Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Marketing Operations › Process

Prioritising Marketing Operations Work

How to decide which operations fixes come first: one backlog fed by rework, errors, audits and requests, sorted by effect, risk and effort, with protected time.
Guideintermediate10 min readUpdated

Why operations work loses to campaign work

A campaign has a launch date, a requester and a visible result. An operations fix has none of those. Adding a field to the intake form, rewriting a procedure, cleaning the campaign taxonomy or naming a deputy for an automated journey all save time and prevent errors later, but nobody is waiting for them today, and nobody notices on the day they are skipped. Put the two side by side in a weekly meeting and the campaign wins every time.

The cost of that pattern is hidden because it is spread out. The rework that a missing intake field causes is absorbed, a little at a time, by everyone who chases a missing detail. The error that a missing check lets through is fixed by whoever finds it. The process hub explains how cycle time and rework show these costs once they are measured, and the marketing operations hub ends its first review by choosing the two fixes that remove most rework. Doing that once is easy. Doing it every quarter, with a list that keeps growing, needs a backlog, a way of ordering it and time to work on it.

The decision here is narrower than general marketing prioritisation. The guide to prioritising work and capacity scores requests against objectives, plans them against capacity and routes urgent work. This guide decides only the operations function's own improvement work: what to change about the way the team works, and in what order.

Where operations work comes from

Operations work rarely arrives as a request. It surfaces as a by-product of other work, and unless someone collects it, it disappears back into that work. Each source below produces candidates for the backlog, and each tends to produce a different kind of item.

SourceWhat it producesWhere it is covered
Rework reasons at review gatesUpstream fixes: intake fields, brief standards, build checksCampaign workflow guide
Error reviewsProcess causes and owned actions after a mistake reached customersReviewing a marketing error
Process mapsWaits, loops back, workarounds and steps with no ownerMapping a marketing process
Approval and compliance auditsGaps in the matrix, missing evidence, reviewers chosen by seniorityApprovals and compliance guide
Data auditsTaxonomy drift, list hygiene, suppression gapsData operations guide
Stack auditsTools to retire, consolidate or reassignMartech stack audit
The risk registerRisks whose prepared response needs a process changeMarketing risk register
Requests from the teamNew work types, fields, templates or reportsThe work management system's intake
Recurring symptomsReports that take days, names that differ by platformOperations troubleshooting

The maturity ladder on the operations hub is a further source of a different kind. It does not list fixes, but it tells a team which problems to expect at its stage, and a backlog full of items from the rung above the one the team is actually on is usually a sign of fixing the wrong things first.

One backlog, separate from requests

Collect every candidate in one list, kept in the same work system as everything else, under a work type of its own. A single list matters for the same reason a single front door matters for requests: improvements scattered across review notes, audit reports and someone's personal notes cannot be compared, so the one that gets done is the one its author remembers to push.

Keep that list separate from campaign requests. Mixing them forces a comparison that operations work cannot win, and it hides how much improvement work is waiting. A separate backlog also lets operations use criteria that suit improvement work, which are different from the objective-fit and deadline questions that decide campaigns.

Each item needs enough to be judged without a meeting. A short, consistent record keeps the backlog usable as it grows.

FieldWhat it holds
The problemWhat goes wrong today, in one or two sentences, with the evidence that shows it
SourceThe review, audit, map, request or symptom it came from
Where it actsThe stage, gate, handover or system it would change
Proposed changeThe smallest change that would fix the problem, if one is known
EffectWhat would improve, and whether it acts on the step that limits delivery
Risk if leftThe consequence if it is not fixed: errors reaching customers, data or consent exposure, lost evidence
EffortA rough size and who would need to be involved
OwnerOne person accountable for the item once it is chosen

Name items as problems first and solutions second. "Requesters do not give a campaign name in the agreed Naming Convention, so names are invented at launch" leads to a better discussion than "buy a naming tool". Merge duplicates as they arrive; the same problem raised by several sources is a stronger candidate, and its record should say so.

Sorting by effect, risk and effort

Order the backlog in two passes. The first pass is a quick sort with the impact-effort matrix: place each item by its expected effect and its effort, and the list falls into quick wins, larger projects, fill-ins and items to drop. The matrix suits a mixed list of fixes because it claims no more precision than the estimates hold.

Effect is not the same as size. The theory of constraints explains why an improvement anywhere except the step that limits delivery adds nothing to the total; it only moves work faster into the queue in front of that step. So weight effect by where the item acts. A modest change to the review queue that holds every launch outranks a larger change to a step that already waits on others. Find that step from cycle-time records by stage or from a process map, not from the loudest complaint.

Risk items need separate handling, because the matrix treats every item as a trade between effect and effort. A missing check on consent, an approval route that leaves no evidence, or a suppression list that does not reach every sending tool may be raised rarely and look like small effort, yet carry consequences far larger than any time saving. Weigh these by the consequence if they go wrong, as the risk register does, not by how often they come up. Findings from approval, data and stack audits fall into this group more often than rework fixes do.

Where the order is contested, because several stakeholders favour different items or two large projects compete for the same people, move from the matrix to a weighted scoring matrix. Agree the criteria and their weights before scoring, such as effect on the limiting step, consequence of the risk removed, effort, dependency on other items and how quickly the benefit appears, and back each score with evidence. The scored order is still an input to a decision, not the decision itself. The comparison of the impact-effort matrix and the weighted scoring matrix sets the two tools side by side, from the number of items each suits to the precision each claims.

