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

Digital Marketing › Troubleshooting

New Marketing Channel Not Working

Why a new marketing channel fails its trial: a test too small to read, the wrong job, borrowed creative, an absent audience, missing tracking or the economics.
Troubleshootingintermediate8 min readUpdated

First, decide whether the test could have answered the question

Before judging the channel, judge the trial. Find the test plan and look for four things written down before launch: the budget, the duration, the single measure of success, and the decision that each possible result would trigger. Kaushik's measurement model makes the point that a measure without an agreed target cannot say whether something worked. If any of the four is missing, the trial was an exploration rather than a test, and its result is weaker evidence than it looks.

Then check whether the plan was followed. Budgets get trimmed mid-flight, start dates slip, creative arrives late, and a test designed for a quarter ends up running for a few weeks. A trial that ran at a fraction of its intended weight cannot return the verdict it was designed for. If the trial fails both checks, the honest conclusion is "not yet known", and the causes in the first half of the table below come first. For other programme-level symptoms, the marketing strategy troubleshooting hub lists the related pages.

Diagnosis table

Diagnosis

SymptomCheckLikely causeFix
Results swing week to week; too few conversions to compareSpend, duration and outcome volume against what the plan neededTest too small or short to readRe-run at a readable weight for an agreed period
Reach and engagement fine, last-click sales near zeroThe job the channel was given against the measure it was judged onJudged on last-click sales when its job is reach or considerationRe-judge on the measure that fits its job
Low engagement compared with what the channel's audience normally givesWhether ads were made for this channel or resized from anotherCreative and offer copied from another formatProduce native creative for the channel
Delivery stalls or costs climb as targeting is narrowed to the real buyerAudience estimates for the buyer, not for broad interestsThe audience is not there in the numbers assumedStop, or reposition the channel's job
Platform shows clicks but analytics shows no matching sessions or salesTagged links, source and medium values and conversion setupTracking never set up for the new channelTag every link; confirm the outcome is recorded
Readable test, clean tracking, cost per result well above the ceilingExpected and actual cost per result against break-evenThe economics cannot workStop the channel and record why

Fix, verify and prevent each cause

A test too small or too short to read

Every channel needs some volume of outcomes before a difference between "working" and "not working" shows through the noise, and many platforms also need time to settle delivery. A trial budget split across several audiences and creatives can end with no single cell large enough to read. The testing and scaling guide sets out how to size and time a fair test.

Fix: recalculate what the trial needed to produce a usable answer, concentrate the budget on one audience and one offer, and re-run for an agreed period.

Verify: at the end, the result is stable enough that a second reading a week later would say the same thing.

Prevent: approve no channel trial without a stated budget, duration and success measure, and do not cut a live test's budget without resetting its end date.

A channel judged on the wrong job

A video or social channel added to reach new buyers will look worthless if it is read on last-click sales, because its effect shows up later and elsewhere. The channel mix guide gives every channel one job, and the measure has to follow the job.

Fix: restate the job the channel was added for, then re-read the trial on a measure that fits it: reach among the target audience, branded search, new-visitor share, or assisted conversions rather than last-click Attribution.

Verify: the measure that fits the job moved in the direction intended; if it did not, the channel has failed at its real job and the verdict is clean.

Prevent: write the job and its KPI into the channel's brief, and report each channel against its own job rather than a single blended sales number.

Creative and offer borrowed from another channel

Resizing a search ad's message into a vertical video, or posting a polished brand advert into a feed built for casual creators, tells the audience at once that this is an advert made for somewhere else. Each format has its own expectations, which the social content formats guide catalogues.

Fix: brief creative for the channel's native format and viewing habits, and adapt the offer to the stage of the buyer the channel reaches.

Verify: engagement and onward clicks rise to levels typical for the channel, before any judgement on sales.

Prevent: budget for channel-specific creative as part of every channel trial, not as a later improvement.

The audience is not on the channel in the numbers assumed

Sometimes the trial is sound and the channel is simply the wrong place. The buyers exist, but not on this platform, or not in a state of mind where they will act. The tell is that delivery is easy to broad interests and becomes thin and expensive as targeting narrows to the real buyer. The choosing channels guide explains how to locate attention and existing demand before picking a channel.

Fix: compare the platform's audience estimates for your actual buyer with the size the plan assumed; if the gap is large, stop the trial or give the channel a smaller job.

