.
An example: Borland marketed Pascal and later C compilers. (Ok, sorry--from the deep dark past.) They outsourced both pieces of technology. So you might think that this is an example of outsourcing core competency.
But their core competency was Marketing--Phillipe Kahn, during that period had glow-in-the-dark PR.
(Incidentally, the technological genius behind both products was none other than Anders Hejlsberg.)
But OXO products are not complex product. The majority of their products are injection molded plastic.
So, maybe their core is not product design at all - maybe their core is product concept and product selection. They choose a product type they want, develop the initial concept far enough along to pass it off to the industrial designer who models it - then its contracted out to low-bid/low-cost (presumabley previously used) plastic fab in china.
Seems like a really good model when you have very simple, typically single-digit-materials, products.
We ran into this dilemma last year at Lucidchart when we were debating how to handle SEO, PR, and general marketing. We flirted with outsourcing but eventually hired two full-timers and we haven't looked back. Took a month or two to get them up to speed on the product and best practices but I now glean off of their expertise and having them in-house makes our team stronger.
The only thing we would outsource is very quantifiable deliverables that need no additional maintenance to be valuable. Ie. data entry, template designs, link building.
I stopped reading at "outsource ... analytics". In this age of data and data science "analytics" means a lot more than web analytics and some javascript on your web pages. This expertise needs to be part of your organization's DNA.
I would argue it needs to - marketing starts with thinking about how to communicate your product value to your customers and ends with strategies to keep them happy and around for as long as possible. A great product is the basis but if you don't spend time thinking about how to market your product, your company will most likely not scale
If you're a cake business, don't outsource cake making, but do get someone else to build your shop.
But so many web companies spend time and energy building CMS, db astraction layers, and tonnes and tonnes of infrastructure that is nothing to do with their business.