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

Marketing Operations › Troubleshooting

Marketing Teams Duplicating Work

Why two marketing teams end up making the same asset, buying the same tool or running the same study: many front doors, scattered work and no shared record.
Troubleshootingintermediate12 min readUpdated

First, trace the last few duplicates back to where each began

Start with evidence rather than a reorganisation. Collect the duplicates people already know about, such as a brochure built twice, two subscriptions for the same kind of tool or a survey commissioned without knowledge of an earlier one. For each pair, note where each piece of work began and who asked for it, how each request reached the person who did it, where each was tracked while under way, and what the second team could have found before starting, had it looked. One row per duplicate is enough.

The answers point to a join. A request that reached two people by different routes points to intake; work nobody outside one team could see points to the work system; a rebuilt asset points to the library, a repeated study to the research store and a second tool to the way tools are bought. The same kind of duplicate in several regions means nobody split shared from local work, and teams building what operations already offers means nobody wrote down what it offers.

Then decide who answers for each shared service. The RACI matrix holds to one rule: each row has exactly one accountable role. Write a row for each join the trace exposed, such as triaging requests, keeping the library or approving a new tool, and name the role accountable for it.

This page is about effort spent twice. If the visible cost is time, with work arriving late rather than twice, the diagnosis for campaigns that keep launching late is the better start.

Diagnosis table

Diagnosis

SymptomCheckLikely causeFix
Two people find mid-job that they were asked for the same thingHow each request reached the team, and to whomRequests reach the team through several routesOne front door for every request, triaged against work already open
Work under way in one team is started again in anotherWhere each team tracks its work, and whether others could see itWork lives in inboxes, chats and private spreadsheetsOne work system with an owner, stage and date for every item, visible across teams
A team rebuilds an approved asset, or reuses an out-of-date oneHow long someone outside the owning team takes to find the current approved fileThe approved asset is harder to find than an old copyOne approved copy of each asset in a library, old folders closed
Two teams commission studies that answer the same questionWhether the first study's findings could be found, trusted and seen to applyFindings sit in one team's deckFindings kept as tagged records and searched at the start of every brief
Card statements or renewals show two tools doing one jobHow each tool was bought, and whether anyone checked what was ownedTools bought by request on a cardOne route for tool requests, triaged against the tools already owned
Each region or product team keeps its own rules, reports and tools for the same purposeWhether anyone has written down which parts of the work are sharedNobody decided what is sharedWrite the split, and give the shared parts one owner
Teams build for themselves what operations already providesWhether a document names operations' services, its customers and how to askNo operations charterA short charter with services, decision rights and a sponsor

Fix, verify and prevent each cause

Requests reach the team through several routes

The guide to prioritising work and capacity puts one front door first: a single route for every request, whoever sends it, whether a form, a queue of tickets or a board everyone can see. When a request can come in through email, a chat message, a meeting or a word in the corridor, each to a different person, two people can accept the same job without either knowing, and as the guide warns, many front doors leave prioritisation to accident.

Fix: make request intake the only route, under the guide's rule that work which has not come through it does not exist. Triage on a fixed rhythm and compare each request with what is open; one that matches existing work goes to whoever is already doing it.

Verify: check that every request completed over a recent period has one intake record; anything done without one shows which side door is still open.

Prevent: answer requests that arrive another way by pointing to the form, every time, and close side doors as they appear.

Work lives in inboxes, chats and private spreadsheets

The work management guide draws the line plainly: work can be managed only when every item sits in one system with its owner, stage and date, and work held in inboxes, chat threads and scattered spreadsheets can only be remembered. A team tracking its work in its own spreadsheet is invisible to the team next door, so something half-built in one place is started from nothing in another.

Fix: move every item into one work system organised by type of work and stage, show the calendar as a filtered view of those items, and give every team read access so anyone can see what is planned and under way before starting something similar.

Verify: sample work finished last period from each team and confirm each item was in the system from the start.

Prevent: keep the guide's rule that a decision reached in chat is not complete until it is recorded on the item.

