What Did Insights Actually Change? Build a Decision Ledger (September 2026)

Sep 22, 2026 by Marcos Dymond, Head of Growth

I'll be frank: attribution math breaks the moment finance points it at insights work. Revenue chains are too diffuse, cost savings depend on a launch that never happened, and hours saved just hands back budget. What survives the conversation is a simple ledger, one row per decision, filled out the week the call is made. Here's how to build it.

TLDR:

• Prove insights ROI by tracking decisions changed, not hours saved or studies produced
• Build a Decision Ledger with six fields per row, capturing the counterfactual the week the call is made
• Log five decision types: go/no-go, reallocation, risk avoidance, speed compression, strategic bets
• Pitch substitution, not addition: name the agency retainer or tracker wave your function replaced
• A connected intelligence layer like Merciv turns the 90-day debrief into a lookup by preserving citations and confidence tiers

Why "What Did Insights Actually Change?" Is the Hardest Question Insights Leaders Get Asked

Finance stopped counting decks. In 2026 budget reviews, the question landing on Heads of Insights is sharper: what did any of this actually change? Study counts and tracker waves no longer clear the bar. Output volume was the proxy finance tolerated until a cheaper one showed up, and insights shelfware is the predictable result.

The reframe that survives the conversation is decisions, not hours. Hours-saved math invites a discount fight ("if it's faster, charge less") and finance builds their own calculator anyway. Decisions changed is harder to fake.

A Decision Ledger is the running record that makes the answer defensible: for each material decision, what the call would have been without the insight, what it became, and the evidence that closed the gap.

The Reason Traditional ROI Math Fails for Insights Teams

Three standard CFO formulas get pointed at insights work, and all three break on contact.

Revenue attribution fails because the causal chain is diffuse. A pricing decision that lifted a category four points had a segmentation study, a competitive read, a merchant's instinct, and a promo calendar behind it. Handing the study full credit is fiction; handing it none is worse.

Cost savings fails because the counterfactual is unknowable. You cannot invoice a launch that never happened because the concept test killed it.

Hours saved fails structurally. Faster work invites a smaller budget, not a bigger one. Research should be judged by the decisions it changes, not by the activity or throughput behind it.

More than 75% of CEOs at consumer-facing companies believe customer insight is critical to accelerating growth, yet measurement of the function itself remains rare — fewer than half of companies regularly measure the ROI of their insights work. Belief without measurement is how a budget line gets cut in a soft quarter.

The Decision Ledger changes the unit of account from output or hours to decisions changed, with the counterfactual captured at the moment the call is made. That is a foundational reset when building a consumer insights strategy from the ground up.

What a Decision Ledger Is and Why It Beats a Research Log

A research log tracks what the team produced: studies, waves, decks. A project tracker records activity. Neither answers the decision question, which is the core challenge behind any customer insights to action strategy.

The Decision Ledger is an operating artifact. One row per non-trivial decision insights touched. Six fields, in a spreadsheet if that is all you have:

The ledger belongs to the insights lead, updated within a week of each decision. Reconstructing it at budget season is how the counterfactual gets rewritten by memory.

The Five Categories of Decisions Worth Logging

1. Go/no-go calls. Launches, reformulations, line extensions, discontinuations. The counterfactual is loudest here: what would have shipped, or stayed on shelf, without the read.
2. Resource reallocation. Media mix rebalances, retailer prioritization, SKU rationalization, promo calendar changes. The dollar amount moved is the natural size stamp on the row.
3. Risk avoidance. Pulling a claim that failed substantiation, delaying a launch after a concept test hit a wall, changing pack after a complaint cluster surfaced. Hardest to defend later because the bad outcome never happened, which is why the counterfactual must be captured the week the call is made.
4. Speed-to-decision compressions. A stalled decision that moved because a readout landed in days instead of weeks. The unit is the decision that would otherwise have slipped a quarter or been made on instinct.
5. Strategic direction bets. Segmentation resets, whitespace commitments, category redefinitions. Log with lower confidence and check at 12 months instead of 90 days.

Skip routine reporting, tracker refreshes with no action attached, and anything where insights was cc'd but not consulted. It's the same logic behind triaging ad-hoc research requests smarter.

How to Capture the Counterfactual Without Guessing

The counterfactual is the fragile part of the ledger. Captured after the outcome is known, it reads as revisionism. Captured before, it holds up under a skeptical read. Three techniques get it on the record at the right moment.

Pre-decision briefs, signed by the decision-owner. Before the research lands, the stakeholder writes one paragraph: what they would do today with the information they already have, and why. The insights lead countersigns and files it. When the readout changes the call, the delta is documented in the decision-owner's own words, dated before the evidence existed.

Recorded stakeholder positions in the readout. Open by asking each principal to state their current leaning before findings are presented. Log it in the appendix with names and timestamps. The room's starting position is the counterfactual; the ending position is the decision.

90 and 180 day debriefs. Calendar the check the week the decision is made. At 90 days, revisit the observable metric; at 180, the strategic ones. A debrief that references a pre-signed brief and a recorded room position is a paper trail a CFO can audit, which is critical for VPs defending consumer insights numbers in budget meetings.

The Metrics That Belong on the Ledger (and the Ones That Don't)

Track on the ledger:

• Decisions influenced, tagged by category (go/no-go, reallocation, risk, speed, strategic).
• Decision-cycle compression: weeks between question asked and decision made, against the team's prior baseline.
• Avoided-loss estimates on risk-flagged rows, captured in the pre-decision brief so the number is not backfilled, a discipline covered in the lean team playbook for insights directors.
• Leadership citation rate: share of QBR decks, brand plans, and capital requests that name an insights source by finding.
• Recommendation adoption rate: how often the call the team recommended is the call leadership made.

