Growth Plays: A Plain Definition and Five a Small Content Team Can Run

Here is the short version. A growth play is a written procedure, built to move exactly one number, that you can run again. Something sets it off, a rule decides the response, work gets produced, and one agreed measure settles whether any of it mattered. A campaign ends on a date. A play goes back on the shelf and comes down again next quarter.

This explainer is aimed at anyone who has now heard the phrase in three consecutive planning meetings and would like a definition they can use on Monday. It defines the term, takes a play apart, and closes with five that a small team can genuinely run beside everything else already on their plate.

A growth play, defined

Treat it as a documented, repeatable method for shifting one metric, whether that metric is conversion, retention or qualified pipeline. BlueConic, a vendor in customer data software, describes the term in its explainer as a use case built in advance and oriented to an outcome, which puts customer data to work against one stated business goal. Its version of a play has four moving parts: the data that signals, the logic that decides, the methods that activate, and a framework that measures.

That reading is shaped by what the product does, which is normal and worth noticing rather than holding against it. Stripped back for a content team with no customer data platform, the same structure survives: one outcome, one trigger, one piece of work, one measure.

Why the phrase is everywhere in planning season

Teams now own more tooling than they own hours. The Content Marketing Institute's 2026 B2B research found nine in ten marketers using AI to produce content while fewer than four in ten reported performance improving as a result, which is the shape of a capacity problem rather than a capability one. More output did not translate into more results.

A play interrupts that by demanding two answers before anybody starts: what outcome is this attached to, and what evidence would show it arrived. Most of what sits in a content calendar cannot survive either question, which is exactly why calendars grow rather than shrink.

There is a second benefit, which is protection from your own ambition. Three people running twelve initiatives produce twelve mediocre things and no readable result. Three people running one play produce something you can judge. That failure pattern is laid out in why the most-recommended growth plays waste a small team's quarter.

The four parts of any play

Part one: the signal that starts it

Every play opens with something observable from outside your own enthusiasm. A page losing clicks. A question asked in three consecutive sales calls. A competitor shipping something. The signal exists to answer when, so that nobody has to decide on instinct whether this is the week.

Part two: the rule that decides

A rule turns a trigger into a response without convening anybody. Written out, it reads like this: when a page gives up an agreed share of its annual clicks, it enters the refresh queue automatically. That form matters. A rule that lives only in one person's head is not a play at all, it is that person's instinct wearing a process costume.

Part three: the work itself

The action is whatever actually gets produced: a rewritten section, an email, a new page, a recorded interview. Keep it small enough to finish inside a week. Actions that span a month stop being plays and become projects, which have different failure modes and worse completion rates.

Part four: the one number

One measure, agreed in advance, settles the question. The worked example on BlueConic's page involves a quiz added to busy product pages for a skincare brand, and the vendor reports that inside two months 12% of people who took it had turned into identified profiles with consent attached. Read that as vendor-reported, because that is what it is. Its value here is the structure it illustrates, not the figure: a single intervention, a single number, a closed window.

A play against a campaign, side by side

	Growth play	Campaign
Lifespan	Repeats indefinitely	Has a start and an end date
Target	A single metric	Usually several at once
Starts because	A signal appeared	A date arrived
Accountable	One named person	A team, which often means nobody
Leaves behind	A method you can rerun	A folder of assets

Five plays that fit a team of three

1. The decay refresh. Starts when a page falls against its own figures from last year. Do: correct the facts and swap in current examples. Judge by: clicks to that address two months on. Detection is covered in spotting pages that are quietly losing traffic.
2. The sales-question page. Starts when one question turns up in three separate conversations. Do: publish a straight answer, including the part where you are not the right fit. Judge by: how often a rep forwards it without being asked.
3. The original data release. Starts when you are sitting on figures nobody outside has seen. Do: publish a short report, ungated. Judge by: how many outside pages reference it. Siege Media's guide to B2B PR makes the case for original research and documented customer work precisely because other people want something to cite.
4. The expert interview. Starts when a subject comes up that nobody on the marketing team has lived through. Do: twenty recorded minutes with a colleague who has. Judge by: the proportion of published pieces carrying a named expert's words.
5. The assistant audit. Starts on a calendar date, monthly. Do: put five buyer questions to two assistants and record what comes back verbatim. Judge by: how often you appear in those answers. Picking the five questions gets easier with the reasoning in when a keyword with no reported volume is worth writing for.