Verify: the decision holds when you check where recent customers say they spend time and found you.

Prevent: confirm audience size for the precise buyer, not for broad interests, before a channel enters the plan.

Tracking never set up for the new channel

A new channel is often launched by someone outside the usual campaign process, and its links go out untagged or tagged inconsistently. Its visits then land in direct or referral traffic, and its sales are credited elsewhere. Google's documentation on URL builders explains how campaign parameters added to links carry source, medium and campaign into analytics, and the naming conventions and UTM guide sets the convention.

Fix: tag every link the channel uses, confirm that its conversions are recorded, and re-read the trial from the date tracking became complete.

Verify: sessions in analytics broadly match clicks reported by the platform, and the channel appears under its own source.

Prevent: make a tracking check part of the launch checklist for any channel, with a named owner.

Economics that can never clear break-even

If the test was readable, the job was right, the creative was native and the tracking was sound, and the cost per acquisition (CPA) still sits well above what a customer is worth, the channel does not work for this business at this time. The performance economics guide derives the ceiling from margin rather than from platform suggestions.

Fix: stop the channel and write down the conditions under which it would be worth retesting, such as a higher-margin product or a lower price of attention.

Verify: the cost per result from the test, set against the ceiling, would not close with any plausible improvement in conversion.

Prevent: check expected cost per result against the ceiling before approving any trial, so channels that cannot pay are never started.

A prevention routine for channel trials

Most failed trials can be traced to decisions made before launch, which is where a prevention routine belongs. Use a one-page trial plan for every new channel: the job it is meant to do, the buyer it should reach, the expected cost per result checked against the ceiling, the budget and duration needed to read the result, the single success measure, and what will happen for each possible outcome. Require a tracking check and native creative before the start date. During the trial, change nothing the plan did not foresee; if budget has to move, extend the end date rather than reading an underpowered result. At the review, ask first whether the trial ran as designed, and only then whether the channel worked. Record every verdict, including the reason and what would change it, in a register the whole team can read, so that channels are not retried on enthusiasm or abandoned for ever on a single poor test. Finally, keep a small reserve in the annual budget for trials, so new channels are tested on purpose rather than squeezed in from leftover money.

A channel that passed its trial and later stops earning its place is a different case, and the guide to stopping an established paid advertising channel covers its evidence, the parts of the plan that depend on it and the wind-down.

Illustrative example — not a real result

A pet food brand tries short video

A pet food subscription brand gave a short-video platform £4,000 over four weeks to "see if it works", spread across five audiences and resized television footage. The review found 11 attributed sales and called the channel a failure. A second look showed that no links had been tagged, so most of its visits had landed in direct traffic, and that the job had never been stated. The team re-ran the trial for eight weeks at £6,000, with one audience, creator-style videos, tagged links and the job set as reaching new dog owners. Reach in the target group and branded search both rose, new-customer sign-ups from direct and branded search increased by 140 over the period, and the channel was kept as a creation channel with its own measures.

Common mistakes

Calling a trial a failure before checking it ran as designed. Spreading a small budget across many audiences and creatives. Reading a reach channel on last-click sales. Launching without tagged links. Keeping a failed channel at a token budget instead of stopping it.

Frequently asked questions

How long should a new channel get before we decide?

As long as the test plan said, and the plan should have said so before launch. The right period depends on the channel's job and on how much volume it needs to produce a readable answer: a capture channel aimed at people already searching can be read sooner than a reach channel whose effect arrives through other channels later. If no period was agreed, agree one now and treat the time so far as set-up.

Is it worth running a new channel at a small budget just to keep a presence?

Rarely. A channel run below the level needed to learn anything costs money and produces no decision. Either fund it properly as a test with a review date, or stop it and record why, so the question can be reopened later with a better design.

The new channel's own dashboard says it is working but our sales did not move. Which do we believe?

Neither on its own. Platforms report under their own attribution rules and tend to credit themselves generously, while total sales may not move for reasons unrelated to the channel. Check the business's own data for the audiences or regions the channel reached, and if the question matters, design a holdout so you can compare outcomes with and without it.

Terms used on this page

Where this page is used

Also related

Sources

  1. Digital Marketing and Measurement Model — Avinash Kaushik
  2. [GA4] URL builders: Collect campaign data with custom URLs — Google Analytics Help

Related