How do Saudi markets change when platforms and ecosystems…

Seifelden Alhajj · Revenue Engineering · Riyadh · الإصدار 1.0 · أكتوبر 2026 · الحالة: مخطط (المنهجية أولًا) / Version 1.0 · October 2026 · Status: planned (method first) Companion to «تصميم السوق» / Market Design (marketdesign_ar.md #research). Every study below is planned. None has run, none has results, and nothing in the essay depends on an outcome from here.

Why a research agenda, and why method first

In less than a decade, Saudi food service and retail have received, almost at once, delivery aggregators, a new entrant that reached about a tenth of delivery order volume within months (Redseer, February 2025), buy-now-pay-later providers licensed by SAMA, merchant-store platforms, POS ecosystems with their own app marketplaces, mandatory e-invoicing integration (ZATCA), an open-banking framework (SAMA), and an active competition authority that has written conditions into a delivery merger and consulted on a delivery-sector guideline. For an industrial-organisation economist that is a rare set of natural experiments, staggered by city, sector and date.

The honest position is that almost none of it has been measured from the restaurant's or retailer's side in public. So the agenda starts with methods that a Saudi group can run on its own data, and publishes only aggregated, anonymised results that pass the rules below. The question is always the essay's: who pays whom, who can leave, and who captures the value — measured, not asserted.

Rules for every study

Method before data, data before claim. Each study is pre-registered here (question, unit, design, what would change our mind) before any data is collected.
No platform market shares. We do not estimate or publish shares between delivery platforms. Where a study needs market structure, it uses published figures from the source that produced them, with the source's own definition and date.
Contracts stay confidential. Partner groups share settlement-statement aggregates, never contract text. No result identifies a platform's terms with a named restaurant.
Descriptive on law. Studies that touch the Competition Law describe published decisions and guidelines; they do not assess any firm's compliance.
Credit. Frameworks are credited to their authors (see evidence.md); Saudi figures to the body that published them, with date.
Status discipline. Status moves planned → designed → collecting → analysing → published. Nothing is cited from a study before "published".

Study 1 · صدمة الداخل الجديد: ماذا حدث لقنوات المطعم حين دخل منافس؟

Entry shock: what happened to restaurants' channel mix and contribution when a new delivery entrant arrived? Status: planned

Question. When a new aggregator entered a city with heavy subsidies, did a multi-branch group's total orders grow, or did orders move between doors (platform to platform, counter to app)? What happened to contribution per order and to the Demand Ownership Ratio?
Unit. Branch × month, 12 months before and 18 after entry.
Design. Difference-in-differences using the staggered timing of entry across cities as the treatment; branches in not-yet-entered cities as controls. Matched branch pairs (format, trade area, age) as in the Channel Ledger's holdout design (#holdout-design).
Data. Partner groups' POS and settlement statements (orders by door, deductions by line), aggregated per branch-month.
Outputs. Share of platform growth that was incremental vs moved (the "transferred growth" failure mode, Ledger #fm-3); change in remaining commission headroom by platform.
What would change our mind. If total orders rose in treated cities by the same amount as platform orders, entry created demand rather than moving it.

Study 2 · من يدفع ثمن رسوم المنصة؟ انتقال الرسوم إلى السعر

Pass-through: when platform fees move, where do they land — app prices, branch prices, or margin? Status: planned

Question. How much of a change in platform deductions is passed into app menu prices, how much into branch prices, and how much is absorbed in margin?
Unit. Menu item × platform × month, plus the same item on the branch menu board.
Design. Monthly public price panel (app prices as displayed to customers; branch menu boards) across three cities; event study around announced fee or programme changes; comparison with the published Al-Eqtisadiah survey (2023) and the gaps Argaam reported in January 2026 as reference points, not as baselines.
Outputs. The price gap by item class (ruler items, regular, channel items — Ledger §6) and its movement after fee changes.
What would change our mind. If price gaps do not move after fee changes, pass-through runs through margin, and the headroom test (not the menu) is the binding decision.

Study 3 · من يستخدم أكثر من تطبيق؟ إحصاء الاستضافة المتعددة

The multi-homing census: how many restaurants and customers use more than one platform? Status: planned

Question. Multi-homing decides who holds the bottleneck (Armstrong 2006). How common is it on each side in Saudi delivery, and does it differ by city, category and income?
Unit. Restaurant brand (listing on each app); customer (apps used in the last 30 days).
Design. (a) Restaurant side: a listing census of public app storefronts in defined grids of three cities, repeated quarterly. We publish the share of brands listed on one, two, three or more apps — never platform shares. (b) Customer side: a consumer panel survey with a validated question on apps used and apps paid through a subscription or wallet.
Outputs. The effective number of platforms per brand (the calculator's eff), and customer single-homing rates.
What would change our mind. High customer multi-homing would mean restaurants hold more leverage than the essay's default reading suggests.

