Guide · Benchmarks and recipes for a new government social account

Your first 90 days running a government’s social media

Whether you have run communications for fifteen years or picked up the accounts last Monday, the same thing is missing on day one: a benchmark. There is no published figure for what a city of 12,000 should expect from a Facebook post, so the number in most people’s heads comes from a neighbouring town’s best week. This guide replaces that with measurement — 1,301,388 Facebook posts from 4,600 U.S. city and county governments, plus 27,365 Instagram and 24,237 X posts, over the twelve months to September 2026. What a typical post earns, which platform repays the effort, and the post shapes that reliably work.

In this guide: Week 1: what normal looks like · The number nobody tells you · Weeks 2–4: show up · Which platform is worth your time · Weeks 5–8: fix the mix · Format, and why it flips · Multi-post campaigns · Four recipes that work · Weeks 9–12: report it · Method

Week 1 — find out what normal looks like for your size

The first mistake is comparing your account to a city ten times your size, or to a brand. Local-government engagement scales with population, and the gap between bands is large. Before you set a single goal, find your row.

Read your row and write those three numbers down. They are your baseline. If you run a village of 1,800 and a post earns 6 reactions, you are above your median, not failing. Ninety days from now you want to have moved the middle number, not chased the outlier.

The number nobody tells you

This is the finding worth passing on to whoever asks you for numbers: a third to nearly half of all local-government posts earn no engagement whatsoever. Not low engagement — zero. No like, no comment, no share.

The instinct after a quiet post is to conclude something is broken — the algorithm, the content, you. Usually nothing is broken. A meeting notice that nobody reacts to has still done its job if the people who needed it saw it. Judge the month, not the post.

Weeks 2–4 — just show up

The clearest pattern in the data is unglamorous: the accounts that post regularly do better than the accounts that post in bursts.

A realistic target for month one is simply a post every week you are at work, at the cadence for your size in the table above. That is one or two posts a week for most small communities. It is a lower bar than most people set themselves, and it is the one that correlates with everything else.

Which platform is actually worth your time

Most governments are told to be everywhere. Almost none have the staff for it, so the real question is where the hour goes. The naive comparison — typical Instagram post earns 11, typical X post earns 2 — is misleading, because the governments we track on Instagram are larger than the ones we track on Facebook. The honest comparison is within the same government: take only the cities that run both, and compare each one against itself.

Facebook and Instagram are close to a coin flip — Instagram wins in 47% of the governments running both. If you have the hour, the second platform is worth it, and the format section below explains what to put there.

X is a different story. The typical local-government post on X earns a single engagement, and 29.9% of them earn nothing at all, against 9.1% on Instagram. Against the same government’s own Facebook page, X loses 93% of the time. That is not an argument for closing the account — it is an argument for changing what you expect from it. X has one job, covered in the recipes below, and it is not reach.

Want to know what your own peers posted?

That is the whole job GovFeeds does. Search every post by U.S. local governments — by topic, state, population band, platform and engagement — and read what actually worked before you write your own version. This is a live search for water main: 28,407 posts.

Take the two-minute tour → Plans from $80/month

Weeks 5–8 — fix the mix before you add volume

Once you are posting consistently, the biggest lever is what rather than how much. Ranking topics by how a typical post performs, against how much of the calendar each takes, shows a clear mismatch.

Meeting notices, agendas and board announcements are the single weakest category we measure — and they take roughly one post in seven. You cannot stop posting them, and you should not. They are statutory, and reach matters more than reaction for a legal notice. But if they are a seventh of your calendar and infrastructure is a twelfth, the correction is obvious: the road closure, the water main, the sidewalk project — the work residents can see out of the window — is the most reliably engaging thing a local government posts, and it is chronically under-posted.

Photos, text and video — and why the answer flips between platforms

This is the section where single-platform advice does the most damage, because the right answer on Facebook is the wrong answer on Instagram.

Read those last two columns together. A typical Facebook video does no better than a plain text post — both earn 2 — but the average video earns more than twice as much as the average text post. That is the signature of a lottery: most videos land flat, a few travel a very long way, and the average is dragged up by the winners. On Facebook, with limited time, take the photographs.

Now the same question for Instagram and X, and note what happens to the video row.

How much does video actually gain you? Not the five-fold the medians above suggest — that gap is inflated by which governments post video, since the ones that do tend to be larger and better resourced. Comparing each government against itself, across the 246 accounts with at least ten image posts and five videos, video beat images in 76% of them, by a median of about 1.6×. Smaller than the headline, and still the most reliable format decision on the platform.

