Industry Oct 12, 2026

An advertiser’s source map: acceptance rules and the first pilot

Build a source card, agree enquiry acceptance and compare full pilot costs. Practical templates and two educational scenarios for products in Russia.

A
AffiliateCore
14 views
An advertiser’s source map: acceptance rules and the first pilot

What you will produce

An advertiser's traffic-source map is a working document connecting an enquiry's origin, the promise made to its author, acceptance conditions and the full cost of a result. Prepare it before selecting a placement. Buying clicks first and deciding later which enquiries qualify is a predictable cause of disputes with partners.

This guide concerns an ordinary permitted product or service in Russia. It does not promise access to every platform or vertical and contains no instructions for recruiting gamblers. Both numerical scenarios are educational, with independent hypothetical USD amounts—not currency conversions, AffiliateCore catalogue offers or customer results.

Your deliverables are a product brief, a shortlist of at most five suitable sources, agreed result definitions and a first-pilot plan. Selecting a channel, creating an offer and launching traffic are separate stages. Do not spend until restrictions, landing-page readiness and event verification are understood.

1. Separate source, format and measurement

A source explains where someone encountered the message: search, a creator's website, a community, an opted-in mailing list or a directory. Format describes the material: article, advertisement, video, comparison or form. The landing page continues the promise. A CRM records handling of enquiries. An affiliate platform connects an approved partner with agreed events within the capabilities of its integration.

Yandex Metrica and a CRM are not independent audience sources. SEO and paid search may serve similar needs but differ in costs, timing and placement mechanisms. A creator's followers are not automatically buyers of your product. A large audience may contain few people with the relevant problem.

Start with five possible rows:

  • Organic search: the need has been expressed; you need a useful page, evidence and development time. Include labour rather than calling the channel free.
  • Paid search: a query exists, but its meaning may not match your service. The promise, permitted advertisement and destination must align.
  • Advertising network: the message appears in another context. Explain its relevance and allow for an earlier decision stage in the test.
  • Creator or community: assess subject, placement rules, actual geography and agreed content. Audience size alone proves no outcome.
  • Your own voluntary mailing list: recipients should expect the message, with suitable subscription context, frequency and an easy opt-out. Buying a database does not establish permission.

This is a research framework, not a recommendation to launch five channels. A platform offering advertising does not establish eligibility for your account, region or category. Check current rules before each placement, including Yandex Direct requirements and restricted categories.

2. Product brief before recruiting partners

Complete one document understandable to someone outside your internal discussions.

Product and problem: what the customer receives. Geography: where purchase or service is genuinely available. Restrictions: age, customer type, equipment, plan or territory where relevant. Promise: what may be said before the click. Result: the action being verified. Sources: allowed, prohibited and requiring separate agreement. Materials: copy, images, rights and validity period. Review: responsible person, deadline and rejection reasons. Economics: reward, fees, handling costs and maximum acceptable cost.

Do not leave “quality lead” undefined. For inventory software, it might mean a demonstration request from a company representative with a suitable business scale and a voluntarily provided contact method. For repair services, it might mean an enquiry about a supported device inside the service area. These differ even if both forms ask for a name and telephone number.

List non-results too: page view, button press, unfinished form or a duplicate of an already recorded enquiry. Do not reject every existing customer's contact without an agreed rule. Acceptance criteria should not be invented after a submission arrives.

Start with this brief in our AffiliateCore workflow. Register as an advertiser, then complete the applicable account requirements and prepare your own offer through the existing process. Creating an offer guarantees neither approval, partners nor traffic. If your scenario is unsupported, do not disguise it with a similar form option: establish an applicable process first.

3. Source card: a comparison template

Use identical fields for each candidate:

  1. Where the material appears and who controls it.
  2. The problem that brings the audience there.
  3. Verified territory versus assumed territory.
  4. The exact material and its promise.
  5. The destination and what the visitor sees next.
  6. Whether current rules permit product, source and format.
  7. How reporting distinguishes the source without personal data in labels.
  8. Costs before the first result and while handling each enquiry.
  9. Who reviews cases and how disputed decisions are considered.
  10. The threshold for stopping or approving another limited budget.

Attach a message and landing-page mockup, not just a channel name. “Contextual advertising” is too broad to validate. “Search advertisement about managing two warehouses → page explaining product limitations → demonstration request” is a route that can be agreed.

Exclude incompatible candidates first. Compare the rest by task relevance, traceability, available materials, cost and handling capacity. Avoid numerical scoring that implies precision without evidence. “Unknown” is more useful than an invented conversion percentage.

4. Acceptance and disputes

Agree enquiry uniqueness, relevant qualifications, review deadline and observation window before testing. For example, the enquiry concerns the specified product and area, the person voluntarily supplied contact details, and a repeat of the same request inside an agreed period is not paid twice. Select that period for the product's cycle; there is no universal number here.

Keep four states separate: under review, accepted, rejected with a reason and cancelled after initial acceptance. The last requires a correction procedure agreed in advance. Do not merge unprocessed enquiries with final rejections. Comparing sources with different proportions of open cases is premature.

