Sri Lanka's Media-Buying Market: Inventory, Service Fees and the Evidence Boundary

Sri Lanka’s media-buying market cannot be measured by adding agency invoices to broadcasters’ or platforms’ advertising revenue. The same inventory purchase can appear in both. The useful distinction is between the money buying advertising space or time, the service fee earned for arranging it, and the advertiser’s wider campaign budget.

The statistical evidence has nevertheless improved since this report’s August update. The Department of Census and Statistics (DCS) published a later structural benchmark: its 2019 survey describes reference year 2018, when Division 73, advertising and market research, recorded Rs28.197 billion of output and 8,165 persons engaged. This is historical evidence about establishments in a broader service division. It is neither 2026 media expenditure nor a measure of digital advertising.

Evidence searched through 20 October 2026. Coverage is limited: the analysis establishes transaction boundaries, a historical statistical comparison, observable supplier offers and dated regulatory conditions. It does not estimate current national spend, channel shares, net media prices or agency profitability.

A newer historical benchmark, with the same measurement warning

The 2017 and 2019 DCS reports permit the following comparison of Division 73. Monetary values are nominal rupees, displayed in billions.

The table is two historical snapshots, not a current growth series. Both surveys cover service establishments with at least five persons; the 50-person-and-above stratum was fully enumerated and smaller establishments were sampled. Division 73 also includes market research. Freelancers and smaller establishments fall outside the frame, while in-house marketing work is not separately identified as an advertising industry. Calendar-year observations may include accepted April–March financial-year returns.

Intermediate consumption increased faster than output between these snapshots. That does not establish margin compression: statistical value added is output less intermediate inputs, before treating wages and other claims on value added as business profit. Nor does growth in persons engaged establish comparable jobs or worker outcomes. The reports do not provide the matched firm panel needed to separate entry, reclassification and changes within continuing businesses.

One campaign, several different money flows

The audit’s core is the paid planning, selection, purchase and management of media for campaigns directed at Sri Lankan audiences. Direct inventory sales are a related transaction examined to explain that service, rather than added to its revenue. Creative-only work, stand-alone PR and market research remain adjacent activities. Payer residence, supplier residence and audience geography must be recorded separately; a foreign platform invoice does not, by itself, identify where an advertisement was delivered. The adjacent advertising-services audit covers the strategy, creative and execution work surrounding that purchase. These are connected analytical boundaries, not additive industry totals: ISIC Rev.4 class 7310 includes creative services, media planning and buying, and media representation.

Here, net media expenditure means inventory cost after disclosed discounts and rebates, with fees and taxes separately identified. Gross billings describes money invoiced before removing identified pass-through items. Neither term automatically equals retained agency revenue. Contractual principal/agent roles and actual accounting policies determine how an individual business records revenue. The ONS/Destatis advertising measurement paper, slides 7–11, explains these statistical measurement problems; it is a conceptual comparator, not evidence of Sri Lankan commission rates.

The commercial implication is direct. A larger invoice can reflect more inventory purchased, a different mix of bought-in services or a change in billing route, without a corresponding increase in the buying service’s income. Comparing agency scale therefore requires the same revenue basis and period. No inspected invoice panel supports a national reconciliation in this edition.

The available offers span managed buying and direct platform use

Two Sri Lankan provider pages show the boundary in practice. Publicis Groupe Sri Lanka lists omnichannel strategy, planning and buying, and sponsorships and partnerships alongside creative and production services. Ogilvy Sri Lanka lists planning, buying, monitoring and analytics alongside campaign strategy and production. These are current website offers inspected on 20 October, with no publication dates stated. They establish advertised services, not delivery quality, national coverage or supplier rank.

A different operating route is a platform billing account. Google Ads’ billing instructions explain the account’s billing country, currency, payment profile and payment settings. They show why an advertiser needs to retain the billing configuration as well as campaign reports. The documentation does not establish how many Sri Lankan advertisers buy directly, which settings every local account can use, or whether self-service is cheaper after staff time and supporting services.

The relevant competitive comparison is consequently the work transferred. Managed buying can include planning and coordination across suppliers; direct platform use leaves those tasks with the buyer or another provider. The evidence does not support a single national buyer-power score, a complete agency count or platform concentration ratio. Offers overlap, and the inspection of two providers is purposive rather than representative.

An official procurement gives a more concrete view of outdoor delivery. SLTPB’s BIA pillar-advertising tender, closing 3 February 2026, sought installation and operation of 33 LED screens, local fault support, maintenance, content-production pricing and playback/uptime reporting. It links access to airport inventory with physical service obligations and a proposed financial return to the rights holder. These are tender requirements, not an award, operational screen count, advertiser price or realised revenue. Conflicting clauses—three years from acceptance versus an end date in November 2028, and an allocation variously described as “50%”, 9 arrival plus 8 departure pillars (17 of 33), and 16 pillars in Submission Form A5 (p.29)—prevent treating the proposal as an internally settled revenue model.