Multi-post campaigns — the shape that actually works

The standard advice is to run campaigns: do not post about the budget hearing once, build a sequence. That is testable. Taking every run of three or more posts by the same government on the same topic inside a week — 51,360 such sequences — the first question is whether posts inside a campaign do better than a government’s one-off posts.

They do not. Campaign posts earn a median of 7.0 against 7.5 for isolated posts, and campaign posts beat solo posts in only 40% of governments. Repetition does not make an individual post perform better. What it does is multiply the number of chances somebody sees the message at all — which is the actual reason to do it.

The useful question is what the sequence costs you. Measuring each post against the opening post of its own sequence:

The entire decline happens between the first post and the second — a drop of about 42% — and then it stops. The eighth post in a sequence does as well as the second. There is no compounding fatigue penalty for staying on a subject; you pay the cost once. And 35% of follow-up posts beat their own opener outright, so “the first one is best” is a tendency, not a rule.

A campaign template: the service-disruption arc

The most transferable sequence in the data is the one governments run when something residents rely on stops working. Providence, Rhode Island ran this over two weeks of a February blizzard, and the shape is worth copying exactly:

Every post answers a different question, which is why the sequence does not read as repetition — and why post five, the resolution, performed as strongly as post one. Note that nothing in the arc is an update for its own sake: each one either tells residents something has changed, gives them something, or asks them for something.

This particular run earned very large numbers, because a blizzard is a very large event. Do not read those multiples as a target. Emergency and disruption communication is judged on whether the people who needed the information got it, not on engagement — the reason to copy this sequence is that it covers everything a resident needs to know, in the order they need it.

Four recipes that work

Every example here is a real post by a real government, chosen because it beat that government’s own median by a wide margin — not because it went viral. These are results an ordinary account can reproduce.

1. Answer the comments you have been getting — in public

Modesto, California had spent months collecting impatient comments about a delayed park. Instead of another progress update, they built a video out of the comments themselves — “Enough with the name, when is it opening”, “It is taking forever to build this park” — and used them as the set-up for the opening announcement. It earned 499 likes against a median Instagram post of 44, roughly 11× their normal.

This works because it proves somebody is reading. The risk people fear — that repeating criticism amplifies it — did not materialise; the comments underneath are congratulatory. If you have a project everyone is impatient about, the impatience is your best opening post.

2. On X, be the operational feed — it is the one thing X is good for

The typical government post on X earns one engagement. The exceptions are almost all the same shape: something happening right now that people need to act on. Titusville, Florida — a city of about 50,000 — posting a bridge closure during the Artemis II launch crowds earned 23× its own median X post.

Note what it is not: not a newsletter, not a ribbon-cutting photograph, not a reshared Facebook post. Road closures, outage updates, weather operations, “this rumour is not true” — that is the X calendar. Judge the account on whether the people who needed it found it, not on likes.

3. Say the thing residents are already saying

Austin, Texas launched a new city brand and it was not universally loved. Their own post opened by acknowledging exactly that — “y’all have had a lot to say about” it — rather than pretending the reaction had not happened. Plain text, no image, about 7× their median X post.

Local-government accounts default to a register nobody actually speaks in. The posts that break out are almost always the ones where a person is audible.

4. Film the work residents can see out of the window

Infrastructure is the highest-performing topic on Facebook and one of the most under-posted. It also happens to be the easiest thing to film. Oxnard, California’s video about resurfacing twelve aging alleys earned 633 engagements against a median of 54 — nearly 12×.

This is the cheapest repeatable win in the guide: a phone video of a crew doing visible work, on the platform that rewards video, about the topic that already out-performs. If you do one new thing in ninety days, do this one.

Weeks 9–12 — put one page in front of your manager

The last piece of a first ninety days is not a post at all. Somebody will eventually ask how it is going, and the honest answer — “we averaged four reactions a post” — sounds like failure to anyone who has not read the table above. Bring the comparison with you.

One page is enough: your posts per week against the median for your population band, your median engagement against the same, and your three best posts of the quarter with what they were about. That reframes the conversation from “is four good?” to “four is the median for a city our size, and here is where we beat it.” Our separate guide on reporting social media to council covers the framing in more depth.

Or have the page built for you

A GovFeeds monthly benchmark is that one page: your volume and engagement against governments your size, month over month, with your statewide rank and the posts that carried the month. Below is a real one.

See a full sample report → See plans

Where these numbers come from

Related reading: What should we post this month?, Should your city turn off Facebook comments?, and Reporting social media to council.