November 10, 2020 11 minutes to read

Setting up end-to-end analytics for a company with a long production cycle: a case from Australia

Prior to contacting us, Payless Promotions only kept accounting records without tracking marketing campaigns. We provided them with a structured data warehouse, a system for end-to-end analytics, taking into account the specifics of the business.

Now the company can make decisions based on data about traffic sources and their costs.

Payless Promotions

Manufacture of branded products to order


  • Founding year2000
  • NicheB2B
  • GeographyAustralia

Tasks

The task that the Client set for us was to set up end-to-end analytics in order to see data in a single system for analyzing the effectiveness of marketing investments.. 

To determine ROAS (return on advertising spend), the system must contain three key indicators in a single interface:

  • traffic sources
  • costs by traffic sources (by groups, campaigns, keywords, etc.)
  • the number of leads, completed orders and income from them. 

The main difficulty of the task was that between the day when the order was received and the date when you can take the profit, there were 40-60 days on average. Because of this, it was impossible to assess the return on investment in advertising using standard e-commerce tools. 

The site does not have a traditional e-commerce way of buying a product, as this is a custom product that is made for the consumer. The final cost depends on the quantity, material, invoice, shipping cost and other characteristics of the goods. 

The transaction consists of the following steps: 

  1. Making an application on the site (via chat, application form, callback).
  2. Calculation of the final cost of the order by the manager.
  3. Negotiations with the client and order confirmation. 
  4. The product is manufactured and shipped to the customer. 
  5. The company takes the profit on the order.

Up to 60 days could elapse between the third and fourth stages. At the same time, the Client could not attribute this order to profit until the moment of its shipment. There were several time periods: 

  1. Between the cost of traffic and making a profit on the order (the Client incurred the costs of attracting a buyer today, and the data on the completed order appeared in the analytics a month later).
  2. Between the application and the execution of the order.

Order data

At the beginning of the project, the company did not have a single system that would contain the structured data we needed. 

Sales data was stored in 2 systems: Suite CRM – for registering applications; Zebra – accounting system for production (similar to 1C).

The client received data on his final marginality from Zebra. The systems did not communicate with each other.

Traffic

Traffic data was collected in Google Analytics, in which only basic goals were configured. E-commerce has not been set up.   

Solution

It was decided to integrate e-commerce in two stages.

Stage 1

Setting up Google Analytics, streaming cost data using OWOX BI.
Google Ads was linked to Google Analytics, the data came through a direct link. To transfer data from Bing and Facebook Ads, we used OWOX Bi.

Setting up the transfer of data on completed orders to Google Analytics via the Measurement Protocol. In this case, the data matches only by source/channel/campaign and there is a significant discrepancy in time. 

In the case of the Client, the configured e-commerce will always provide data with an offset in the reports. There are several reasons for this:

  1. When placing an order, how much it costs is not clear. In the process of discussion with the manager, the volume of the order may change.
  2. Order fulfillment varies. 
  3. In the process of making a purchase, the customer can edit the components of the order. 

We implemented data transfer through the Measurement Protocol knowing about this feature in advance. Even when sending data via the Measurement Protocol, due to the logic of the system, the order is tied not to the session date, but to the date the request was sent to Google Analytics. 

We made this decision because the Client agreed to evaluate the data in Google Analytics based on a period of 1 quarter.

3. Implementing the SourceBuster script

We implemented the SourceBuster script for all forms and chats on the site.

For each request, the Client’s CRM system saved data with the following set of fields: 

  • source
  • medium
  • ad group
  • term
  • content
  • technical information for Measurement Protocol operation (Google Analytics clientID, IP, User-agent browser).

For this reason, we applied waterfall logic (sequential check). At the time of receiving the order, the system checked whether we had a card for this buyer. If so, whether we know the acquisition source. Together with the Client, we decided to attribute all orders to the original source that brought the client, regardless of whether this is the first order or a repeat one. Data from SourceBuster was written to the lead card and transferred to the order entity (job), and then to the client card (account) in CRM.

As a result, we received a summary report for CRM with the following data: 

  • request id (lead)
  • lead creation date
  • order id (job), which is linked to the order id
  • SourceBuster fields
  • order status
  • turnover on order
  • profit on order.
The table shows an excerpt from the report with the data being changed.

This data is already enough to quickly assess the contribution of traffic sources to leads and orders. 

Stage 2

Combining data based on exports from CRM and Google Analytics

Our task was to combine data on costs, payback and traffic for a specific date in one interface. 

We used a simplified principle of cohort analysis. The data is refined over time and it is possible to evaluate the payback of the cohort by the date the traffic costs are incurred.

That is, in August, you can clearly analyze the profit from advertising investments for May. 

ROAS Summary Report Architecture

The process of generating the final ROAS report is as follows.

  1. Once a week, sales data is uploaded from CRM using a connector, in which there are fields by traffic sources (marked under SourceBuster).
  2. Similarly, once a week, Google Analytics uploads data on traffic sources with the following fields: date, source/channel, clicks, sessions, costs.
  3. With the help of massive transformations, 2 tables are formed, which we connect by the key “date/source/channel/campaign/»
  • from GA, we take data on costs and traffic for the date
  • from CRM, we take data on applications (leads) on the date and subsequent completed orders with data on profit.

Prepared export from Google Analytics after transformations looks like this:

The table shows an excerpt from the report with the data being changed.

Result 

Before contacting the agency, the Client was only accounting. He took the amount spent on marketing, divided it by the profit received, and approximately saw the total marketing payback. Based on this data, it was impossible to make decisions at the source or campaign level. 

Final report:

The table shows an excerpt from the report with the data being changed.

After we have introduced the changes described above, the Client began to conduct high-quality optimization of his advertising campaigns according to parameters that he had not tracked before: Cost Per Lead, Сost Per Order, ROAS and the share of the advertising campaign in the total profit received.

By connecting end-to-end analytics, the Client was able to refine the analysis to the level of an ad group and a keyword. The Client started working on optimizing campaigns in Google Ads. He saw, for example, that within the same ad group there is a mixed demand — commercial (B2B) and end-user (B2C) requests. It was decided to cut bids on campaigns and groups that did not pay off in the quarter. 

The client received a structured data warehouse, a system for end-to-end business-specific analytics, which is an accurate tool for deciding on traffic sources and costs. 

Need powerful analytics?

We know how to customize it to your needs

We don't work with:

  • russia
  • CBD
  • Tobacco
  • Politics
  • Gambling
  • YMYL
  • Crypto
  • Pornography

Certificates
and Awards

Google Analytics Individual Qualified since 2009

Meta Business Partner. Only ~16 in Ukraine

eSputnik Partner since 2019

MailChimp Experts since 2010

UpWork Top Rated

#364 in the TOP-1000 companies worldwide, 2025 by Clutch

TOP Pay Per Click Company in Ukraine, 2025 by Techbehemoths

Clients

We have worked with 281 Clients and helped them enhance online marketing and earn more.

Clients
ABOUT US

Why did we choose Roman.ua?
Because they are geeks.