Naming convention template for Ecommerce

The naming convention with three segments a store needs: which slice of the catalogue a campaign sells, which promotion it sits inside, and which creative set an ad belongs to.

Everything else is the general convention, unchanged. One row per segment, fixed positions, a closed list of values with an owner, one separator and a change rule that does not break reporting history.

Adopt it before the first season is built. A store's account grows faster than most, because every promotion and every segment tends to produce new campaigns, and an account built for a year without a convention cannot be grouped by anything useful afterwards.

Use it again whenever the range is restructured, because the catalogue segment list is the one that goes stale.

Ask what a store wants to know at the end of a month and the answer is almost never how a platform performed. It is how new arrivals did, or whether clearance moved, or what the promotion cost across everything it touched.

Those questions are filters if the catalogue segment and the promotion window are in the name, and they are a manual exercise if they are not. That is the whole argument for this version of the convention: the two things a store reports by are the two things most accounts leave out of their naming entirely.

It also changes what the account looks like. A convention with a segment position encourages campaigns that are about one segment, which is the structure a store wants anyway, because the margin, the target and the creative all follow the segment.

The catalogue segment position needs a closed list with a named owner, usually in the trading team rather than in marketing.

Most stores need four or five values: best sellers, new arrivals, a hero designation, the long tail, clearance. A list that has grown to fifteen almost always means somebody has started encoding products rather than segments, at which point the position stops being a grouping and becomes a second product identifier that nobody maintains.

Where a campaign genuinely covers several segments, give it a value that says so rather than a compound token. A campaign named for two segments cannot be attributed to either, and writing mixed in the position is a more honest record than pretending otherwise.

The promotion position identifies which promotional window a campaign belongs to. It should name the window — the season, the event, the drop — and not the offer inside it.

The reason is comparability. A token containing a percentage or a dollar amount produces a new value every time the offer is tuned, so a store ends up with a dozen tokens for what is really one sale, and the year-on-year question that the position exists to answer becomes impossible.

Full-price trading needs a value too, rather than an empty position. A campaign with no promotion token is indistinguishable from one where somebody forgot, and an explicit value removes the ambiguity.

The third addition is a creative set token on the ad level, and it is the segment that connects the account to the production process.

Give it the same identifier that appears on the creative brief row and on the asset file. Once those three agree, a question like which of this season's shoots is actually working becomes answerable without anybody opening a folder, and a fatigued set can be identified and replaced as a set rather than ad by ad.

This is also the position that makes a test readable afterwards. A test identifier in the creative set token, matching the row in the test plan, is what lets a finding be traced back to the assets that produced it.

A store's account turns over faster than most, and the temptation to rename campaigns each season is strong: the same prospecting campaign, renamed for spring, then for the sale, then for summer.

Renaming breaks the history in most reporting, and it also hides something worth seeing. A campaign that has carried four seasons and three offers is not one campaign whose performance can be trended; it is four campaigns in a trench coat. Creating new campaigns per season produces a cleaner record and lets the old ones be compared with their successors.

Where renaming has already happened, record the old name, the new name and the date in the notes so that two exports from either side can still be joined.

Store campaign names get long, because there are more segments to carry, and several surfaces truncate.

Check the limit on the platforms you actually run, and order the positions so that the most important ones survive truncation: the channel and the segment early, the flight period late. Where a name is regularly being cut, the fix is shorter tokens rather than fewer positions — a four-character segment code is fine if the meaning column explains it.

Write the truncation limits in the notes, because the person building campaigns in six months will not know them.

The positions in a campaign name are only half of the benefit. The other half arrives when the same tokens appear in the tracking parameters, because that is what joins advertising spend to whatever the store measures downstream.

Build the parameters from the same values: the channel, the segment, the promotion window, the creative set. Then a question about which segment produced which orders can be answered in the store's own analytics rather than only in a platform's reporting, which matters because the two rarely agree.

The common failure is that parameters are built by a different person at a different moment, with slightly different values — plural where the campaign name is singular, a different token for the same segment. Put the parameter format in the where-it-appears column for every position, and check one live link before a season starts.

Seven checks. One separator everywhere, never inside a value. The catalogue segment list matches what the trading team actually uses, and it has an owner there. The promotion token names the window rather than the discount, and full-price trading has its own value. The creative set token matches the creative brief and the asset file. Every level has one example name that splits correctly. Names are set once, with renames recorded. And the truncation limits of your platforms are written down.