Setting a publishing cadence between editions that keeps a show brand alive

The show closes on a Thursday in March. The team publishes four recaps over the following fortnight, an award winners piece, and a save-the-date. Then the site goes quiet until the middle of January, when somebody remembers that registration opens in three weeks and asks the agency for a content push.

That is the default publishing cadence between editions for a large share of B2B show brands, and it produces a site that is functionally new every year. The audience has to be reacquired, the search positions have to be rebuilt, and the crawler has to be persuaded to come back.

What actually happens to a site that stops publishing

Google's own Search Central documentation on managing crawl budget, as it stands in 2026, is unusually direct about how recrawl frequency is decided. It states that demand "varies based on a site's size, update frequency, page quality, and relevance, compared to other sites", and that "URLs that are more popular on the Internet tend to be crawled more often to keep them fresher in our systems". On staleness it says the systems "want to recrawl documents frequently enough to pick up any changes".

Read that as an operating constraint. Update frequency is one of four inputs Google names, and it is the only one an editorial team controls week to week. A site that publishes nothing for eight months is telling the recrawl system that returning often has no payoff, and the system responds accordingly.

The reader half of the problem is less measurable and probably larger. A trade audience checks a small number of sources on a rough weekly rhythm, and a source that produces nothing for two quarters drops out of that rhythm without anyone deciding to drop it. When the site starts publishing again in January, the returning-visitor share of its traffic is low, because the people who used to return have stopped checking. You can see this in your own analytics by comparing the returning share in the month registration opens against the month after the show closes.

The same documentation gives thresholds worth knowing, described in the original as a rough estimate rather than a rule. Crawl budget becomes a real concern for large sites of a million or more unique pages changing about weekly, and for medium sites of ten thousand or more pages changing daily. Most single show brands sit well below both, which is the point. Your problem is rarely that Google cannot crawl everything you publish. Your problem is that it has stopped checking.

How often should a show brand publish between editions?

Two pieces a week is the floor I would defend, and the reasoning is arithmetic rather than doctrine.

A year has 52 weeks. Take out the eight weeks around the show, when output is high and driven by the event itself, and 44 weeks remain. At two pieces a week that is 88 pieces in the quiet period, plus whatever the show period produces, so roughly 110 to 120 a year. One full-time editor commissioning, editing and publishing can carry that with freelance writers. One writer producing all of it cannot.

Below two a week the gaps start to matter. At one piece a week, a single sick fortnight and a Christmas break turn into a five-week gap, and the pattern the crawler sees is indistinguishable from abandonment. At two a week, the same absences leave a gap of two or three weeks that the following month repairs.

The rate matters more than the total. A brand publishing 100 pieces in six bursts looks less alive than one publishing 80 at a steady weekly rhythm, because habit is built on predictability and so is recrawl scheduling. If your editorial calendar for the quiet months exists as a list of pieces with no dates against them, you have a total rather than a cadence, and the year round content calendar problem is the one to fix first.

What does the crawl stats report actually tell you?

Stop guessing about crawl frequency. Google's 2026 documentation for the Search Console crawl stats report describes what it holds: total crawl requests, total download size and average response time, broken down by response code, file type, Googlebot type and crawl purpose. Crawl purpose splits two ways: discovery, meaning a URL Google has not fetched before, and refresh, meaning a return visit to a URL it already knows.

Turn that into a rate per URL, because the raw request count moves with site size and tells you little on its own.

Say your show media site has 2,400 indexable URLs. In the eight weeks around the edition, the report shows an average of 480 crawl requests a day. Over 56 days that is 26,880 requests, which is 11.2 requests per URL, or one visit per URL every five days.

Now take the dormant quarter. Average daily requests have fallen to 130. Over 90 days that is 11,700 requests, or 4.9 per URL, one visit every 18 days. The ratio between the two periods, 11.2 against 4.9, is a little over two to one, and that ratio is the number to track edition over edition. If it widens to five to one, your quiet period has gone quiet enough that new pages take weeks to be seen at all.