The approved asset is harder to find than an old copy

The asset library tutorial gives the library one job: the version that is current, approved and cleared for use should be the easiest to find. When an old campaign folder is quicker, people reuse what they find there or rebuild the asset, and the tutorial blames near-identical copies above all for the wrong file going live.

Fix: build an asset library with exactly one approved master of each asset, carrying approval status, version, rights and expiry, and archive superseded versions rather than deleting them. Make the old folders read-only, then archive them, and point templates, intake forms and procedures at the library.

Verify: ask someone from a team that did not make a common asset to find its current version; they should reach the library copy first and recognise it as approved.

Prevent: give the library an owner and an upkeep routine, so new approved assets arrive with their fields complete and expired ones leave.

Findings sit in one team's deck

The tutorial on building a research repository explains why a study gets repeated, and between teams the first reason it gives is the one that matters here: nobody outside the team that ran the study knew it existed, because its findings lived in that team's slides. A single study repeated because it changed nothing the first time belongs to the research section's diagnosis for findings that are not acted on.

Fix: keep findings, not decks: each finding stored as a separate record beside its evidence, tagged consistently, linked to the decisions it informed, with an owner and a review date. Keep names, recordings and contact details out, since they are personal data that belongs in the study's restricted store, and add the past studies teams still rely on.

Verify: the next research brief opens with what the repository already holds, linking the records found.

Prevent: make that search the first step of every brief, as the tutorial does.

Tools bought by request on a card

The tool requests guide describes stacks that grow one request at a time, a tool spotted at an event, a trial started, a card used, until the business learns of it at renewal. Two tools doing one job for different teams is among the results it names.

Fix: route every tool request through one front door and triage it against the stack record: the job is already covered, an owned tool can be extended, the tool goes to evaluation, or the request is declined. Treat a card sign-up as a trial by another name, logged and triaged afterwards, and ask finance to flag software on card statements to whoever triages. Each extra tool holding customer details is one more place where personal data is processed under the GDPR, which is why the guide holds any trial or card purchase that receives real customer data until its privacy review is done.

Verify: every tool in the stack record has an entry in the request log, and card statements show no software the log does not.

Prevent: publish how quickly requests are triaged and keep to it, since a request left waiting becomes a card payment.

Nobody decided what is shared

The comparison of centralised and decentralised operations lists what it costs to leave each team to run its own way of working: reports that cannot be combined, two tools bought for one job, and a lesson learned in one team that never reaches another. The guide to marketing team models finds the same cost in teams organised by product or segment, where each grows its own half-specialist and standards drift apart unless something central holds them.

Fix: list what each team has built for itself, rules, reports, templates and tools, and sort it with the comparison's hub-and-spoke split: the hub takes whatever is damaged by variation, such as data rules, naming, tools and governance standards, and the teams keep whatever is damaged by uniformity, such as their own messages, timing and choice of channels. Write the split down, because the comparison says it holds only in writing. Where the duplication shows up as one campaign under several names, the diagnosis for inconsistent campaign naming owns that part.

Verify: the next region that needs a report or tool the hub now owns asks the hub first.

Prevent: give each team a named operations contact, a route to propose changes to shared standards and a periodic review of the split.

No charter saying what operations provides

The operations charter tutorial calls operations the function most easily defined by accident: without a written boundary, two teams can claim the same work while nobody claims something important. Teams that do not know what operations provides build their own versions.

Fix: write a short charter with the sponsor's agreement, naming what operations owns, what sits outside its remit and whose it is, the internal customers it serves and the services each receives, the decisions it takes, recommends or is consulted on, and how work reaches it, through the same front door as other marketing requests. Write the decisions behind the traced duplicates as rows with one accountable role each.

Verify: a team lead asked which services operations provides, and how to request one, gives the answer the charter gives.

Prevent: give the charter a review date, and redirect requests outside its boundary rather than absorbing them.

Every duplicate on the list began the same way: someone started work with no means of seeing that the same work existed or was already under way. Prevention puts that check where work starts and keeps the places it looks worth searching.

