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

Social Media Marketing › Expert

Checkpoint: Evidence for What Social Channels Add

Scenario questions set at an invented lunchware maker test the first expert social media module: telling credit from cause, keeping view-through apart, choosing a geo holdout or a platform lift study and using one to read the other, treating branded search and customer answers as signals, reporting an organic video channel as a range and matching the evidence to the budget and stage. Each answer gives the reasoning and the page behind it.
CheckpointExpert levelModule 1: Proving what social channels add4 minUpdated

What this checkpoint covers

Simuxa Lunchware is an invented maker of reusable lunch boxes and flasks. It sells on its own website and through kitchen shops and a department store group that share weekly sales by region, runs TikTok ads and keeps an organic YouTube channel of packed-lunch ideas. The questions apply the guide to measuring what TikTok adds beyond last click, the guide to measuring YouTube's contribution to the business and the comparison of geo holdouts and conversion lift studies.

Illustrative example — not a real result

Simuxa Lunchware: the figures the questions use

MeasureFigure
TikTok spend on lunch boxes£12,000 a month
Average web order£140
TikTok orders on last click in web analytics90 a month
TikTok's reported conversions on the account's chosen windows610 a month, 420 of them view-through
Share of units sold through shops45%
Spring lift study at £12,000 a month: incremental web orders TikTok reported160 a month
Autumn geo holdout at the same spend: incremental orders, web and shops290 a month

September, the back-to-school weeks, carries Simuxa's year.

Task

  • Answer each question with its reason before reading the explanations; a right choice for the wrong reason is a miss.
  • Reread the linked page for every miss.

Questions

  1. The finance director's last-click report shows TikTok barely paying for itself; the account manager's TikTok column shows it paying several times over. Which should she believe?
  2. The monthly report's headline is "TikTok conversions" from TikTok's reported total, with social conversions from web analytics added into a "social total". What changes?
  3. Finance asks whether the TikTok budget pays, given how much sells through shops. Geo holdout or TikTok lift study?
  4. With both test results in hand, the analyst proposes dropping geo tests for a lift study every quarter. Is that sound?
  5. During a September TikTok push, branded searches rose sharply and more checkout answers named TikTok. The paid social manager calls it proof. What is your reading?
  6. The owner wants one revenue figure for the YouTube channel for the board. What do you give instead?
  7. A new flask range has just started on modest TikTok spend, and the agency proposes a mix model to measure it. What fits this stage?

Answers explained

  1. Neither. Last click never sees an ad watched without a tap, and TikTok's column credits buyers who would have bought anyway: both describe Attribution credit, not cause. Her question is what the business would lose without TikTok, the Incrementality question, which a test with a Holdout Group answers first, and a mix model calibrated with one can answer later. See why last click and TikTok's own column disagree.
  2. Split the lines and stop adding. Report last click, click-through and View-Through Conversion credit separately: view-through is unproven rather than wrong, and the part most likely to be counted elsewhere. Write the attribution setting and its date into the measurement plan and put the test that checks view-through beside its line. See keeping view-through credit on its own line.
  3. A geo holdout read on Simuxa's own regional sales. A lift study counts only what TikTok observes, so with so much sold in shops it would understate the effect by design; regional web and shop sales capture it, and independence suits a contested budget. It needs targetable regions, retailer sales by region and enough comparable regions. See when to choose a geo holdout test.
  4. Partly. Run in different periods at similar spend, the two results show how much of TikTok's effect the platform can see. Use that ratio to read routine lift studies, date both results, and repeat the geo holdout when something material changes, such as a large budget shift or a new retail listing. See running one design to check the other.
  5. Signals to explain, not proof. The push ran in Simuxa's strongest season, branded search also moves with press and other channels, and answers show where attention came from, not what TikTok caused. Read branded search against its baseline and earlier years, and let the answers choose what to test, not what to spend. See branded search and how-did-you-hear answers.
  6. A range with its limits. Tagged last-click orders are the lower end, assisted journeys widen it, and the upper end adds customers who named YouTube but arrived untagged, with branded search as support. Name the gaps, including shop sales the site cannot see and no holdout, which a findable library rarely allows, and end with a decision. See reporting the channel to the business.
  7. Cheap, always-on evidence first. Keep view-through on its own line, take a branded search baseline before the push and read flask buyers' how-did-you-hear answers. A mix model needs history and varied spend; it comes third, calibrated with a test. See choosing the evidence by budget and stage.

Where to go next

If the first three questions caught you out, rework the exercise on evidence for a fishing tackle retailer; the creator measurement it builds on sits at the advanced social media level. Then start the next module with the exercise on a hijacked account and a wrong ad price.

Pages this lesson applies

Terms used on this page

Also related

Related