Study 4 · كم زبونًا يتبعك لو خرجت؟ قياس كلفة التحول

Follow-through: measuring switching cost from the restaurant's side Status: planned

Question. If a branch pauses one platform for a defined period, what share of the lost platform orders reappears through other doors (branch, takeaway, own app, another platform)? This is the port input of the Dependency & switching calculator.
Unit. Branch × week.
Design. Controlled, time-limited pauses in matched branch pairs, only where the contract allows it and after legal review; synthetic-control comparison. Brand-name search share (pull) recorded from merchant reports where platforms provide it.
Ethics. No pause that breaches contract terms; customers see normal service in other doors; no pause in Ramadan or peak season weeks.
What would change our mind. A follow-through near zero would mean the platform, not the brand, owns the demand in that trade area.

Study 5 · اشترِ الآن وادفع لاحقًا: هل يزيد السلة أم يستعير من الغد؟

BNPL at the till: does it raise the basket, or borrow from next month? Status: planned

Question. For retailers and F&B groups that accept BNPL, what is the incremental effect on conversion, basket value, returns and 90-day repeat, net of the merchant fee?
Unit. Transaction and customer cohort.
Design. Randomised or rotated availability of BNPL at checkout (online) where the provider's terms allow; matched stores offline. The Ledger's increment test applied to a payment method.
Outputs. Contribution after the merchant fee per incremental basket, compared with the group's floor.
What would change our mind. If the basket uplift is mostly timing (purchases pulled forward), BNPL is a financing cost, not a demand engine.

Study 6 · السكك العامة: هل تخفض الفوترة الإلكترونية والمصرفية المفتوحة كلفة التحول؟

Public rails: do e-invoicing and open banking lower switching costs between software ecosystems? Status: planned

Question. ZATCA's Phase 2 integration waves and SAMA's open-banking framework standardise data that used to be locked inside vendors. Do they make it easier for a restaurant or retailer to change POS, ERP or payment provider?
Unit. Firm × quarter.
Design. Survey plus vendor-churn data where available; compare firms by the date their integration wave began (staggered by ZATCA's published thresholds) to study before-and-after churn and integration costs.
Outputs. Reported switching time and cost by software layer; share of firms that changed provider within four quarters of integration.
What would change our mind. If churn falls after integration, compliance work itself became a new switching cost.

Study 7 · خريطة المكمّلين: من يعيش على منصات نقاط البيع والمتاجر؟

The complementor map: who lives on Saudi POS and store-builder platforms, and who captures value? Status: planned

Question. How many third-party apps and integrations sit on Saudi POS and e-commerce platforms' marketplaces, in which categories, and how fast do they enter and exit? Where does the platform itself enter a complementor's category?
Design. Quarterly census of public app-marketplace listings (category, launch date, pricing model), coded with Jacobides, Cennamo and Gawer's (2018) complementarity types; event log of first-party entry into a category.
Outputs. A public dataset of categories and entry/exit; no revenue estimates.
What would change our mind. If first-party entry is followed by complementor exit in that category, the platform is acting as a dominator rather than a keystone (Iansiti and Levien 2004).

Study 8 · أسواق التوريد للمطاعم الصغيرة: السعر والائتمان

B2B supply marketplaces and small restaurants: price, credit and dependence Status: planned

Question. For independent restaurants and small groceries, do B2B marketplaces lower input prices and working capital, and do they create a new dependency (credit tied to one marketplace)?
Design. Price-basket comparisons across marketplace, wholesaler and distributor for a fixed list of SKUs, monthly; short survey on credit terms and switching.
Outputs. Price dispersion by channel; share of buyers whose credit ties them to one marketplace. Feeds the "B2B supply for smaller restaurants" option in the essay's third move.

Study 9 · سجل عام لقرارات المنافسة المتعلقة بالمنصات

A public register of Saudi competition decisions and guidelines on platforms Status: planned

Question. What has the General Authority for Competition decided, conditioned, blocked or consulted on in platform markets, and with what conditions?
Design. Legal-descriptive: a dated register built from GAC releases, Istitlaa consultations and reputable press, each entry with its primary source. No assessment of any firm's conduct.
Seed entries (verified, see evidence.md): conditional approval of Jahez's acquisition of The Chefz with four conditions (14 September 2023); draft guideline on competition in food-delivery platforms put to consultation (March 2025); GAC study reported by Argaam (29 January 2026); GAC review of Uber's offer for Delivery Hero (reported 24 September 2026, pending).

Sequencing

Order	Study	Why first	Data owner
1	S3 multi-homing census	Public data, no partner needed; sets the vocabulary	Public listings + panel
2	S9 decisions register	Public sources; frames every other study	Public sources
3	S4 follow-through	Directly calibrates the calculator; one group can run it	Partner group
4	S1 entry shock	Needs 2–3 partner groups across cities	Partner groups
5	S2 pass-through	Needs 12 months of price panel	Public prices
6	S5, S6, S7, S8	Sector partners (retail, software, suppliers)	Mixed

What this agenda will not do

It will not rank platforms, publish shares between them, estimate any platform's commission, or advise on legal compliance. It measures what a platform's entry does to the economics of the restaurant and the retailer, with methods they can repeat on their own data.

Back to the paper