End-to-end analytics and optimization of paid promotion on message boards for a network of service centers

The initial situation is a network of 6 service centers for repairing household appliances in 3 cities. Now we place about 120 ads on Avito and Yula, we use paid promotions and promotions, but we don’t understand what ads and services actually bring in requests and sales. Transactions are recorded in CRM (amoCRM), but the source is often lost. There is call tracking (Calltouch), some of the leads come through messengers. The budget for paid promotion on message boards is about 2,500 USD/month, the current cost of the application is unstable.

The task is to build a correct measurement of the effectiveness of paid promotion on message boards and provide a set of specific optimization solutions: what to turn off, what to scale, which ad-service-geo combinations provide a margin.

Technical environment and data - access to Avito Pro (company account), Yula (office), amoCRM, Calltouch, GA4 (available on the site, but the site is only used as a business card), Google Sheets tables with expense history and a list of advertisements. Placement automation is not used. Access to site APIs may be limited - the solution should work without an API, but with priority on semi-automation where possible.

Stages and expected result
1) Audit of the current lead generation scheme - analysis of all entry points (calls, chats, forms, messengers), how sources are currently identified, where attributions are lost, what reports are available in the site’s accounts.
2) Designing a tracking scheme for bulletin boards - UTM and tag rules, unified naming of ads and campaigns, ad-service-city linkage, mapping to CRM fields.
3) Setting up and checking the end-to-end connection - correctly identifying the source at the lead/deal level in amoCRM for calls and requests from chats, setting up the transfer of parameters to Calltouch/CRM where technically possible, documenting site restrictions.
4) Collection and integration of data into a single showcase - Google Sheets or Looker Studio: promotion costs, impressions/contacts/requests from accounts, leads and sales from CRM, calls from call tracking. Deduplication of leads by phone and time of contact. Calculation of CAC, CPL, CR per sale, gross margin by type of service.
5) Analytics and optimization recommendations - identification of unprofitable ads and groups, recommendations for budget redistribution, a list of 10-20 specific actions (pause/restart/change of geo/schedule/promotion packages), test plan for 2 weeks.
6) Transfer of the result - a ready-made dashboard, a reference table of naming and tags, brief instructions (up to 3 pages of text) on how to maintain the system and how to make decisions weekly.

Restrictions
- You cannot change CRM (amoCRM remains).
- You cannot ask clients to go to the site - the main flow should remain inside the message boards.
- The solution should not require the development of a separate server; webhooks/integrations/scripts in Google Apps Script are acceptable.
- Access to accounts is provided for the duration of the work; all settings must be reproducible according to the instructions.

Acceptance criteria (measurable)
- In CRM, at least 85% of new leads from message boards have the following filled in: site, city, service/category, ad identifier or its unambiguous naming.
- The dashboard shows in one place for each site and city: expenses, requests, leads, sales, CPL, CAC, CR, gross margin. Data update - at least once a day manually or semi-automatically.
- A list of optimizations with a forecast of effect: for each item, the current metrics and the expected change are indicated (for example, a decrease in CPL by X% or an increase in CR by Y%) and test conditions.
- A control check was carried out on a sample of 30 leads - the correctness of the source and link to the ad was confirmed using Calltouch/CRM/office logs.

It is important that you are interested in the accounting system and management conclusions regarding paid promotion on message boards, and not in rewriting ad texts or design.