The paper that inspired this illustration can be found here.

For a robust debate about this topic, check out this LinkedIn thread. Here's my take on when/how I think STP should be used, following all the discussion in this thread:
 
1. As a framework for understanding how brands compete? No. It doesn’t fit the scientific data that I have seen. The idea that a BRAND must strive to be differentiated and appeal more to some and less to other types of buyer isn't true.
 
2. As a tool to decide which types of people your BRAND should and shouldn’t try to sell to? Well, research is useful for establishing consumer needs, priorities and attitudes related to a job to be done and therefore what/who your brand competes with. Research also stimulates ideas of how best to address the needs. 'Segmentation’ projects often do this. But segmentation projects are also often conducted amongst category buyers in an attempt to identify a 'type’ of consumer the brand should focus on and try to build a loyal following amongst. This is where STP often falls down IMO, especially if the chosen group(s) represent a small proportion of all category buyers.
 
3. To determine your brand’s price positioning? Research can help with this and it’s an important decision for a brand to make. This is a basic form of segmentation... but you don’t need a huge segmentation project to obtain the required insight/data. The reason I think that a brand needs to decide where to play in terms of price positioning is my suspicion that brands CANNOT successfully come to mind readily for Category Entry Points associated with very different price points. I’ll need to explain more about my thinking about this in another, dedicated post.
 
4. As a method to help you come us with an inspiring/unifying brand identity (/positioning)? You don’t need to do a big segmentation project to do this, but it’s a common side-benefit of doing one. It can be helpful for a brand to define a clear identity (/positioning) because doing so may inspire stronger creative ideas for marketing activities, making them more memorable, effective and consistent with each other (thanks to recurring themes/assets). Consistency makes the marketing increasingly effective over time. The trouble comes if the brand’s clear identity is one that is polarising – engaging and memorable for some category buyers but irrelevant and forgettable for others. The best brand identities (/positionings) lead to marketing activities that are engaging and memorable for most category buyers rather than a (small) subset. Brand identities (/positionings) that can help a brand be memorable across many different Category Entry Points are perfect, IMO.
 
5. As a tool to decide which products in your portfolio to target at different people? STP can be helpful here, IMO. But, to be honest, I think it makes more sense to segment a market based on the Category Entry Points and make sure your portfolio of products is used well to address as many CEPs as possible. Segmenting based on CEPs is better than segmenting based on individuals IMO. After all, the same individual often buys the category for different reasons on different occasions. You can learn more about this here and here. Demand space studies, whereby individuals can belong to multiple groups, depending on their combination of needs, also make sense to me.
 
6. To help plan your media for a campaign in order to reach lots of different people in an efficient way. This is a great use of segmentation techniques IMO. It allows you to reach lots of different category user segments (different ages, for example) using the media best suited to reach each group. You can also tailor the creative content, product(s) featured and message to maximise relevance to each segment. Byron Sharp refers to this as 'sophisticated mass marketing'. It is a far cry from the theories that gave birth the the original STP approach that says a brand has to strive to be differentiated to appeal more to some and less to other buyers/situations.

One last thing to mention. Even if tight targeting might seem illogical for a brand with big growth ambitions.... NEW brands usually DO need to claim that they are aiming to appeal to a niche audience thanks to a USP because this is the only way to convince retailers to partner them. This means, I think, that brands often need to launch with a differentiated message but then broaden their relevance once they have become established. The trick is to choose a brand identity (/positioning) and DBAs at the beginning that will work during the launch phase AND later as the brand expands.