Cut from the ledger:

• Hours saved. Invites a discount conversation.
• Studies completed, waves fielded, reports produced.
• Dashboard views, portal logins, deck downloads.
• NPS on the insights function.

Every metric points at a decision, not an artifact.

Building the Business Case: Substitution, Not Addition

Additive asks lose in flat budgets. Substitution asks win because they name the line item you are replacing, not the one you are adding to, a framing especially relevant in the first 90 days as insights head.

Line the ledger up against costs finance already sees:

Ranges above are illustrative order-of-magnitude figures based on typical enterprise contracts we've seen across CPG and retail insights teams. Verify against your own vendor invoices and statements of work before using them in a CFO pitch.

The pitch reads: here is what these three lines cost last year, here are 14 decisions our function absorbed against them, keep the function and cut the line.

How to Present the Ledger to Different Stakeholders

Different stakeholders read the same ledger through different lenses. Rebuilding the artifact five times is how the exercise dies. Route cuts of one underlying record instead.

For the CFO, surface substitution rows against named line items alongside avoided-loss estimates from risk-flagged decisions, and lead with dollars substituted and dollars protected. For the CMO, surface marketing calls influenced (media mix, campaign, claim, pack) with cycle compression, and lead with share of brand-plan decisions touched and weeks compressed. For the CEO, surface the two or three strategic bets the function unlocked (tagged as direction rows), and lead with those named bets, counterfactual attached. For the board, surface citation rate on the year's major capital decisions, and lead with the percent of capital-request decks citing a finding by source.

One ledger, four routings. Only the columns surfaced and the summary line change. If the record cannot support four cuts without rework, it is not structured well enough yet, a sign that leadership buy-in for consumer insights may still be missing.

Common Failure Modes When Standing Up a Decision Ledger

Five failure modes take out most ledgers before the second budget cycle.

Retroactive scoring. Rows filled in after the outcome is known read as revisionism, and any CFO who has sat through an attribution debate will spot it in one pass. If the counterfactual was not captured the week the call was made, the row does not go on the ledger.

Attribution inflation. Claiming a launch that had seven inputs and one insights read. Log the decision only when the recommendation is on record before the call, and mark contribution as sole, primary, or contributing. That standard matters when producing board-ready consumer insights without black-box AI.

Undercounted risk wins. Avoided-loss rows are invisible by design. Fix at capture: the pre-decision brief names the loss size in the decision-owner's words, before the risk is retired.

Compliance drift. When the ledger becomes a form filled out for legal cover, entries get thin. Keep it short, owned by the insights lead, and reviewed live in the QBR.

The harder failure sits underneath all four: a function that logs decisions but never routes findings to the person who owns them. A complaint cluster surfaces on a hero SKU and lands in a workspace nobody was assigned to read. The decision never happens because the signal never arrived. Signal ownership per SKU or category, with a named recipient and a same-week routing commitment, is the prerequisite.

What Changes When the Ledger Runs on a Connected Intelligence Layer Like Merciv

The ledger discipline is yours to hold regardless of tooling. What a connected intelligence layer removes is the manual reconstruction that usually kills the practice after two quarters.

When every output carries a citation, retrieval date, and three-tier confidence score (High, Directional, Exploratory), the evidence column becomes a clickable link, not a rebuilt PowerPoint appendix. It is the same rigor behind AI-powered share-of-voice reporting for marketing leaders. A Q1 decision logged against a Directional read can be re-queried in Q3 against fresh signal, because prior tracker readouts compound as reusable context in the knowledge base instead of decaying in a shared drive. The 90-day debrief becomes a lookup.

Final Thoughts on Defending Your Insights Team's Value in Budget Season

The teams that keep their headcount are the ones who can point to decisions changed, not decks produced. Start the ledger this quarter, capture counterfactuals before outcomes land, and pitch substitution against line items finance already sees. Merciv's enterprise setup is where citations, confidence scores, and prior readouts compound into the evidence trail this practice needs.

FAQ

How do you prove consumer insights team ROI without falling into the hours-saved trap?

Anchor the business case on decisions changed, not time saved. A Decision Ledger captures the counterfactual (what leadership would have done without the read) at the moment each call is made, then pairs that record with substitution math against named line items (agency retainers, tracker waves, redundant listening seats) so finance sees the function replacing spend, not adding to it.

Decision Ledger vs research log: what's the actual difference?

A research log tracks outputs; a Decision Ledger tracks decisions influenced, with counterfactual and outcome per row. See the full breakdown above.

Can I build a Decision Ledger in a spreadsheet, or do I need a tool like Merciv?

Yes. The six-field structure (decision, question asked, recommendation, decision made, counterfactual, observable outcome) runs in a spreadsheet, and the discipline is yours to hold regardless of tooling. What a connected intelligence layer removes is the manual rebuilding of the evidence column at debrief time: when every finding already carries a citation, retrieval date, and confidence tier, the 90-day check becomes a lookup instead of a reconstruction project.

What's the best way to capture the counterfactual so it holds up to CFO scrutiny?

Capture it before the readout lands, in the decision-owner's own words. Pre-decision briefs signed by the stakeholder, recorded room positions logged in the readout appendix with timestamps, and calendared 90 and 180 day debriefs give you a dated paper trail. A counterfactual reconstructed after the outcome is known reads as revisionism, and any experienced CFO will flag it in one pass.

Which decisions belong on the ledger and which should you skip?

Log go/no-go calls, resource reallocations, risk avoidance, speed-to-decision compressions, and strategic direction bets, anywhere the recommendation was on record before the call was made. Skip routine reporting, tracker refreshes with no action attached, and any decision where insights was cc'd but not consulted; padding the ledger with weak rows is how attribution inflation gets the whole artifact discounted.