Marketing & GrowthFree
Competitive Positioning Map With Real Axes
Score every candidate attribute from published evidence, then let spread and correlation pick the two axes rather than picking the two that flatter you.
Ashwell scored itself and eight competitors on eight candidate attributes, which is 28 pairs of axes it could have drawn. The board deck used two of them: module breadth against modern architecture. Across all nine players the architecture score spans ten points of a hundred, so nothing on that axis is a difference. Once the axis is rescaled to fit the box, the six points between Ashwell and Fernhurst become 60% of the chart height.
The obvious repair is to swap the fake axis for a real one, so the second draft used module breadth against enterprise proof. Both spread the set properly, which is why the draft looked rigorous. They correlate at 0.958, so a single component accounts for 97.9% of the variance between them. The map shows it: five players in the top right box, four in the bottom left, and the other two boxes empty.
The pair that survives every test is how you buy it against how deeply you can configure it, correlated at negative 0.11, with all four boxes occupied. Ashwell is not alone in a corner: it shares its box with Baverstock and Ilmington. Written for whoever presents the landscape slide and then takes questions on it. The competitor teardown is where one player's score gets sourced properly, review mining tests which complaints you can honestly claim, and the lost deal patterns say who you actually lose to.
Why the empty quadrant is not white space
Two axes that move together are one axis. NIST's engineering handbook is direct about the situation: when an analysis starts with a substantial number of correlated variables, principal component analysis is the dimension-reduction tool for it. On Ashwell's set the five attributes that cleared the spread test collapse into two families. Inside a family the correlations run from 0.851 to 0.958. Across families they run from 0.050 to 0.292. Two dimensions, so one map, and which attribute labels each axis is a choice worth stating.
The maps that get taken seriously publish their scales. Forrester's description of its own Wave methodology says the analyst writes scale explanations for every criterion and scores each vendor against them. It also weights the criteria by importance and hands clients an interactive version they can re-weight. That is the standard an internal landscape slide has to meet. So the deliverable here is the scoring Sheet, with a source and a date in every cell, and the picture is downstream of it.
Ease of use is the attribute everyone asks for and nobody scores. G2 publishes how it does: satisfaction is assembled from named review questions, ease of use among them, weighted by review volume, recency and completeness. That is a measurement. The public proxy across this set runs 4.2 to 4.6 out of five, ten percent of the scale, so it cannot orient anything. Entry price does spread, from 19 to 72 dollars, but only five of the nine publish one.
How it works
Score from published evidence
Every cell traces to a page, a document or a price list, and carries the date it was read.
Test each attribute
Coverage first, then spread, because an axis the set does not differ on carries nothing.
Measure independence
The correlation matrix across surviving attributes, which is where the flattering pair usually dies.
Check the boxes
A pair that leaves two quadrants structurally empty is a ranking wearing a map's costume.
What you get
- Every player scored on every candidate attribute, each cell carrying its published source and date.
- A spread test per attribute, so an axis nobody in the set differs on never reaches the chart.
- The full correlation matrix, with the pairs that are secretly one measurement named.
- Box occupancy at the midpoint split, including the boxes that are empty by construction.
- The map drawn on the surviving pair, plus the same map under the alternative labellings.
- A Sheet of the scoring, a diagram of the map, and the written rationale for both axes.
Common questions
Shouldn't the axes come from what buyers care about?
Buyer criteria are the right source for the candidate list and the wrong place to stop. Ashwell's buyers care about ease of use, and across these nine vendors the only public proxy for it runs 4.2 to 4.6 out of five. An attribute buyers weigh heavily but nobody differs on explains nothing about where anyone sits.
What is actually wrong with a correlation of 0.958?
The two axes are one measurement, so the map has one real dimension. A single component accounts for 97.9% of the variance between Ashwell's second pair, all nine players land on a diagonal, and two of the four boxes hold nobody. That chart is a ranking drawn at 45 degrees.
Is an empty quadrant not the white space?
Sometimes, and the test is whether the box is unclaimed or unreachable. On the second draft the empty boxes were wide module breadth with no enterprise proof, and its mirror. Neither exists in this category because case studies accumulate as products broaden. Nobody is leaving that space open for you.
How do you score something like configurability?
From documented API endpoints and whether custom objects are supported, both countable in public developer documentation, with the count and the date in the cell. Where a vendor keeps its reference behind a login the cell is left unscored, and an attribute scoreable for only some of the set cannot be an axis. The same scored matrix feeds an RFP fit sheet.
Do we need more than one map?
It depends how many independent families survive. Ashwell's five attributes collapsed into two, so one map carries the set; three families would need a second. Worth knowing: Ashwell's two nearest neighbours are the same under all three labellings of those families, which is what makes the placement worth defending.
Where does price go?
On the map as an annotation, not as an axis. Entry price spreads well here, from 19 to 72 dollars per user, but only five of the nine vendors publish one, so an axis built on it would be estimating four positions. Packaging structure belongs in the pricing page review.
What actually gets delivered?
A Sheet with a row per player and a column per attribute, every cell carrying its source and date, plus the correlation matrix and the gate outcomes. A diagram of the map. A Doc arguing both axes and listing the rejected pairs. Rep-facing framing belongs in the battlecard pack.
Competitive Positioning Map With Real Axes
Fill in the form and your workspace opens with the work already underway.