Video summary

Фид в Гугл Таблицах. Новое обучение с обсуждением

Main summary

Key takeaways

Technology

Overview

This video is a tutorial/training session about building and operating a Google Sheets–based product feed (“fit”) for Yandex Direct (and related ad formats such as product galleries/smart banners). The approach is presented as more automated and customizable than Yandex’s manual feed builder.

What’s new / why this approach

  • The speaker argues the template has been refined through accumulated experience and real “case” learnings from other users.
  • The goal of the “second stage” (a freer training format) is to collect audience feedback publicly (not only in private chats) and share working results.

Core product: Google Sheets template for feed generation

  • A special Google Spreadsheet template is used.
  • It loads JavaScript scripts inside the sheet; after filling data, it auto-generates the feed structure.
  • The file is typically a small multi-tab template; users mainly fill in commodity/product matrix cards.

Feed structure and required workflow

1) Duplicate the template

  1. Duplicate the template copy to work with it.
  2. The link may open in reader mode; users must make a copy and edit that copy.

2) Start with the “matrix” sheet

Product-level fields include (paraphrased from subtitles):

  • product title/name
  • description (character limits are mentioned)
  • image links
  • product link
  • currency (USDT or ruble)
  • price fields, including optional old price
  • category / model / manufacturer / competitor-related fields
  • additional parameters and keywords

3) Fill store-level fields

  • store name
  • company name
  • primary URL
  • currency

4) Optional “collections”

  • “Collections” act like product catalogs (aggregations of items).
  • They require IDs that link back to cards.
  • Purpose: connect catalog pages with cards via card identifiers.

5) Run “Fit verification” / validation

After authorization (allowing scripts), the sheet runs checks:

  • spelling/punctuation in titles/descriptions
  • link validity (e.g., 404 checks)
  • image link validity (also 404 checks)
  • structural errors such as missing required fields for collections/cards

If errors are found, the sheet points to the specific cells that need correction.

6) Generate the feed (“Fit generate”)

  • Produces the feed file for download.
  • The speaker recommends a better approach next.

7) Automated feed deployment and serving via link

  • Use “deployment” in Google/Sheets to get a web application URL.
  • This URL becomes the feed endpoint that Yandex Direct can scrape as a “full-fledged fit.”

Key technical recommendations / constraints

Text length and formatting

The speaker states practical constraints:

  • headings/title: max around 56 characters
  • description: around 81 characters

Even if long fields help relevance, the ad presentation may still truncate.

Image handling is critical

  • Images must be direct links (not local files uploaded somewhere in the sheet).
  • Two suggested options:
    1. Ask the client for a zip/pool of creatives, host them on the site, and provide direct URLs (possibly in a hidden folder).
    2. Request access/help to place images on the speaker’s infrastructure (time-consuming; possible time-out).
  • Warnings:
    • If images are missing or deleted (e.g., removed from an external host), ad visuals can break.
    • In some cases ads may still run without pictures, but visuals can look ugly/white/blank; moderation and presentation risks exist.
  • Note: the speaker mentions a previous use of an “Apex” upload service, but it became less reliable due to deletion/sanctions-related changes.

Category / model / manufacturer fields

The sheet expects users to fill:

  • category
  • model
  • manufacturer

Strategy described:

  • For non-ecommerce projects, fill broad but maximally relevant values that match a niche (based on queries).
  • Test from broad → narrower keyword coverage using category/model/manufacturer keys.

Brand recommendation:

  • Add not only the brand but also a few strong competitor brands.
  • Competitors may not be broadcast into ads automatically, but they still help coverage (as noted by an asterisk).

Ads parameter mapping (important)

During campaign setup, ensure:

  • title is taken from “name/title”
  • description is taken from “description”

The speaker recounts a mistake where incorrect mapping caused the feed fields to appear in ads in a messy way.

Price logic

  • Old price vs new price:
    • Use an “old price” to represent discounts.
    • If discount isn’t possible, set old price = new price.
  • Risk:
    • Missing/incorrect old price fields can generate errors; the sheet may enforce old price as mandatory for discount formatting.
  • Update behavior:
    • Changes can propagate quickly after updating the sheet data.

Type goods/services

  • A “product type” field distinguishes general vs private (goods vs services style logic).

Additional parameter blocks

The template allows adding arbitrary card parameters to influence:

  • audience selection
  • coverage
  • traffic volume

Keyword/competitor-related blocks:

  • competitors (comma-separated)
  • key phrases / target phrases (broad testing recommended)

Collections vs cards: what the speaker claims in practice

  • The speaker emphasizes that collections (catalog-style entries) can outperform or complement product cards in some niches.
  • Collections connect to cards via IDs and can enable catalog-level targeting, expanding where ads show (e.g., catalog pages).

Real-estate experiment (example)

  • For real estate-specific experiments, the speaker duplicated catalog-page content into collections/cards.
  • Catalog pages reportedly became the most effective promotion source, even when product information was not “enhanced.”
  • They also note that landing-page patterns matter:
    • some tests worked better on multi-page landing pages / catalog duplicates
    • than on a site-wide duplicate-company-pages approach

Performance/analysis claims (case study highlights)

The speaker provides anecdotal metrics:

  • In an e-commerce context, custom feed strategies reportedly achieved stronger KPIs than typical baselines (example cited: 17% cost per result vs lower single-digit benchmarks).
  • In non-commercial niches, they report success reaching a working campaign state after iterative testing—often within ~1 month and ~90–100% of cases.

Funnel behavior discussed:

  • Broad queries expand coverage and show product/gallery formats more effectively than narrow requests.
  • Smart banners/product ads can attract broader “awareness” traffic that later converts.

Dynamics mentioned:

  • After changing feed data, Yandex can update ads within seconds/minutes (“within a short time” after Google processes).

Notification concept:

  • The sheet can include a “notification test” / email alerts concept:
    • if the generated feed breaks, the system can notify by email which feed/task failed.

How it’s maintained

  • After initial setup, Yandex Direct can re-fetch the feed automatically from the deployment URL.
  • Updates to the sheet propagate to the feed endpoint, avoiding manual re-upload of the file each time.

Main speakers / sources

  • Lesha — primary instructor / template author-like speaker
  • Tanya — interrupts/questions to show the screen; participates
  • Semyon — participant; appears in Q&A and examples
  • Andrey — participant; asks/engages on collections/real estate topics
  • Nikita — mentioned as a participant asking about experience
  • Ilya — mentioned in a question about experience
  • Additional commenters/attendees — briefly referenced in Q&A

Original video