A budget meeting becomes easier when every activity can be ranked by one number. The apparent precision reduces disagreement, shortens the conversation and gives the decision a defensible-looking record. But ease is not the same as adequacy. My position is that attribution should inform a marketing allocation without becoming the rule that determines every investment. The final recorded interaction may describe part of a customer's route to an application. It does not automatically identify the additional business created by that activity, or settle the contribution of work the recording system cannot assign.
This position has an uncomfortable implication for marketing leaders. If the model does not decide the budget, we must explain our own judgement more clearly. It is not enough to say that brand matters or that some effects take time. Those statements can become an excuse for preserving expenditure indefinitely. A leader who rejects a narrow ranking must offer a better decision account: the objective, alternatives, amount at risk, evidence available, uncertainty remaining and occasion for reconsideration. Otherwise the objection to false precision merely protects an equally weak form of intuition.
Ask what the report is permitted to decide
In Veldoran's fictional 2025 planning round, a proposal sought to move C$2.6 million from brand activity into channels ranked by the last click attached to completed applications. The route from report to recommendation looked direct. Activities with a recorded conversion received more funding; activity without the same assignment received less. I asked the analysts to describe what the model observed before we treated the ordering as an investment instruction. This was not a request to discredit the analysis. It was a request to establish which question the analysis had actually answered.
An application may follow several encounters, including some the recording system sees poorly or not at all. The last recorded event can still be useful for understanding a path. The problem begins when the event is described as the cause of the application, then used to infer what another increment of spend will produce. Those are further propositions requiring their own support. A clear report should distinguish an observed sequence, an attribution convention and an estimate of incrementality. Combining them in one heading makes the budget look more certain than the evidence allows.
The analysts were right to challenge me in return. If I believed the model missed material influence, what decision-relevant evidence could I provide? It would be unreasonable to demand a rigorous case from acquisition while allowing the brand account to rest on familiarity. We examined the purpose of the activity, its audience and the alternatives for the disputed authority. Finance helped separate the commercial question from the reporting convention. We did not suddenly possess a complete causal model. We had a more explicit description of the limitations within which the allocation had to be made.
Unmeasured value still needs a bounded case
I retained C$1.8 million of the disputed amount for a six-month brand programme and moved C$800,000 to the proposed acquisition routes. The split could look like a compromise designed to satisfy everyone. The CEO asked that question directly. My response had to explain why the retained programme still deserved examination, what the reduced amount would fund and why six months was a reasonable point for review. I could not present the absence of attribution as evidence that the programme was valuable. Missing evidence creates uncertainty; it does not automatically favour the expenditure I would like to preserve.
The authority was therefore tied to a defined programme and a fresh decision date. We recorded the evaluation questions and the conditions under which the activity would be narrowed or stopped. This made the argument more demanding for my team. A general account of long-term value could no longer substitute for explaining the actual work. It also made the acquisition account more honest, because the transferred C$800,000 was not accompanied by a claim that the existing ranking had proved the incremental return from that future amount. Both allocations carried assumptions.
There is a practical risk in making every decision conditional: the organisation can spend more time reviewing than acting. I do not propose a committee meeting for every placement. The depth of the account should reflect the amount, duration and strategic importance of the commitment. In this case, C$2.6 million was sufficiently material to deserve a documented judgement. Smaller decisions could use agreed principles within delegated authority. The point is to prevent a reporting convention from silently replacing executive accountability, not to require an elaborate essay before routine marketing work can proceed.
Record the judgement that survives the model
At six months, the programme's activity and spend had been reconciled, but we still could not attribute an incremental return confidently. That was an important result to retain. It would have been tempting to assemble favourable observations into a success narrative and carry the authority forward. Instead, we continued a narrower programme under a new decision. The original amount did not renew itself merely because the work had been completed. Completion, commercial effect and the case for another period remained separate questions in the account presented to colleagues.
This discipline also helps a board understand what management is asking it to accept. Directors may see a strategy built around customer growth while the underlying budget depends on several different forms of evidence. They should be able to ask which claims are observed, which are modelled and which represent management judgement. They need not select channels or design the attribution model themselves. Their interest is whether the resource commitment is being made with a proportionate account of uncertainty and whether the organisation can recognise evidence that should change its course.
I would start the conversation earlier in a future planning cycle. When a dispute about attribution appears only at the budget deadline, it can sound like the owner of an established programme resisting accountability. Agreeing the decision questions in advance would allow analysts, finance and marketing to prepare evidence for the same purpose. It would also reveal where a different research design might be worth undertaking before the next allocation. Some uncertainty can be reduced through better preparation; some must remain part of the judgement. The two should not be confused.
A leadership website should preserve that distinction rather than use the case to advertise an instinct that supposedly outperformed the numbers. My contribution was to challenge the permitted inference, accept a partial transfer, define a bounded continuation and return to the decision without claiming a return that had not been established. The model remained useful. It simply did not become the constitution of the budget. A responsible marketing account leaves room for analysis and judgement while making it possible for another executive to see where one ends and the other begins.