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

Digital Marketing Foundations › Intermediate

Drafting a Channel Mix for a Language-Learning App

An exercise in turning a single-channel dependency into a deliberate mix. Working from an invented language-learning app's data, you give each channel one job, a journey stage, a success measure and a share of effort, and explain how search and content reduce the app's reliance on paid social.
ExerciseIntermediate levelModule 2: Designing the channel mix5 minUpdated

Why this matters

A business that relies on one paid channel has handed its growth to someone else's auction: when costs rise, nothing underneath absorbs the shock. This exercise applies the method in the channel mix guide to a lopsided starting point and a small team.

The app and its current mix

Everything about Lingopath is invented for the exercise.

Illustrative example — not a real result

Lingopath: last quarter's position

Lingopath is a subscription app teaching conversational German, Spanish and Dutch, sold in the UK. It has a free trial and a monthly plan. Research last quarter found its best subscribers are adults preparing to move abroad for work, who need everyday language for their first months in a new country. Its position: "the app that gets you through your first month abroad". The team is three people: a marketer, a content writer and a lifecycle specialist shared with product.

ChannelLast quarter's spendShare of spendWhat it was doing
Paid social£51,00085%Everything: prospecting, retargeting and trial offers in one campaign
App-store search ads£6,00010%Bidding on the brand name only
Email and in-app messages£3,0005%A single welcome email
Blog£0 media—Forty posts on grammar topics, little traffic

Signals from the data:

  • Cost to acquire a paying subscriber through paid social rose from £14 to £23 over three quarters.
  • Trial-to-paid rate is 18%; most trials that lapse never open the app after day two.
  • Search Console shows the site appears for queries such as "German phrases for a job interview" and "how to open a bank account in the Netherlands" but ranks too low to be clicked.
  • Keyword research shows steady demand for relocation-language queries in all three languages; competitors' pages for them are thin.
  • Finance has capped next quarter's media at £60,000 and asked for a lower cost per paying subscriber.

Giving each channel a job

Lingopath's problem is not paid social itself but one campaign doing every job against one number, so nobody can tell which job is failing. Write a job sentence per channel: stage, segment and success measure. Use the stages of the RACE framework (reach, act, convert, engage) to check coverage, because Lingopath has almost nothing in act or engage.

Then decide the balance between search channels. The SEO versus PPC comparison is the reference: paid search (SEM) captures existing demand within days and stops when spending stops; organic search (SEO) takes months and compounds. The question is which queries each covers, and when. A content programme costs writer time even when its media cost is nil.

Task

  1. Write a job sentence for every channel you keep or add, naming the RACE stage, the segment and the success measure.
  2. Build a channel mix table with the columns channel, role, stage, success metric and share of effort.
  3. Split paid social into separate campaigns where it currently does more than one job.
  4. Decide which search queries paid search covers now and which organic content should take over, and when you would test pausing paid on them.
  5. Write a short rationale explaining how search and content reduce Lingopath's dependence on paid social and what you expect to happen to the cost per paying subscriber (CAC) over the next few quarters.
  6. Note one sequencing rule and one risk.

Worked answer

Illustrative example — not a real result

Lingopath: next quarter's channel mix

ChannelRoleRACE stageSuccess metricShare of effort
Paid social, prospectingCreate demand among people planning a moveReachCost per trial; brand search lift45% of media
Paid social, retargetingBring back site visitors who did not start a trialActTrials from retargeted visitors8% of media
Non-brand paid search and app-store search adsCapture relocation-language searchesConvertCost per paying subscriber at or below £1825% of media
Brand app-store search adsDefend the brand nameConvertCost per paying subscriber5% of media
Organic content: relocation guidesRank for the same queries; offer a trialReach and actOrganic trialsHalf the writer's time; no media
Email and in-app lifecycleTurn trials into subscribers; reduce early lapseConvert and engageTrial-to-paid rate up from 18%7% of media; the lifecycle specialist
Relocation partnersReach people at the decision to moveReachReferred trials5% of media
Testing reserveNew creative and query testsAllDecision per test5% of media

Paid social falls from 85% to 53% of media; total media stays at £60,000.

Search and content. Paid search goes live first on relocation queries: it captures existing demand and shows which topics convert. The writer turns the best-converting queries into guides (job interviews, renting, banking) instead of grammar posts. As each guide ranks, a pause test on the matching paid queries frees money for queries the site does not yet rank for. The distribution channels guide adds the other half: each guide also goes to the email list and to partners, so it earns visits before search picks it up.

Expected effect. The blended cost per paying subscriber should fall as organic trials grow at no media cost and lifecycle work lifts trial-to-paid. Paid social is now judged on the demand it creates, not on last-click sales.

Sequencing rule and risk. Capture channels go live before the prospecting campaign changes, so the demand it creates is caught. The risk is the writer's capacity: if guides slip, organic never takes over and paid search cost rises with volume.

Common mistakes

  • Moving budget from one paid channel to another and calling it diversification. Dependence on any single auction is the same risk.
  • Giving content a share of media but no time. Content is mostly effort, so record it as effort.
  • Judging prospecting on last-click sales. It will look worse than capture channels on that measure by design.
  • Leaving engage empty. The early lapse after a trial starts is the biggest leak in Lingopath's data.

When the table is done, test yourself with the channel mix checkpoint, or return to the intermediate level overview.

Pages this lesson applies

Terms used on this page

Also related

Related