Choosing your first one

Most teams reach this point holding four or five candidates, and want to start three of them at once. Rate each candidate out of three against three questions. Could it begin this week, with the people already employed here? Can it be measured using data you can reach today, without buying anything? Does the number it moves connect to revenue rather than to attention? Add the three, take the highest total, and when two candidates tie, choose whichever costs less.

Then defend it from your own enthusiasm, which is the harder part. No second play begins until the first has produced a result, and a disappointing result counts. Record it in three lines: what was done, what moved, what changes next time. By the third play you own a small library that belongs to the team rather than to whoever happened to be running things, and it survives a resignation. The library is the asset. No individual play ever was.

Two caveats worth stating before you commit a quarter. BlueConic suggests giving a play somewhere between a month and three months before stacking another on top, and anything that depends on how search results behave should get the slower end of that window. Beyond that, results shift with market, budget and timing, and nothing described here promises traffic or revenue to anybody. For a read on which plays held up across a full year rather than a quarter, see which growth plays actually moved anything.

Price the fifth play before you put it on a monthly calendar

The assistant audit looks free right up to the second run, when somebody notices that putting five questions to two assistants by hand, reading the replies properly and logging them eats most of an afternoon. The market noticed first. Otterly, a tool built for precisely this job, publishes its rates: the entry tier is $29 a month and watches fifteen prompts daily across ChatGPT, Google AI Overviews, Perplexity and Copilot, with the hundred-prompt tier at $189. Claude, Gemini and Google's AI Mode are sold on top rather than included, so a realistic bill depends on how many engines your buyers actually use.

Hold those two numbers against the manual version before anything goes on a recurring calendar. Fifteen prompts checked every day by software is a different instrument from five prompts checked monthly by a person: one detects movement, the other takes a snapshot. A snapshot is enough for a first run, and it is the cheapest way to discover whether anybody acts on the answer before money starts arriving every month to collect it.

The decay refresh hides a quieter version of the same problem, which is that its signal needs a year of history to compare against. Google's own documentation on the bulk data export is explicit that the tables begin accumulating only once you switch the feature on, and it points you at the API for anything predating your setup. A play whose trigger depends on data nobody has been storing is not runnable this quarter, whatever it scored.

The survey puts the hard part where the measure goes

The research release quoted earlier has a fuller write-up behind it, and the detail moves where a first play should point. The published write-up of the study describes answers from 1,015 B2B marketers collected between late June and mid-August 2025, and measuring content effectiveness was picked by 33% of them, placing it third, behind weak conversion at 40% and a shortage of time, people and budget at 39%.

Read as one problem rather than three, that list explains itself. A team that cannot measure effect cannot say which content prompted an action, so it cannot argue for the resource that would produce more of that content, so it stays busy and unpersuasive. The same write-up reports 89% of respondents using AI content creation tools against 39% who say content performance improved, and 12% who say quality went down. Volume was never the constraint.

This is the argument for the four lines, stated more precisely than usual. A play is the smallest structure that forces a measure to exist before the work starts, which makes it an answer to the third item on that list rather than another initiative queuing behind it.

Where we would overrule our own scoring

The third scoring question above asks whether a play's number reaches revenue rather than attention. For a team's first play, we would ignore our own question. Choose on how fast and how cleanly the number arrives, even where the ceiling on that number is plainly modest.

The reasoning comes out of the figures just quoted. Measurement ranks third among the challenges teams report, which says the binding constraint for most of them is not selecting a high-impact play, it is finishing one and being able to state what happened. A revenue-linked play fails that twice. Its number lands a sales cycle later, by which point three other things have changed underneath it, and the remaining argument is with somebody in finance rather than with the data. The sales-question page, judged by how often a rep forwards it without being prompted, is a smaller prize than a pipeline figure and it is a prize you can collect inside six weeks without anyone's permission. A first play exists to prove the method works in your team. It does not exist to rescue the quarter.

The ordering has a specific blind spot. A team already running a working attribution column, logging outcomes by habit and carrying finished plays behind it should rank by impact, because measurement is not what stops them and our ordering would cost them genuine upside. The same holds where a sale closes in days rather than quarters, which makes the revenue number the fast number as well as the big one. Everybody else gets more from one finished play than from the right play abandoned halfway through.

No measure, no play

The sporting language flatters the work and hides the part that actually helps, which is constraint: a single metric, a closed window, one person answerable. The test is uncomfortable and takes four seconds. Say the measure out loud in one sentence, without hedging and without a conjunction. If that sentence will not form, what you are holding is not a play. It is an idea, and ideas have never been cheaper than they are this quarter.