Register your interest: Tag @Cody, get an agent
BlogResources

Pinterest and CodeWords: a catalogue problem, not a social one

Pins have a long tail and behave like search results. Catalogue feeds, rich pins, and why the analytics timescale is completely different from every other platform.

Rithul PalazhiRithul Palazhi11 min read

Summarize with AI

Pinterest and CodeWords: a catalogue problem, not a social one
On this page

Treating Pinterest like a social network is the mistake that makes automating it disappointing. People arrive at Pinterest looking for something — a recipe, a room, a product — which makes a pin far closer to a search result than to a post, and it makes the timescale entirely different.

A post on most platforms is finished within a day. A pin can accumulate traffic for a year, which changes how you should publish, how you should measure, and what is worth automating.

What we'll cover

A pin is not a post

Discovery is search-led. People find pins by looking for something, which means text — title, description, board name — does work that a caption elsewhere does not.

The tail is long. A pin can drive traffic months after publication, so judging one on its first week is judging it on a fraction of its life.

Volume is not the strategy. Pinning the same thing repeatedly to many boards is a pattern the platform discourages, and it produces a worse account rather than a bigger one.

The image is the entry point and its aspect ratio matters more than on most platforms, since it determines how much space the pin occupies in a feed.

Links matter. Pinterest sends traffic onward, which is the reason most businesses are there, and a pin whose link is broken is worse than no pin.

Everything below follows from these. Publish consistently, write the text properly, keep links correct, and measure over months.

What the Pinterest API reaches

Pins can be created, read, updated, and deleted, with their image, title, description, link, and board.

Boards and sections can be created and managed, which is how a catalogue is organised.

Analytics for pins, boards, and the account, including impressions, saves, and outbound clicks.

Catalogues accept product feeds, generating product pins from your own catalogue automatically.

Ad accounts and campaigns are available with the corresponding permissions.

Rich pin validation confirms that your site's markup is providing the metadata that makes a pin richer.

Access requires approval, with different products carrying different requirements, which is worth confirming before scoping.

The catalogue, which is the real integration

For anyone selling products, this is where the value is and it is not a publishing integration at all.

A feed generates product pins automatically from your catalogue, so the pins reflect what you actually sell.

Price and availability stay current, which is what stops customers arriving at a sold-out product — the failure that loses both the sale and the trust.

Feed errors are the thing to automate. A rejected feed, or rows rejected within an accepted feed, stops products appearing and reports nothing to anybody who would notice.

Keep the feed generated from the source of truth, not maintained separately, or it diverges within a month.

Group products sensibly, since the organisation of the catalogue affects how it surfaces.

An automation that generates the feed, submits it, and reports rejections by reason is worth more than any amount of manual pinning.

Connecting it to CodeWords

CodeWords connects to more than 3,000 integrations, and the connection is made once and reused.

  1. Open CodeWords and start a new automation.
  2. Describe what should happen in plain language to Cody, the automation builder: which board or catalogue, what triggers a pin or a feed submission, and where the reporting goes.
  3. Authorize the connection with an account holding the necessary business permissions.
  4. Describe the exceptions: a rejected feed row, a link that no longer resolves, an image of the wrong shape.
  5. Publish to a test board before anything reaches a live account.

You describe the outcome; Cody builds it, connects it, and deploys it. The free plan covers light use, with Pro at $39 per month and Business at $100 per month as usage grows; details are on the pricing page.

Seven automations worth building

Catalogue feed generation and submission. From your product source, on a schedule, with price and availability current.

Feed error reporting. Rejections grouped by reason, since a category of products silently absent is the characteristic failure and nothing announces it.

Publishing from your content pipeline. When something is published, create the pin with the right image, text, and link, rather than somebody doing it later or not at all.

Link checking. Pins whose destination no longer resolves, reported for fixing. Pins outlive the pages they point at, and a long tail is only valuable while the link works.

Performance reporting on the right timescale. Monthly and quarterly, since a weekly report on a platform with a months-long tail measures almost nothing.

Cohort reporting by publication period. How pins published in a month performed over the following six, which is the only view that reflects how the platform actually works.

Board audit. Boards with few pins, boards that duplicate each other, and pins on the wrong board, since the organisation affects discovery.

Measuring on the right timescale

Weekly reports mislead here. A pin's first week is a small share of its eventual performance, and judging content on it produces wrong conclusions consistently.

Report by cohort. Pins published in January, measured over the following six months, compared with February's — this is the view that shows whether the content is improving.

Outbound clicks matter more than impressions for most businesses, since traffic is the reason to be there.

