GolOps
Back: SIBUR
Other studies in this categoryIndia · Consumer goods manufacturerRussia · Packaging manufacturer
Decision journey study

How Russian polypropylene appears in the algorithmic choice of Turkish converters

Packaging manufacturer · feedstock purchase for the production cycle

19/ 100

19 out of 100 — this is how much the object influences the choice in this scenario0 — the buyer never sees you, 100 — you are recommended first

Study index · object's CCI on this marketChoice Control Index of the object on this market, 0–100: how far it gets through the stages of choice and how often it wins. Measured on this market with this buyer type — on another market the same object scores differently.
-5 vs Q4 2025Turkey · Q1 2026 · AI systems: 7

What to change to be chosen

We found three possible directions for growth. We estimated how each of them could affect the buyer's choice and potential revenue.

  • Logo of SIBURSIBURTurkey
    H18 mo

    Enter application-based choice.

    Shortlist1222AI answers out of 100 that name SIBUR
    +1 a yearnew annual contracts with convertersScenario estimate
  • Logo of SIBURSIBURTurkey
    H28 mo

    Pass evaluation with Turkish converters.

    Shortlist1221AI answers out of 100 that name SIBUR
    +1 a yearnew annual contracts with convertersScenario estimate
  • Logo of SIBURSIBURTurkey
    H38 mo

    Remove the settlement barrier before the shortlist.

    Shortlist1219AI answers out of 100 that name SIBUR
    +1 a yearnew annual contracts with convertersScenario estimate
  • Nizhnekamskneftekhim
    Not assessed

    The market is measured, but there is no report or growth hypotheses for Nizhnekamskneftekhim. We build them to order.

What task is the buyer solving?

Pick a choice scenario — a specific task of the buyer.

See which options AI systems consider, whom they recommend and how the participants' positions change across the stages.

Language shapes the choice

We study the same task in Turkish and English to see how language affects AI-system recommendations and companies’ positions.

MeasurementA measurement is one round of the study: over a few days we put the buyer’s questions to the AI systems and recorded every answer. It is repeated every quarter. The latest measurement is shown by default. Pick an earlier quarter to see how positions have changed. The number next to it is how many measurements there are.
AI systems: 7
AI systemThe measurement runs in every AI system at once. The combined result is shown by default. Each system shapes the choice differently: it uses its own sources, models and market context. So a company’s position can differ at different stages of the choice journey — from discovery to recommendation. Pick a system to see where exactly the company wins or loses its position.
Choice journeyThe choice journey is how the buyer starts the conversation with AI. One describes their problem, another asks straight about the product category, a third asks about specific suppliers, a fourth starts with constraints — budget, timing, requirements — and a fifth asks for a ready recommendation. The start of the conversation shapes whom AI suggests. Pick a journey to see where the company is stronger.
LanguageEach scenario is put to the AI systems in the local language and in English. The combined result is shown by default. In different languages AI relies on different sources, so a company’s position can differ. Pick a language to see where the company is stronger.

Why does SIBUR appear in 72% of answers but get only 4% of recommendations?

We traced how SIBUR passes the five stages of choice in AI-system answers. See where its position weakens most and how competitors perform.

Measured

Measurement conditions: S1 «Choosing a polypropylene grade for packaging production» · journey «Problem-first» · Turkey · languages TR, EN · AI systems: 7 · Q1 2026

01DiscoveryDoes the object appear in the answer space at all?
SIBURRU
72%
Borouge
88%
LyondellBasell
91%
SABIC
87%
02ConsiderationDoes the AI present it as a real alternative worth considering?
SIBURRU
59%
Borouge
81%
LyondellBasell
85%
SABIC
80%
03EvaluationDoes it enter a substantive comparison against the choice criteria?
SIBURRU
33%
Borouge
70%
LyondellBasell
75%
SABIC
69%
04ShortlistDoes it enter the final set of candidates?
SIBURRU
12%
Borouge
55%
LyondellBasell
60%
SABIC
54%
05RecommendationDoes the AI recommend it outright as a preferred option?
SIBURRU
4%
Borouge
37%
LyondellBasell
42%
SABIC
36%
Where the choice is lostSIBUR
−13
−26largest loss
−21
−8

Height of a column is the share of answers where the object passes the stage. The narrowing between columns is the loss.

Participation index55first three stages
Win index8shortlist and recommendation

A stage is analytical markup of a step, not a required order of conversation: a journey may start with a recommendation and come back to comparison.

Key finding

SIBUR is recognised as a large supplier with a short haul, but when a grade is chosen for a specific application the AI systems rely on Western producers' technical documentation.

Point of maximum loss

−26pp

The largest loss occurs between the stages «Consideration» and «Evaluation»

SIBUR · journey «Problem-first»

Why the position is lost

Choice factors that keep the object out of the final recommendation.

The grade portfolio is not described in machine-readable form
63%
No grade datasheets in English or Turkish
55%
Restrictions on settlements and currency operations
47%
Weak on-site technical support for the converter
41%
The sanctions context in the models' reasoning
34%

Position depends on the journey and on the scenario

Probability of a final recommendation for SIBUR for each decision journey, in both scenarios.

S1S2

The scenarios are different economic tasks. Their levels may be set side by side but not subtracted: the gap between scenarios is not a deficit to be closed.