The buyer, payer and audience have different needs

For this audit, customer segments are defined by the buying task: a single-platform purchase, a coordinated campaign across media, or a regulated-product campaign needing additional approval. These are analytical segments, without estimated sizes. The advertiser specifies the commercial objective; the payer settles the bill; an agency may plan and execute; the audience receives the message. One organisation may occupy several roles. The airport tender adds a distinct relationship: a rights holder selects an operator, which is different from an advertiser buying an impression or placement.

This distinction matters when assessing outcomes. An invoice can substantiate cost. A placement record can substantiate delivery. Neither proves incremental sales, unique audience reach or customer satisfaction. No representative buyer survey, common campaign-outcome dataset or worker survey was located in the specified search. Claims about sector-specific adoption, regional reach and the success of large versus small advertisers remain outside the conclusions. Telecom connections, network traffic and card e-commerce transactions do not answer those questions and are not used as advertising-expenditure proxies.

Integration is observable; effectiveness needs another test

Ogilvy Media’s undated Sprite case describes weather-responsive creative on LED displays at eight Colombo locations, with named backend, integration and creative contributors. Taken as a company account, it illustrates a delivery dependency: buying the display space alone is insufficient when creative changes depend on a working data feed and implementation partners. It does not establish campaign timing, incremental sales, a national adoption rate or superiority over conventional outdoor advertising.

That distinction tests the wider technology thesis. More interconnected delivery can create additional coordination work, but the evidence does not show whether its benefits exceed the fees and operational cost. A useful observation would be a dated campaign record joining approved creative, live delivery, inventory invoices and an independently interpretable outcome measure. The absence of that record prevents an efficiency claim, not a description of the disclosed implementation.

Rules affect the offer and the data used to deliver it

CAA Direction 91, dated 17 May 2023, addresses defined e-commerce entities and platform operators marketing and selling goods. Its pre-purchase provisions cover misleading advertising, offer conditions and extra charges, false or hired reviews, incentives and data notices. That scope must be checked before applying the direction to a campaign; it is not a universal advertising approval code. For medicines, devices and borderline products, the NMRA’s published guidance separately requires prior written advertising approval.

The immediate dated transition is Gazette 2498/16, issued 22 July 2026. It appoints 1 January 2027 for PDPA sections 2 and 3, Part I and Part III. This order does not commence every part of the Act. The 2025 amendment expressly permits different appointed dates. Audience-list and campaign-data arrangements therefore require a provision-specific review; this report does not infer universal consent, messaging or penalty rules from that single order.

These records support a need for dated offer, approval and data-role documentation. They do not show the industry’s compliance rate or the financial effect of enforcement. National tax rates likewise cannot be treated as a uniform effective charge on all inventory and agency services; invoice and taxpayer treatment must be established before comparing net costs.

What would change this assessment

The assessment date is 20 October 2026, with an outlook to 20 April 2027 and earlier reassessment if rules or material evidence change. This outlook is conditional rather than a spend forecast. If buyers require invoice-to-delivery reconciliation, suppliers able to separate inventory, fees, rebates and results have an observable basis for differentiation. If buying moves into a platform account or an in-house team, local agency billings can fall without proving that total advertising demand fell. If new commencement orders change the data-protection timetable, affected campaign workflows must be reassessed against the new text.

Three market requirements follow from the evidence: traceable cost reconciliation for intermediated buying; verifiable execution across the suppliers needed for a campaign; and applicable-rule checks for regulated offers and personal-data use. Tests are whether invoices reconcile, delivery records identify the responsible suppliers, and dated approvals and processing-role records exist. These are proposed market success conditions, not claims that named firms already satisfy them. They cease to apply in their present form where the transaction has no intermediary, no integration dependency or no relevant regulated activity.

Methods, limits and publication history

The DCS table uses Table 5.2 in the 2017 report, PDF p.31, and the 2019 report, printed p.19. The publications appeared in August 2019 and August 2022 respectively. Differences use (2018 value / 2016 value − 1) × 100, calculated from unrounded rupees and rounded to one decimal. Output less intermediate consumption equals reported value added exactly in both rows. No inflation adjustment, currency conversion, missing-year interpolation or current-market extrapolation is made.

The source search covered DCS annual surveys and service-survey materials, CBSL advertising-expenditure searches, official classifications, regulator records and named provider pages. No current national net-media-expenditure series was located in that search through 20 October. The 2018 survey landing page did not expose a usable report; targeted retrieval of the detailed SLSIC PDF failed. The inspected UNSD class is used for conceptual classification, while the DCS division remains the actual statistical population. A search failure is not evidence that the activity or a private dataset does not exist.

The original publication date remains 23 June 2025. The August 2026 correction removed unreproducible US$462–500 million market estimates, a US$1 billion-plus 2034 forecast, precise shares, margins and numerical competitive scores. Those withdrawals remain. This edition adds the reference-2018 DCS benchmark, replaces connectivity-led inference with transaction analysis and narrows absolute claims about missing national measures to the documented search. It does not reinstate the legacy figures. The earlier Online Safety discussion is not carried forward as a current legal-status conclusion: the inspected parliamentary record establishes a 23 June 2026 second-reading/committee stage, not a full October status check.

Future research should prioritise a consistently defined expenditure panel, including direct platform purchases and disclosed discount treatment, then comparable campaign delivery and outcome data. Until those exist in an inspectable form, this edition’s coverage limits remain material.