Reasons should be specific: unsupported model, location outside coverage or an evidenced duplicate under the agreed rule. “Poor quality” is insufficient for investigation. Suspicious behaviour is a signal to investigate, not proof against a partner.

Give partners a dispute route containing event ID, reason, date and responsible reviewer. Share only necessary information; do not put user contacts into public spreadsheets or sub-IDs. Advertisers remain responsible for their data and consent processes; a platform name does not transfer that responsibility.

5. Educational example: inventory software demonstrations

Assume ordinary software is available to target companies in Russia, with agreed search placements and a specialist creator's materials permitted. Verify that for a real product. The result is a demonstration enquiry matching the brief, not any visitor's registration.

Materials comprise a short search message about two warehouses, a page with import and access-right limitations, and a creator's comparison of use cases. Example promise: “Check whether the software fits two warehouses: limitations and questions for a demonstration.” Do not promise higher profits or free access without evidence.

Educational route: prepared AffiliateCore offer → approved partner and source → verified material → advertiser landing page → agreed event through the working integration → review → final result. Resolve missing event delivery before traffic starts. A read-only API is not automatically an endpoint for submitting new enquiries.

Plan: agree two sources separately, assign limits, test the technical transition without fictitious payable events, observe four weeks and finish reviewing every enquiry. Give sources separate nonpersonal codes. Attribution windows and definitions match. Advertisers must not generate fake enquiries to improve the figures.

Channel S: 60 enquiries, 40 finally accepted, 12 rejected and 8 cancelled. Assume USD 10 per accepted result: USD 400. An illustrative fee of 10% of that amount adds USD 40; this is not an AffiliateCore tariff statement. Materials cost USD 80, handling all 60 enquiries at USD 1 costs USD 60, and tools cost USD 20. Total USD 600, or USD 15 per accepted result.

Channel C: 45 enquiries, 30 accepted, 10 rejected and 5 cancelled. Rewards USD 300, illustrative fee USD 30, materials USD 120, handling USD 45 and tools USD 15. Total USD 510, or USD 17 per accepted result. Additional taxes and charges are assumed zero; add actual ones separately. Cancellation payments have already been corrected and no cases remain open.

With a limit fixed beforehand at USD 16, S passes the educational threshold and C does not. This is not a universal claim for search: it describes specific materials, costs and a cohort. Give S a limited repeat with quality checks. For C, investigate fit and whether preparation costs can fall without reducing usefulness. Changing reward, product and audience together would obscure the comparison.

6. Educational example: a local repair service

Assume a provider serves one city and specific printer models. The result is an accepted diagnosis request. Consider a local directory and a relevant community. Check both formats' availability and each platform's rules before placement; a category label is not permission.

Materials include supported devices, service area, pricing explanation and a “model → fault → district → preferred contact method” template. A community placement needs useful preparation advice, not disguised bulk messaging. Never invent branches.

Test sequence: verify facts, agree placement, check the mobile journey, define the event and source codes, run a bounded observation period, check technical errors daily and close the cohort's review. Cheap clicks are not sufficient evidence. If agreed handling deadlines are missed, fix operations first.

Directory D: placement USD 120, preparation USD 30, handling 24 enquiries × USD 1.25 = USD 30, tools USD 10. Total USD 190. Of 24 enquiries, 15 are accepted, 6 rejected for territory or model and 3 cancelled. Cost per accepted result is USD 12.67.

Community M: placement USD 80, material USD 50, handling 30 × USD 1.25 = USD 37.50, tools USD 12.50. Total USD 180. Of 30 enquiries, 12 accepted, 12 rejected and 6 cancelled. Cost per accepted result is USD 15. Neither route has an additional performance reward: this example pays for placement. Do not add an imaginary CPA reward, but include one if the actual agreement specifies it.

The limit set beforehand is USD 14. D passes this educational filter; M does not, despite bringing more raw enquiries. Investigate whether territory was unclear, unsupported models attracted attention or processing was slow. Do not allege artificial traffic without evidence. Evaluate revised material in a new cohort rather than rewriting past results.

7. Launching, stopping and expanding

Before spending, have a working landing page, processing owner, acceptance rules, permitted materials and verifiable measurement. Check the actual integration with a safe technical event where supported; do not create fictitious paid conversions. External analytics and platform figures may differ in windows and definitions: document those differences.

Stop immediately for a prohibited source, misleading material, unjustified data transfer or unverifiable events. Stop spending on economic grounds when the limit is reached, a closed cohort fails the agreed threshold and no concrete corrective hypothesis exists. Insufficient data proves neither success nor a right to expand indefinitely.

A small next step is appropriate only with confirmed quality, understood cancellations, processing capacity and acceptable full costs. Review the next cohort instead of only the lifetime average. Inform partners before new conditions take effect.

Assignment: complete the product brief, two source cards and budget, then prepare your own proposal. Sign in to the AffiliateCore advertiser account and apply the conditions when creating an offer draft. Do not publish an educational offer in the real catalogue merely to test this article. The publisher traffic-source map explains the other side of the journey; your advertiser decision should rest on acceptance and full economics rather than click count.