Dependencies cut across both passes. A fix to the campaign taxonomy may need to land before any reporting fix is worth doing; a new intake field may be pointless until the intake form is the only route in. Order dependent items together and do the enabling item first, even if it looks modest on its own.

Protecting time for the backlog

A well-ordered backlog with no time to work on it is a list of good intentions. Operations work needs time that campaign work cannot take, agreed in advance with whoever plans the team's capacity, and the agreement matters more than the amount. There is no correct share that suits every team; start with what the team can keep through a busy period, record what it delivers and adjust it with that evidence at the next planning session.

Three practices keep the time protected. Plan operations items into the same capacity view as campaigns, so they are visible when someone proposes taking the time back. Give each chosen item one owner, an expected finish and a way to show it worked, exactly as a campaign has. And when urgent work does take the time, record it, so the cost of borrowing becomes visible rather than routine.

Measure the backlog's effect on the system, not on the people in it. Cycle time by stage, rework by gate and the share of work passing its gates first time are operational measures; turning them into a KPI for individuals invites people to skip gates to hit them. Record a Baseline for the measure a change is meant to move before the change goes live, so its effect can be read afterwards.

Sequencing changes the team can absorb

A team can only take in so much change at once. Every new field, step, template or rule has to be learned, and several arriving together blur into one disruption, make it impossible to tell which change helped and give people a reason to work around all of them. Choose a few items for each period, finish them and let them settle before starting the next batch.

Prefer the smallest change that acts on the problem. An intake field, a definition of complete at a handover, an owner named for a step or a check added to an existing procedure can be made quickly, explained in a sentence and undone if they do not work. A new tool, a restructured workflow or a new approval tier may be the right answer, but they arrive slowly, change many things at once and are hard to reverse. Try the small change first and reach for the larger one when the small one has been shown not to be enough.

Tell the people affected what is changing, why and from when, and update the procedure or template the change lives in on the same day. Then look again at the measure the change was meant to move once enough work has passed through, and record the result against the backlog item before closing it. Items that did not work go back on the list with what was learned.

A logistics software company orders its backlog (illustrative)

Illustrative example — not a real result

Tobbrique Logistics Software orders its operations backlog

Tobbrique Logistics Software, an invented company selling route-planning software to freight firms, had a marketing team of 9 and its first operations manager. She found operations ideas in 5 places: rework reasons recorded at the review gate, the actions from 3 error reviews, an approvals audit, requests from the team for new work types and a long list of complaints in the team channel. Gathered into one backlog and merged, they came to 26 items.

The work system's stage history showed that the review queue limited launches: over the previous quarter, campaigns waited a median of 7 working days for review, against 2 for any other stage. The top rework reason at that gate, on 11 of 30 returned items, was a customer claim with no supporting evidence attached, which sent the work to product marketing and back.

The first sort put a new review tool, requested by 2 senior marketers, among the major projects. It did nothing about missing evidence. A change to the intake form, adding a required field for any customer claim with a link to its evidence, was a quick win acting directly on the queue. The approvals audit had also found that approvals for partner co-marketing left no record; it was raised once, but because an unrecorded approval of a partner's claim carried a serious consequence, it went into the first batch on risk.

The team chose 3 items for the quarter: the claim field, an approval record for partner work and a short procedure for attaching evidence. It agreed a fixed half-day each week for operations work. A quarter later, claim-related returns had fallen to 3 and the median wait for review to 4 working days. The review tool stayed on the backlog, re-sized against the new baseline.

Common mistakes

  • Mixing operations work into the campaign backlog. It loses every comparison. Keep its own list and its own time.
  • Fixing the loudest complaint. The step people complain about is often not the one that limits delivery. Find that step in the records first.
  • Weighing risk by frequency. A rarely raised gap in consent, evidence or access can matter more than any time saving. Weigh it by consequence.
  • Starting with a tool. A new tool changes many things slowly. Try the smallest change that acts on the problem first.
  • Too many changes at once. The team cannot absorb them and nobody can tell which one worked. Finish a few, then start the next.
  • Protected time that is not protected. If campaigns can take it without anyone noticing, they will. Plan it in the capacity view and record each borrowing.
  • No baseline. Without the measure recorded before the change, its effect is a matter of opinion.
  • Closing items without checking they worked. Record the result against the item; an item that did not help goes back on the list with what was learned.

Frequently asked questions

Should operations improvements go in the same backlog as campaign requests?

No. Campaign requests are judged against marketing objectives and deadlines, and almost any improvement loses that comparison in a busy week, however much time it would save later. Keep operations work in its own backlog, sorted by its own criteria, and give it time that campaign work cannot take. The two lists meet only when a change needs people who also deliver campaigns, which is a capacity decision for the marketing lead.

What should a new operations lead fix first?

Whatever most limits the team's delivery or carries the most serious risk, which is rarely the loudest complaint. Trace a few recent pieces of work to find where they wait and what is redone, collect the findings from any error reviews and audits, and sort them by effect on that limiting step, by consequence and by effort. Small changes at intake or at a handover often come before a new tool, because they act sooner and cost less to undo.

How much time should a team set aside for operations improvements?

There is no standard share that suits every team, and a figure borrowed from elsewhere will not fit your demand or staffing. What matters is that the time is agreed in advance, protected from campaign pressure and spent on the top of the backlog. Start with an amount the team can keep, record what it delivers and adjust it at the next planning session with that evidence.

Terms used on this page

Frameworks

Where this page is used

Also related

Related