The marketing operations systems hub separates two kinds of system that fail in different ways. Systems of work, such as the work tool, the calendar and the intake form, fail by being ignored, as work drifts back into chat and email. Systems of record, such as the asset library, fail by being polluted with entries nobody can trust. The first needs adoption and good design; the second needs rules and routines. Give each an owner from the RACI rows and judge them on those different terms.

Then ask one question at every front door: what did the requester find already? The intake form for marketing work, the research brief and the tool request form can each carry it, with a link to what was searched. A request that arrives with the answer filled in has checked the work system, the library, the repository or the stack record before anyone spends effort; one that arrives blank tells the triager where to look.

Keep a short note of each duplicate that still gets through, with the join that failed, and read the notes at the charter's review: a repeat in the same place shows which owner or routine needs attention, and one in a new place shows a join nobody has drawn yet. Where marketing runs across several countries, the playbook for multi-market marketing operations applies the same discipline to the decisions that recur in every market, each with one accountable role. The marketing operations troubleshooting hub lists the other symptoms this collection diagnoses.

A builders' merchant traces three duplicates to their source (illustrative)

Illustrative example — not a real result

Wendiquaxt Building Supplies

Wendiquaxt Building Supplies, an invented builders' merchant with a head office marketing team and 4 regional teams, found three duplicates in one quarter. Two regions had commissioned near-identical surveys of trade customers 6 weeks apart, 3 teams were paying for separate design tools, and a third region had rebuilt the head office trade brochure because it could not find the approved file.

The operations lead traced each one. Both surveys had been requested by email to whichever researcher the regional manager knew and tracked in each team's own spreadsheet, and the first survey's findings sat in a deck on one region's drive. Each design tool had been bought on a card, and nobody kept a list of tools. The approved brochure sat 4 folders deep on head office's drive beside 3 older versions. Nobody had written down what head office provided to the regions.

Over the next 2 quarters the business opened one request form for all marketing work, moved every team into one work system with a shared calendar view, set up an asset library and started a research repository with the findings of its last 5 studies. Tool requests now go through one triage against a new stack record, which kept 1 of the 3 design tools. A 2-page charter with a responsibility matrix makes head office accountable for the library, the repository, the stack record and request triage. In the following 2 quarters the trace found 1 further duplicate, a regional price list built from an old template, and the template moved into the library.

Common mistakes

  • Reorganising before tracing. A restructure moves the boundaries without repairing the joins between them.
  • Buying a tool to stop duplication. A system nobody is required to use becomes one more place to search.
  • Centralising everything. Messages, timing and channels chosen close to the market suffer when forced into one shape.
  • Cancelling the second tool first. Find out what it holds and who depends on it before closing it.
  • Leaving shared records without owners. A library, repository or tool list nobody maintains stops being searched, and the duplicates return.

Frequently asked questions

Isn't some duplication simply the price of letting teams move quickly?

Some variation is deliberate and worth keeping. Teams close to a market are better placed to decide their own messages, timing and channels, and two regions writing different campaigns for different markets is not duplication. The duplication worth removing sits in things that suffer when every team does them differently: approved assets, research findings, tools, data rules and reporting definitions. Write down which side each one falls on, so the teams keep their speed where it matters and stop paying twice where it does not.

Who should be responsible for spotting duplicates before they happen?

No single person can see everything, so give each shared record an owner and put a check at the point where work starts. Whoever triages marketing requests sees new work first; whoever keeps the list of tools sees requests for software; the owner of the research store and the owner of the asset library know what already exists in each. Give each of those checks one accountable role, write the roles into the operations charter, and ask every request to say what the requester already looked for.

We found two teams paying for tools that do the same job. Should we cancel one straight away?

Not before you know what each tool holds and who relies on it. Treat the pair as a consolidation rather than a cancellation: decide which tool serves the job, move the users, the work and any customer data across, and close the other at its contract date. The site's tutorials on reviewing licences and seats and on retiring a marketing tool set out that order step by step.

Terms used on this page

Frameworks

Sources

  1. Regulation (EU) 2016/679 (GDPR) — EUR-Lex

Related