Saves indicate durable interest, which correlates with the long tail better than immediate engagement does.

Expect a skewed distribution. A small number of pins will account for most of the traffic, so averages describe nothing real — report the distribution.

Keep your own history, since analytics availability is bounded and the timescale that matters here is longer than most platforms retain in detail.

Boards and metadata that do work

Boards are a taxonomy, and a well-organised set helps discovery in a way that an arbitrary one does not.

Write descriptions as though for search, because that is roughly what they are, without stuffing them with keywords.

Titles carry weight. A descriptive title outperforms a clever one, which is the opposite of the instinct on most platforms.

Use rich pins where your site supports the markup, so pins carry current metadata from your pages.

Keep the link correct above everything else, and validate it when the pin is created rather than discovering months later that a whole board points at a retired category.

Images, which are the whole entry point

Pinterest is a visual index, and the image decides whether anything else you did matters.

Aspect ratio determines the space a pin occupies. A tall pin takes more room in a feed than a square one, which is why the platform's recommended ratio is worth following rather than using whatever the product photo happened to be.

Generate the image rather than reusing the product shot. A pin with the product, a short line of text, and consistent branding performs differently from a bare catalogue photograph, and generating it from a template is exactly the kind of work automation does well.

Check the image loaded. A pin created with a failed or placeholder image is worse than no pin, and it is a failure that produces no error.

Keep resolution adequate. An image that looks acceptable at thumbnail size and poor when opened undermines the click it earned.

Be consistent across the account. A board where every pin looks like it came from the same place reads as a catalogue rather than a collection, which is what people are looking for.

Building it so it survives

Make pin creation idempotent, keyed on the source content, so a retry does not produce a duplicate pin.

Validate before creating. Image aspect ratio, description length, and a link that resolves.

Handle token refresh properly, which is where these integrations break after a few weeks of working.

Record what was published, since your own record is more reliable and longer-lived than querying the platform later.

Report the outcome. Pins created, feed rows accepted and rejected, and links that failed validation.

Limits worth knowing about

API access requires approval, and different products have different requirements — confirm what you have before planning.

Rate limits apply per application and per account.

Analytics retention is bounded, which is awkward given the timescale that matters, and is the reason to keep your own history.

Duplicate pinning is discouraged, so an automation distributing the same pin widely is working against the platform.

Feed processing is asynchronous, so submission is not acceptance and the result needs to be checked rather than assumed.

What to build first

Catalogue feed generation with error reporting: the feed produced from your product source, submitted on a schedule, with rejections grouped by reason and sent to whoever owns the catalogue. For anyone selling products, it keeps price and availability accurate and it surfaces the silent failure where a whole category stops appearing.

Two habits make the difference. Report rejections by reason rather than as a count, since ninety rejections with one cause is one fix. And check the feed result rather than the submission, because submission succeeding says nothing about whether the products were accepted.

Frequently asked questions

Why is our weekly Pinterest report so volatile?

Because a week is the wrong timescale. Pins accumulate traffic for months, so a weekly view measures a small and unrepresentative slice. Report by publication cohort over months instead.

Should we pin the same image to several boards?

No. Repeated pinning across boards is discouraged and produces a worse account rather than wider reach. Publishing consistently with well-written text does more.

What is the highest-value automation here?

For a product business, the catalogue feed and its error reporting. It keeps price and availability correct, and it catches the case where a category of products silently stops appearing — which nothing else reports.

Why did products disappear from Pinterest?

Almost always feed rejections. Submission succeeding is not acceptance, and rows can be rejected within an accepted feed. Check the processing result and group the rejections by reason.

Which metric matters most?

Outbound clicks for most businesses, since traffic is the point, with saves as the leading indicator of a long tail. Impressions are the least useful of the three and the most frequently reported.

Yes, and it is routinely skipped. Pins outlive the pages they point at, and the long tail is only worth having while the destination works. A periodic link check on your own pins is short work.

Why does the integration stop working?

Token refresh, usually, which is the standard failure for every social integration. Exercise the refresh path during testing rather than meeting it several weeks into production.

Should we use product photos directly as pins?

Usually not. A generated image with the product, a short line of text, and consistent branding suits how people browse, and generating it from a template per product is straightforward automation. Aspect ratio matters too, since it decides how much space the pin takes.

What is the most overlooked check?

That the image actually loaded and the link resolves. Both fail silently, both make the pin worthless, and neither produces an error anywhere — which is why they need checking at creation rather than discovering months later.

Get started today

Your first workflow is free to build.

Describe what you need. Cody handles the build, the connections, and the deployment.