MKT5610 Chap.13 Marketing Intelligence from Platform Data
Marketing Intelligence from Platform Data
The marks are in the design, not in the tool
One of the graded activities asks you to assemble a body of public platform material: posts from a set of brand accounts, and product listings with their reviews from an online marketplace. The brief states that no coding is required because the collection tools are covered in class, and that the document lists requirements and deliverables rather than teaching a method.
That division tells you where the judgement sits: in what you choose to collect and whether the result is fit to be compared.
Matched comparators are the whole design
The brief asks for a focal set and, for each focal brand, competitors chosen by rule: same category, and officially verified accounts rather than imitation or fan pages.
On the marketplace side it asks for one comparable competing product per assigned product, same category, similar price, favouring listings with more sales and reviews, never the same item twice.
A difference between two accounts differing in category, size and authenticity at once cannot be attributed to anything.
Completeness and schema fidelity are requirements rather than ambitions
Platforms return recent material first, so a collection that stops early is a sample of the recent past rather than a smaller sample of the same thing.
And files must match the supplied templates field by field, because work sharing a schema can be pooled and work that does not is stranded.
What this chapter covers
- 01
Focal brands and self-selected comparators, chosen by rule
- 02
Verified accounts as a construct validity requirement
- 03
Collecting to the last page, and why stopping early biases the sample
- 04
Comparable product pairs at similar price, never duplicated
- 05
Recording failures rather than dropping them silently
- 06
Schema fidelity, and what makes a collection poolable
- 07
Verification checks that all fail silently
- 08
Data handling conduct and credentials
Audit a comparator set against the stated rules
- 3Judge each of the three against the stated rules.
- 3State the analytical cost of each breach.
- 2Say what the repair is, and what to do if it is unavailable.
Key terms
- Focal Brand
- A brand assigned to you as the subject of a collection, around which comparators are selected by rule rather than by preference.
- Matched Comparator
- A competing account or listing selected to differ from the focal item on as little as possible other than the thing being compared. It is what makes any later difference attributable.
- Verified Entity
- An account confirmed as belonging to the organisation it claims to represent. The requirement protects construct validity, because an unverified page reflects enthusiast behaviour rather than marketing decisions.
- Exhaustive Collection
- Continuing until the last available page rather than stopping at a convenient depth. Platforms return recent material first, so partial collections are biased towards the recent past.
- Error Record
- A file naming the items that could not be collected and why. It is what distinguishes a gap in coverage from a silent omission, and it costs nothing because the tool produces it.
- Schema Fidelity
- Matching a supplied template field by field, in column names, order and content structure. It is what allows separately collected work to be pooled into something larger than anyone assembled alone.
- Encoding Failure
- Damage to text that leaves a file which opens and looks plausible while being unusable for analysis. It is one of several checks in this task that fail silently.
Marketing Intelligence from Platform Data FAQ
Why do the comparator rules matter so much if the tool does the collecting?
Because the tool cannot repair a badly chosen set. If a comparator differs from the focal brand in category, in size and in authenticity at once, then any difference the analysis later finds is confounded with all three and can be attributed to none of them. The rules exist to hold everything constant except the thing being compared, and that is the only part of the task where your judgement changes the result.
What happens if some accounts cannot be collected?
Submit what you have together with the error record naming each failure and its reason. A collection with items missing and no explanation is indistinguishable from one where items were skipped because they were inconvenient, and anyone pooling your work would read the missing brands as having no posts rather than as not collected, which is a different and false fact. The gap is not the problem; silence about it is.
How strict is the requirement to match the sample file format?
Strict, and for a reason that becomes obvious once many submissions arrive. The brief asks for the same column names, the same order and the same content structure as the supplied examples. Work sharing a schema can be combined into a dataset far larger than any individual collected; work that does not is stranded. A file that is correct and differently shaped has produced private value only.
What are the rules about what I may do with the data afterwards?
The brief states that the material may be used for coursework and academic study and for nothing else, with no commercial use, nothing passed outside and nothing published. It also states that login sessions are credentials which must not be shared or included in a submission, which is a practical hazard as much as a rule, because credentials can end up inside an exported file without anyone intending it.
Add a check for that to your verification list.
Assessment move
Before you collect anything, write your comparator list with one sentence beside each entry saying which rule it satisfies. Entries you cannot justify in a sentence are the ones that will be wrong. Then, after collecting, open the combined file and read twenty rows rather than checking that the row count looks right.
Working through Marketing Intelligence from Platform Data in MKT5610? Sia is AskSia’s AI Marketing tutor — ask any MKT5610 Marketing Intelligence from Platform Data question and get a clear, step-by-step explanation grounded in how MKT5610 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.
