Case study · Malaysian EV market

Could we have seen it coming?

Can social media predict which EVs will sell in Malaysia? What held up, what didn't, and why it matters if you forecast promotions and launches.

23,383
social media comments
15
EV models tracked
28
dated market events
10 months
locked forecast test

EVs went from 1 in 100 new cars to about 1 in 10

Battery EVs as a share of all new cars, by quarter.

BYDTeslaProtonOther brands
Source: DOSM car registrations (data.gov.my). Latest bar is Jul–Aug 2026 only.
01 · Test against a simple rule

A complex model did worse than a simple rule of thumb

A machine-learning model built on comments, against one rule: next month = this month. The rule was three times more accurate.

Machine-learning model83% off
Same as last month28% off
Average miss as a share of actual sales. Lower is better.

In your business: check every forecast beats last month or last year.

02 · Fix the messy data

The second version fixed the messy data

Two Chery models share a name across petrol and electric. The first version counted both: 43% of its “EV sales” were petrol.

Chery Omoda 5
15,693 counted
1,141 electric
Chery Jaecoo J7
16,526 counted
601 plug-in

In your business: shipments or sell-out? Gross or net of returns? Choose the target on purpose.

03 · Why did sales move?

Events explain the big moves. Comments don't predict them.

28 dated events, each checked against normal swings. Pick a model, then an event.

BYD Atto 3 · registrations per month
Unusually large moveNormal swing, or a launch
Price is the strongest lever
Deep cuts (17–33%) lifted sales 3–4×. Small cuts did nothing.
Comments move with sales, not ahead
Comments rise in the same month as sales, never before.
Sentiment scores mislead
The cuts that worked drew angry comments. People complain and buy anyway.
04 · The locked test

A forecast frozen at October 2025, scored against the next 10 months

A statistical baseline plus human adjustments. “Locked” means neither could change once real figures arrived.

  1. STEP 1
    Only what was known then
    Data to Oct 2025, news before 1 Nov 2025.
  2. STEP 2
    Rules set before the result
    Method chosen by a fixed rule. Every adjustment has a source.
  3. STEP 3
    Freeze, then score
    Fingerprinted, then scored. No re-running.

Run in Sep 2026 as if it were Oct 2025.

ActualBaselineBaseline + adjustments
Battery EV registrations per month, all brands.
1

The tax-deadline rush: caught

Two adjustments for the Budget 2026 deadline cut error from 35% to 15%.

Assumption held
2

The Proton e.MAS 5: missed

Assumed about 600 a month. It sold 3,068 in January. The log shows which assumption broke.

Assumption broke

The whole scorecard, all 10 months

Average miss as a share of actual sales. Lower is better.

15 models, combined
Same as last month
56%
Baseline
56%
With adjustments
52%
EV market total
Same as last month
31%
Baseline
22%
With adjustments
23%

Adjustments helped models a little and the market total not at all. They worked where the evidence was specific.

Swap “EV” for your category

Every driver here has a twin in FMCG and retail.

In the EV studyIn your business
Price cutsPromotions and price-offs
Model launchesNew product launches
Competitor launchesCompetitor activity
Tax-deadline rushLoading before a price or excise increase
Comment reactionRetailer and consumer feedback

Curious how this would work on your numbers?

The same approach, for FMCG and retail teams.

Data: DOSM vehicle registrations via data.gov.my; public Facebook and YouTube comments, used in aggregate only.