Something changed in the last two years that most positioning advice has not caught up with.
A feature used to be evidence. Shipping something hard signalled that you had the team, the data, and the time to build it, and buyers read that signal correctly. The feature list was a proxy for capability, which is why feature-led marketing worked for two decades.
Then the cost of shipping an impressive-looking capability collapsed. Summarisation, natural language search, an agent that drafts the thing your user was going to write manually. Six competitors shipped the same four features in the same two quarters, and every homepage in the category started to read like a copy of the others.
Buyers noticed before marketers did. They stopped treating feature claims as information.
What buyers do when everything sounds the same
They fall back on proxies that are harder to fake.
Who else uses this. Who is the vendor obviously built for. Does this company understand my specific situation better than the others, or are they describing a generic version of my job. How confident are they about what they will not do.
That last one carries a surprising amount of weight. In a market where everyone claims everything, a vendor who openly rules something out is providing the only scarce commodity in the room, which is a real opinion. Everything else in the evaluation is available in unlimited quantity.
This is also why the demos that land now open with a point of view rather than a tour. The tour assumes the features are the argument. When six vendors have the same features, the tour is just a slower way of proving you are interchangeable.
Features are not positions
The test is simple. If a competitor can copy it in a quarter, it is not a position, it is a temporary lead.
Positions are expensive in a way features are not, because a position requires giving something up. You claim one buyer and lose the ability to speak naturally to another. You claim one definition of the problem and the deals that define it differently stop fitting. That cost is exactly what makes it defensible. A competitor can add your feature next month, and copying your position means abandoning their own market.
April Dunford's sequence in Obviously Awesome is still the clearest way to do this work: what would customers use if you did not exist, what can you do that those alternatives cannot, what value does that create, who cares about that value the most, and what market frame makes that value obvious. AI changed the inputs to that sequence, not the sequence. The difference is that step two got much harder to answer honestly, because the number of things you can do that alternatives cannot has shrunk.
Answer it honestly anyway. A weak but true differentiator you can build a frame around beats a strong-sounding claim that six competitors also make.
Reframe, do not rename
The reflex, once a category feels crowded, is to invent a new one.
Most companies should not. Naming a category means funding the education of an entire market, because nobody searches for a term they have never heard, nobody has a budget line for it, and no analyst has a quadrant for it. That is a well-capitalised, multi-year project, and companies at Series A and B who attempt it usually end up with a term only their own team uses.
Reframing is the cheaper move and it works faster. Keep the category the buyer already searches for, and change the criteria inside it. Not a new species, a better argument about what good looks like in the existing one.
In practice that means saying which kind of team the standard approach fails, and why, and naming the moment it fails. It means a comparison that argues about criteria rather than counting rows. It means the homepage headline being unusable by the competitor whose product is genuinely different from yours, which is a test worth running on your current headline right now.
Specificity is the last advantage
When production is cheap, judgment about what to produce becomes the scarce thing. That applies to software and it applies to the story around it.
Anyone can generate a page of competent positioning copy in ninety seconds. What cannot be generated is the decision about which buyer to serve, which market to walk away from, and which unfashionable opinion to stake the company on. Those require knowing something specific about a specific set of people, and knowing it well enough to bet on it.
The companies pulling ahead right now are radically specific about who they are for, and they make that specificity the pitch rather than a qualifier buried on the pricing page. It reads as narrow from the outside. From inside the buyer's situation it reads as the first vendor who was clearly built for them.
Where to start
Do not start with the category name. Start with the losses.
Go through the last ten deals you lost and find what the buyer was comparing you against and what they believed the decision came down to. You will find one or two criteria that keep appearing, set by somebody else. That set of criteria is your current category frame, whether you chose it or not.
Then decide whether you can win on those criteria. If you can, say so much more loudly and specifically than you currently do. If you cannot, the work is to change which criteria the buyer arrives with, and that work starts with naming a buyer narrow enough that your real advantage becomes the thing the decision turns on.
Either way, the differentiator is no longer in the feature list. It has moved into the frame around it, and the frame is the part your competitors cannot ship next quarter.