The discovery share is the second thing to read. A site publishing two pieces a week generates a small but continuous stream of discovery crawls. A site publishing nothing generates almost none, and the report makes that visible without anyone having to argue about whether the site feels neglected.

Why the crawler mix stopped being one thing

There is a second reason cadence has become worth arguing about, and it has nothing to do with rankings.

Cloudflare published an analysis of crawler traffic across its network in July 2025 covering the year from May 2024 to May 2025. Googlebot's share of combined AI and search crawler traffic rose from about 30 per cent to about 50 per cent over that year, with raw Googlebot requests up 96 per cent, against overall crawler traffic growth of 18 per cent. GPTBot reached 7.7 per cent of all crawling traffic in the same analysis, having grown 305 per cent in request volume.

Two consequences follow for an event media brand. The first is that the crawl your site receives is now shared between systems with different purposes, and the ones building answer sets are fetching a lot more than they were. The second is that a site with a slow refresh cycle is stale in more places at once. A show whose site last changed in April will be described by an answer engine in November using April's agenda, April's speaker list and April's ticket prices.

What you allow those crawlers to fetch is a separate decision with real trade-offs, and it belongs with an ai crawler access policy rather than here. The cadence argument holds whichever way that decision goes.

Making the floor survive December

A cadence fails in the same two places every year. The first is the month after the show, when the team is exhausted and the recaps have run out. The second is the December period, when the venue is quiet and half the desk is on leave.

Bank against both. Commission four to six pieces in the fortnight before the show that have nothing to do with the show, hold them, and schedule them for the weeks when the desk has nothing. The subjects that work here are the ones with no date attached, and the ratio of those to timely pieces is worth setting deliberately, which is the evergreen and timely content mix question.

The second protection is a standing slot. One recurring format, published the same day each week, with a named owner and a template that removes the commissioning decision. A monthly numbers piece from your own registration data, a supplier or product roundup, a five question interview. The format does the work that enthusiasm was doing, and enthusiasm is the thing that runs out in November.

Cost the slot before you promise it. A five question interview takes about an hour to commission and chase, twenty minutes to edit, and fifteen minutes to publish and check, so call it two hours all in. Two of those a week is four hours, which is half a day of one person's time, every week, for a year. Say that number out loud in the meeting where the cadence gets agreed, because a cadence agreed without an owner and a half day reserved is a cadence that lasts until the first week somebody is at a show.

The third protection is a rule about what happens when the slot cannot be filled. Publishing the piece late on Thursday is better than skipping it, and skipping it deliberately with a note in the calendar is better than letting it slide unremarked, because the slide is what turns one missed week into a quarter.

Where this stops

Cadence does not cause traffic, and the correlation between publishing rate and organic performance is easy to read backwards. Sites that publish steadily tend to be sites with a funded editorial desk, an audience that already exists and a subject with continuous search demand. The cadence is one visible feature of that arrangement.

The failure mode is treating the floor as a quota. Two thin pieces a week, written to hit a number, produce a site full of pages nobody reads, and Google's spam policies treat pages generated mainly to have pages as scaled content abuse. If the honest choice is between two weak pieces and one good one, publish the one and let the crawl rate drop. A shorter cadence with a real reader is recoverable. A hundred pages of filler is a cleanup project, and how to run that cleanup sits with the content and media work on the content studio side.

There is also a size effect. On a brand with 300 URLs, crawl frequency is rarely the binding constraint, and the argument for cadence rests on reader habit and on being present when somebody searches for your subject in August. On a portfolio site with 20,000 URLs, the crawl argument does real work.

Open the crawl stats report for your show's media site this week and write down two numbers: average daily crawl requests for the eight weeks around your last edition, and the average for the quietest 90 days that followed. Divide each by your indexable URL count. The gap between those two figures is the cost of the silence, measured on your own site, and it takes twenty minutes.