Who takes the place

In AI recommendations (60–100)
Medium (30–59)
Low (10–29)
Critical (0–9)
Netherlands · USA#1 · leader
A technology licensor with a grade portfolio.
LyondellBasell
51
UAE#2 · −5 pts
Gulf polyolefins, a short haul in the Mediterranean.
Borouge
46
Saudi Arabia#3 · −6 pts
A broad grade range and technical support.
SABIC
45
RussiaRU#4 · −43 pts
SIBUR
8

Tile area and colour show the win index on this market — shortlist and recommendation. The participation index is in the table view.

For companies · from analysis to results

What to change to be chosen more often?

Logo of SIBURRecommended for SIBUR · Turkey

We identified what keeps the company from winning against competitors and proposed concrete actions to strengthen its position in the market.

Choose the solutions you want to implement. GolOps will help organise the work and check the result.

  1. 01

    Show buyers what each grade is best for

    Publish technical data sheets for every grade so converters can see which products each one suits.

    Forecast+10pts to the index
  2. 02

    Speak to Turkish buyers in their language

    Prepare materials in Turkish and set up on-site technical support for converters in Turkey.

    Forecast+9pts to the index
  3. 03

    Remove buyers' doubts about payment terms

    Explain openly how settlements work and which currencies Turkish buyers can use to pay for deliveries.

    Forecast+7pts to the index

Every solution you add raises the plan’s potential

Your planHand the approved hypotheses to the Control CentreThe Control Centre turns decisions into tasks and does most of the work itself: it draws on your team's context, prepares materials, ships publications, and coordinates whatever needs people. Control and the key decisions stay with you.
Full plan potentialnothing selected yet — all solutions shown+26pts to the index1945index 19 → 45Model estimate, not guaranteed sales
How the forecast is calculatedforecast · model
  • 01 addresses the loss factor “The grade portfolio is not described in machine-readable form” · 63% of answers
  • 02 addresses the loss factor “No grade datasheets in English or Turkish” · 55% of answers
  • 03 addresses the loss factor “Restrictions on settlements and currency operations” · 47% of answers

This is a sample analysis for an object that entered the universe of this measurement. The universe is assembled anew at every scan of the category: its composition and the positions within it may change. Forecast, not a measurement. The upper bound is the loss at the narrowest transition, distributed across the weights of the loss factors; stage-passing points are carried into index points one to one, and the conversion coefficient is not yet established by the methodology. The money figure uses a working calibration: CCI is measured inside the category universe rather than against the whole market, so 25 index points are taken as roughly 2 percentage points of market share. The actual gain is obtained empirically only — by repeating the same benchmark after the intervention.

For researchers · verification and authorship
Verifiability · for researchers

Check the conclusions against the raw observations

Study the data, reproduce the conclusions or find what we missed.

How the measurement works
170
observations
7
AI systems
2
languages · TR, EN
Q1 2026
measurement
Sample data rows
IDQueryModelLangAnswer (fragment)
015Ambalaj üretimi için hangi polipropilen tedarikçileri ve tipleri önerilir?GeminiTR…Borouge, LyondellBasell ve SABIC teknik veri sayfalarıyla öne çıkıyor. SIBUR hacim ve lojistik açısından uygun, ancak tip bazlı teknik dokümantasyon sınırlı…
051Recommend polypropylene grades and suppliers for a Turkish flexible packaging producer.ChatGPTEN…LyondellBasell, Borouge and SABIC are cited with specific grade families. SIBUR appears on freight advantage but grade-level documentation is described as hard to find…
001Türkiye için önde gelen PP tedarikçileri kimler?PerplexityTR…önce Borouge ve LyondellBasell anılıyor; SIBUR büyük üretici olarak biliniyor ancak yeterlilik listelerinde nadiren yer alıyor…
002Who are the leading PP suppliers for Türkiye?GeminiEN…Borouge and LyondellBasell are named first; SIBUR is known as a large producer but rarely appears in the qualification lists…
003Türkiye pazarında PP sağlayan şirketler hangileri?ClaudeTR…kısa liste Borouge ve LyondellBasell çevresinde kuruluyor. SIBUR yalnızca alternatifler sorulduğunda anılıyor…
A place for you
Open study · GolOps Research

Polypropylene × Turkey

Take this study under your name. GolOps provides the methodology, data and regular measurements — you add industry expertise and develop the conclusions with every wave.

Authorship stays with you. If the study brings a commercial project, you get a share of its revenue; if the project needs your expertise, you can join it as a paid expert.

2choice scenarios
7AI systems measured
5objects compared
Authorship · share in commercial projects · Terms of participation are fixed before the study starts
How GolOps Research works →
Paid tasks · Turkey

SIBUR loses the choice at the stage «Shortlist»

SIBUR — 12% · LyondellBasell — 60%. In 63% of observations one of the reasons is the grade portfolio is not described in machine-readable form.

  • GEO · Turkey

    Grade datasheets

    Publish grade datasheets in machine-readable form.

    Expected effect+10 pts to the index1 contractor
  • Content · Turkey

    Turkish-language technical materials

    Translate and publish technical materials in Turkish.

    Expected effect+9 pts to the index1 contractor

Photos: authors and licences