Differentiation

The comparison table has been around forever. Your features on the left, the competitor's on the right, count the check marks. Work from that table long enough and the conclusion always points the same way: fill in the blanks.

So roadmaps used to fill up with someone else's feature list. Sales relays that the other product has something you don't, that item lands in next quarter's plan, and the other side is doing exactly the same thing in reverse. A year later the two feature lists are nearly identical and price is the only lever anyone has left. Both sides cut, and neither has much left over. The table isn't wrong, it just answers the wrong question: a blank anyone can fill is rarely worth the quarter spent filling it.

Differentiation is a way of reading the same table against a different measure: not how many items you have, but how long it would take the other side to catch up on one. A feature someone can match in a week by adding a button isn't differentiation even if you have ten of them. Recommendation quality that only emerges after two years of accumulated usage is differentiation even if it's the only thing on the list.

The longer it holds, the more it costs whoever wants to copy it. That cost usually comes from one of three places.

1. Time. Anything that has to accumulate is hard to buy your way past. Three years of documents linked to each other, recommendation accuracy that comes out of hundreds of thousands of logged actions. A competitor can build the same screen tomorrow and have nothing to put in it.
2. Structure. Copying it would cut into the copier's own revenue. If a competitor charges per user, opening read-only access for free is a move they can't match. Not because they lack the engineering, but because matching it costs them money, and that kind of barrier lasts longer than a technical one.
3. Relationships. When a product gets more valuable as more people use it — a network effect — switching stops being one person's decision the moment one team starts exchanging documents with the next. What the competitor now has to clear isn't a feature, it's the cost of moving an entire customer organization.

Differentiation wears down if you leave it alone. What only you did two years ago is now done by three companies, one of which gives it away. So the same question comes back every quarter: how much would it cost someone to catch up on this item today? When the answer shrinks to "two months," that item has stopped being differentiation. This is why it travels next to the idea of a moat — unless today's advantage gets turned into a structure that thickens over time, it simply gets spent.

Checking it is straightforward. Ask recently signed customers why they chose you. If the answers match the item written in the strategy doc, that item is doing real work. If half of them say "it was cheaper" or "you called first," there isn't one yet.

It isn't a line that sits in a strategy document. It settles three kinds of decisions every week.

1. What to build yourself. Build-buy-partner is the call between building a capability, buying it, or wiring in someone else's, and the rule of thumb is single: build what you listed as differentiation, buy or wire in everything else. That's the argument for not spending six months on a payment integration.
2. What not to build. A feature request is a feature a customer asked for directly, and at 200 of them a quarter you can neither do them all nor cut them at random. Drawing the line at "requests unrelated to our differentiation wait" gives a rejection an actual reason, and the same line moves a request that does line up ahead of older tickets.
3. How to sell. A value proposition — the one sentence for why a customer should choose this product — is built out of the differentiation. Positioning, which decides what place the product takes in a customer's head, splits along the same line. In the field, a competitor objection from a buyer weighing two products gets answered with this one item and not a feature list.

"Is this really our differentiation? They could bolt that on in two months."

Pushing the question from whether a feature exists to how long it takes to copy.

"That one isn't differentiation, so let's just buy it."

A judgment that it isn't worth spending headcount to build in-house.

"We asked ten new customers why they picked us, and half gave the same answer."

Confirming in customers' own words that the written differentiation actually works.

Pagedeck rewrote a roadmap on this question last year. Twelve features the competitor had were sitting in the backlog, and Grace asked how many of them needed to come from Pagedeck specifically. Two survived. The other ten existed in every product in the category and could be matched in a month.

So the ten went to the back of the queue, and the headcount went into tying meeting schedules to meeting notes so past discussions surface when they're relevant. That only becomes useful on an account with a pile of documents behind it, so a newer product can copy the screen and still not produce the same experience. Sofia has taken that single item into competitive deals ever since.

The most common mistake is writing a feature name in the differentiation slot. Feature names are the easiest thing to put in the blank. But a feature is a screen and some code, and copying starts the moment it's visible. The differentiation isn't the feature — it's the condition that stops others from building it, the accumulated data or the business structure underneath. Write the feature name instead and the strategy doc goes blank the day a competitor ships a similar screen, and the team is back to filling in a comparison table.

Treating it as settled for a while is just as common. Strategy docs get opened once a year, so the assumption is natural. In practice the time to catch up shrinks every quarter. What took two years last year takes six months this year, and three cheaper competitors show up in the meantime. Skip the quarterly re-check and roadmaps and sales decks keep getting built on an item that already wore through.