Study · CPG · Three instruments, registered before the data

Most line extensions are too small to lift their parent.

The case for a line extension often counts a halo: the new product sends buyers back to the brand they already know. We screened 45 extensions across the same aisle, a new aisle and licensed partners. 4 drew enough attention at launch for a lift to be measurable at all, and none of the ones tested cleared the bar.

Read this first. Two of the three search passes were voided by flaws in their own design, named below. “No 20% lift” is not “no lift”: a smaller effect could sit inside normal monthly swings, and search is curiosity, not sales.

Ipalibo Da-Wariboko · CULT+MATH · November 11, 2026 · 45 extensions · 3 instruments · public data only, nothing commissioned

4 of 45 line extensions were big enough to test. Licensed: 0 of 11.

What the evidence does support comes from the other two instruments. AI assistants name the extension when asked about the parent, and describe owned extensions from the owner’s own site. Licensed extensions are described by everyone except their owner. In reviews, licensed products mostly replace the restaurant visit rather than add to it.

gate: extension searches in the launch month ≥ 20% of the parent’s · lift bar: +20% · 3 tested units, 0 lifted

The decision this is for

Whether to approve a line extension partly on the promise that it lifts the parent. The buyer is a business-unit director deciding where the next extension sits: the same aisle, a new aisle, or a licensed partner.

If extensions lift the parent, part of the parent's revenue belongs in the extension's business case. If they don't, each extension has to pay on its own. The answer here is the second, for every extension we could test.

Most extensions can’t move the parent

An extension can only lift its parent's searches if people search for it. We set a gate before looking: in its launch month or the month after, the extension had to draw at least 20% of the parent's average monthly searches.

Extensions screened and measurable, by where they sit, across three passes
ArmScreenedMeasurable
Same aisle191
New aisle153
Licensed110
Total454
The four that passed the gate, with the launch date checked against a public source
ExtensionArmGate ratioLaunch monthDate check
Heinz Pickle KetchupSame aisle1.14Nov 2023announced November 6, 2023
Oreo CakestersNew aisle3.33Jan 2022back at retail January 7, 2022
Liquid Death iced teaNew aisle0.34Mar 2023launched March 1, 2023
Oatly ice creamNew aisle0.32Jun 2019NYC retail by the end of June 2019

Even the candidates picked because they looked big mostly failed: Skittles Gummies drew 1.7% of Skittles' searches, Bud Light Seltzer 2.8% of Bud Light's. No licensed extension passed; the closest was Chips Ahoy cereal, made by Post, which reached the gate but in a month four months off its dated relaunch.

The ones we could test showed no clean lift

Parent search interest in the launch month against the three months before, Google Trends, two pulls
ParentExtensionMonthChangeSame month, other yearsReading
OreoOreo CakestersJan 2022-11% / -8%Every January falls 4% to 34%No lift
Liquid DeathLiquid Death iced teaMar 2023+51% / +51%Median of other Marches +13%; March 2024 +49% with no launchMatched by a launch-free year
HeinzHeinz Pickle KetchupNov 2023+18% / +16%No other November above +8%Under the bar

Oreo fell in the Cakesters month, as it does every January after the holidays. Liquid Death rose 51% when its iced tea launched, well above its typical March (+13%), but March 2024 rose 49% with no launch at all, so this test can't call it lift. Heinz is the one launch-shaped move: the only November in seven years above 8%, in the month Pickle Ketchup was announced, and still under the bar.

Assistants carry the extension forward

Asked about the parent's range, does ChatGPT name the extension, and is the owner's own site a source?
ExtensionArmNamed (runs)Owner's site cited
Coca-Cola SpicedSame aisle2 of 2Yes
Pepsi PrebioticSame aisle1 of 2Yes
Old El Paso Fiesta TwistsNew aisle2 of 2Yes
Ghost EnergyNew aisle2 of 2Yes
Chick-fil-A saucesLicensed2 of 2No
Taco Bell Cravings KitsLicensed2 of 2No
IHOP syrupsLicensed2 of 2No

6 of 7 extensions were named in both runs. The four owned ones were described from the owner's site in all 8 answers. The 3 licensed ones cited the restaurant or the licensee in 0 of 6, drawing on retailers, scraped pages and an unofficial menu site instead. For Ghost, a lookalike domain was the first source cited in one run. Both Coca-Cola answers listed Spiced as a recent launch; it has been discontinued.

That is the opposite of the TV result, where assistants left the newest show out (the spin-off study). A product question is answered from the brand's product pages, and those list the range.

Licensed products replace the visit

Reviews coded for comparing to the parent, and for changing what the reviewer thinks of or does with it
ArmReviewsCompares to parentChanges behavior toward parent
Same aisle1010 of 101 of 10
New aisle104 of 102 of 10
Licensed109 of 104 of 10

In the licensed arm, three of the four reviews that changed behavior were about where the reviewer eats next: two said the kit replaces a Taco Bell visit. Same-aisle reviews compared every product to the parent and changed nothing but which version they buy.

What we got wrong

The first search pass used a bar the data couldn't see. Google reports monthly volume in steps of about 22%, so a 20% lift can read as zero. The second pass moved to Google Trends, which has no steps, but compared each launch month with the three months before: that reads a seasonal peak as lift. Its control, IHOP's Pancake Day, never cleared 20% in seven years, so it could only void the pass. The third pass fixed both and found too few measurable units to score.

Registered predictions that missed, and the one that held
PredictionRegisteredResult
Assistants name the extension: same aisle55%Miss (named for 2 of 2 units)
Assistants name the extension: new aisle40%Miss (2 of 2)
Assistants name the extension: licensed60%Miss (3 of 3)
Reviews: licensed transfer under 15%LikelyMiss (4 of 10)
V2.1 control (IHOP Pancake Day) valid50%Miss (0 of 7 years)
V2.2 same aisle has 2 measurable units60%Miss (0)
Licensed draws the least attention80%Hold

What this cannot tell you

Search is not sales. A lift in purchases with no lift in searches is not ruled out.

Ten reviews per arm, partly incentivized, coded from search results. One assistant, ChatGPT's budget tier, two runs.

The tested units are the big ones. The verdict covers extensions large enough to test, which is the case most favorable to a halo.

How it was built

Three instruments, each registered before its data: search demand, AI assistant answers, and customer reviews. An arm gets a verdict only when two of three agree. Launch months were read from each extension's own searches and checked against a dated public source. Nothing was commissioned and no company was contacted. Runs 064618b5 · 58abbd30 · c49293d2 · df4df5c5 · f46e2a22 · 98c7b037 · 3c8d5bf4.

Whether to extend is one of six decisions

This study tests one assumption in an extension's business case. The CPG Decision Brief takes the six decisions a brand manager owns and answers each from filings, with the limits of the evidence stated.

See the six decisions

$495 standard, $995 custom. Pre-made and delivered on payment. If the decision turns out not to be decidable from evidence, it is refunded.

CULT+MATH LLC · Houston, Texas · Ipalibo Da-Wariboko
Standing research. Figures are re-measured on the schedule stated on each page and the page is corrected in place, dated. Corrections are published, not applied quietly.
Cite as: CULT+MATH LLC, Ipalibo Da-Wariboko. Attribution to cultmath.com is requested.