How to set growth targets without fake precision
Guide · Frameworks · 6 min read · last verified 2026-07-27
A growth target is a commitment to move a measured position in a stated direction, within a stated range, by a stated review date. Fake precision is what replaces that commitment in most planning documents: a number with decimal places and no lineage. Nobody in the room can say what was measured to produce it, which assumptions connect today's position to the promised one, or what anyone will do differently if it is missed. The decimals signal rigor; the missing lineage guarantees theater. Setting targets without fake precision means doing the opposite work: derive the target from a baseline you actually measured, name the assumptions that connect baseline to target, label each assumption as evidenced or hypothesized, and agree in advance that a miss triggers an interrogation of the assumptions rather than the team.
Where fake precision comes from
Fake precision is specificity that exceeds its evidence, and it enters plans through three doors. The first is negotiation: a confident point number ends a debate, while a range invites one, so planning meetings quietly reward whoever states the most decimal places. The second is spreadsheet arithmetic: a model multiplies one guess by another guess, and the output carries decimal places the inputs never contained — an artifact of the math, not a property of the knowledge. The third is anchoring: last cycle's number plus an ambition increment, a formula that requires no contact with the market at all.
All three doors share one property: the number exists before any measurement does. You can diagnose fake precision with a single question, asked kindly, of whoever owns the target: "what's this number derived from?" A real target has an answer that opens — a baseline scan with a date on it, a documented experiment, an assumption ledger. A theatrical target has a story about how the number felt achievable but ambitious. Precision and accuracy are different properties, and a plan needs the second far more than the first.
The honest target: direction, range, review date
An honest target has three parts, and none of them is a decimal.
Direction names the measured position that should move, and which way. Not "grow awareness" but "increase the number of benchmark buyer questions whose answers cite us." Not "improve win rate" but "raise the share of evaluations in the named segment that end in a win." Direction forces the target onto something that can be measured twice.
Range replaces the point estimate with a floor and a ceiling: the floor is the result that would justify the effort, the ceiling the result you would defend to a skeptic given the evidence in hand. A range is not hedging — it is a confidence statement, and stating confidence honestly is the whole job. Point targets pretend the future has one value; ranges admit variance while still committing to a direction.
Review date is the day the position gets re-measured — not the day the team hopes to feel good. Without a fixed re-measurement date, a target can be renegotiated forever and is therefore not a commitment.
"Grow pipeline meaningfully this half" is a wish. "Move from being cited on four of our forty locked benchmark questions to somewhere between ten and sixteen by the April re-scan" is a target: direction (cited presence, upward), range (ten to sixteen), review date (the April re-scan). The precision lives where it belongs — in the measurement protocol — and the humility lives where it belongs, in the range.
Derive the range from a measured baseline
A range without a baseline is still fiction; it is merely fiction with error bars. Before any target-setting conversation, measure where you stand. For visibility in AI answers, the protocol is the one described in how to set an AI visibility baseline: fix a set of your buyers' real questions, record which answers cite you today, and lock the set so the later re-scan measures movement rather than drift. The same locked-benchmark discipline transfers to any position worth targeting — if the thing you re-measure in April is not the thing you measured in January, the comparison is decoration.
The baseline does two jobs. It converts direction from aspiration into arithmetic: you know the starting value, so movement is subtraction, not narrative. And it disciplines the range: the floor becomes the movement you would be embarrassed to miss given the work planned; the ceiling becomes the movement you could defend given the evidence available. If the current position can't be stated with a source attached, the target meeting is premature — schedule a measurement instead and set the target next week.
Name the assumptions, then label them
Every target is a conclusion resting on premises, so write the premises down. An assumption ledger is a short table attached to the target, where each entry carries one of two labels: evidenced, meaning it comes with a source that opens — a dated scan, an interview note, a documented test — or hypothesis, meaning it is believed, plausible, and untested. There is no third label.
| Assumption behind the target | Label | What would upgrade or test it |
|---|---|---|
| Buyers in this segment research the problem before contacting vendors | Evidenced — question scan, dated and archived | Re-scan next quarter |
| Publishing implementation guides will change which sources get cited | Hypothesis | A documented experiment with a before-and-after re-scan |
| Sales will use the new material on live evaluations | Hypothesis | Call notes reviewed in the weekly review |
Each hypothesis then becomes a test with a verdict date — how to document growth experiments describes the format, and the discipline matters more than the template. The labels are the point. Teams rarely fail because they hold assumptions; they fail because assumptions travel unmarked through a plan and quietly acquire the status of facts.
When you miss, interrogate the assumption
A missed target is information, and the ritual that follows the miss determines whether the information is ever collected. The default ritual blames the team. Its output is predictable: next cycle everyone sandbags, ranges widen for political rather than evidential reasons, and the planning process loses contact with the market entirely.
The alternative ritual reopens the ledger. Walk the assumptions in order and ask which one failed. Three verdicts are possible. The assumption was wrong: the evidence changed or was thinner than labeled — update it, and check what else rests on it. The execution never happened: the experiment the hypothesis required was not actually run — the documentation trail shows this in minutes if the trail exists. Or variance: the position moved, just inside the noise — which means the range was set too narrow, and the fix is a better range, not a louder effort. Each verdict points at a different repair, which is exactly what the blame ritual can never produce.
A leader sets the tone in one sentence: "we didn't miss a number, we disproved an assumption — let's find which one." Teams that hear that sentence bring bad news early, which is the only time bad news is useful.
Keep the loop short
Targets fail quietly in the space between review dates, so give them a cadence. A weekly growth review checks the leading signals and the state of each running experiment; the review date delivers the verdict; the verdict resets the baseline for the next cycle. For targets about answer presence, how to measure whether content changed AI answers covers the re-scan mechanics. In the five questions every growth plan must answer, fake precision is a device for dodging question five — did it work? — by making the target too vague in substance to ever be falsified, however precise its decimals. A market growth intelligence platform such as Magrios exists for the opposite habit: it locks the benchmark, re-scans it on the review date, and attaches an openable source to every position claim, so the target conversation is about assumptions and evidence rather than beliefs and blame.