Institutional and advocacy advertising, the four jobs a company ad does when it is selling nothing, with worked examples

Institutional advertising advertises the organisation rather than any product it sells. Advocacy advertising is the branch of it that argues a position on a contested question. Both are paid for by a company, neither asks anyone to buy anything, and both are judged on whether a specific audience thinks differently afterwards. The four jobs they do are to build a reputation in advance of needing one, to recruit, to hold a licence to operate in front of regulators and communities, and to argue for or against a change that would affect the business. Those four need different copy, different media and different approval chains, and the fourth one is the only one that can put your account in front of an ad platform's election verification process. Below is each job, a worked example of what the ad actually says, and the way each one goes wrong.

Job one, reputation banked before you need it

This is the classic corporate ad. It names the company, says what it does at the scale it does it, and asks for nothing. The purpose is to make the name familiar and the character of the business legible before anybody is choosing a supplier, applying for a job or reading a story about you in a bad week.

What the ad says, worked: a regional construction firm runs a thirty second spot in which a site manager explains that the company has rebuilt the same three schools over twenty years and now employs the children of people it trained. No service list, no phone number, one line of brand at the end.

How it fails: it becomes a montage. The failure mode of reputation advertising is an ad with sunrises, a piano and the word possibility, which is indistinguishable from the ad any other company in the sector could run. The test for this job is whether a competitor could put their logo on your ad without changing a frame. If they could, the money bought nothing.

Job two, recruitment

Institutional advertising aimed at people who might work for you is the version with the clearest payback, because applications are countable. It advertises the employer rather than a vacancy, so it runs continuously rather than when a role opens.

What the ad says, worked: a logistics company advertises that every depot manager in the business started on the yard, names three of them, and links to a page listing the actual progression and the actual pay bands at each step.

How it fails: it describes a culture the first week does not deliver. This is the one job in the four where the audience verifies the claim within days of responding to it, and where the gap between the ad and reality comes back as reviews on employer sites. Pay bands and a named route are durable claims. Words like family and passionate are not.

Job three, the licence to operate

Businesses whose operations affect people who are not their customers run this one. A quarry, a utility, a hospital group, a food processor, a bank after a sector-wide scandal. The audience is a planning committee, a regulator, a local newspaper, a community that lives near the site, and the measure is whether the next hearing starts from a neutral position rather than a hostile one.

What the ad says, worked: a water company publishes its own overflow figures for each named site, the target for each, and the engineering work scheduled against each one, with the date the figures were read, and advertises the page rather than a slogan. The ad is an invitation to a number, not a claim about care.

How it fails: it is run during the crisis rather than before it. An institutional campaign that appears the week after bad news is read as a response to the news, which it is, and it strengthens the story it was bought to soften. The other failure is a claim that is true and unverifiable, because in this job an unverifiable claim performs worse than no claim.

Job four, advocacy, where you argue a position

Advocacy advertising takes a side on a contested question that affects the business. A trade body arguing against a proposed levy. A manufacturer arguing for a change to grid connection rules. A retailer arguing for a change to business rates. The audience is legislators, officials, journalists and the part of the public whose opinion those three read.

What the ad says, worked: a trade association for independent pharmacies buys press and digital advertising setting out the number of its members that have closed in the past year, the specific reimbursement mechanism it says caused it, and the specific change it is asking for, with its own name as the funder at the top rather than at the bottom.

Three rules separate advocacy advertising that works from the kind that becomes the story.

• Name the funder prominently and without being asked. An argument whose sponsor has to be discovered is an argument that will be reported as having been hidden.
• Ask for one specific change. An ad that complains has nothing anyone can act on. An ad that names a clause, a threshold or a date can be agreed with.
• Use your own numbers and date them. This is the category where the opposing side will check, and an undated or sourceless figure is the easiest thing in the world to discredit.

The platform rule that catches advocacy campaigns, and the common misreading of it

Here is the practical distinction most writing on this subject gets wrong, read off Google's own advertising policy on 31 October 2026. Google's definition of a US election ad is specific and narrow. It covers ads featuring a current officeholder or candidate for a federal elected office, a current officeholder or candidate for a state-level elected office, a current officeholder or candidate for a US territory-level elected office, a federal, state or territory-level political party, or "a state or territory-level ballot measure, initiative, or proposition that has qualified for the ballot in its state or territory".

So a corporate advocacy ad arguing for a change to grid connection rules, naming no candidate, no party and no qualified ballot measure, is not a US election ad under that definition. The moment it names a senator, a party, or a proposition that has qualified for the ballot, it is, and then Google's stated requirement applies: "advertisers may run US Election Ads in the United States only if the advertiser completes US Election Ads verification". Google also states that election ads in these regions are subject to disclosure requirements and targeting restrictions, that Ad Grants accounts are not eligible to run them, and that an advertiser wanting to run them from outside the region needs either election ads or advertiser verification for where it is based.

The operational consequence is simple and it has a deadline attached. If there is any chance your campaign will name one of those five things, start verification before the creative is finished, because verification takes time and a campaign that cannot run is worse than a campaign that ran plainer copy. Google notes that most ads are reviewed within one business day with complex reviews taking longer, which is the review step and not the verification step.

Policies differ by platform and by country and they change, so read the current policy of each platform you intend to buy, on the day you plan the campaign. Nothing here is legal advice, and a question about election law, lobbying disclosure or charity campaigning rules in your jurisdiction belongs with a lawyer rather than with a media plan.

What this is not

Two neighbouring things get called institutional advertising and are different purchases.

• Cause-related marketing ties a donation to a purchase, which turns the ad into a sales promotion with its own rules about what you must disclose. Our page on cause-related advertising and what tying a donation to a purchase changes covers that case specifically, and it is the one most often confused with advocacy.
• Charity and nonprofit fundraising advertising asks for money for a cause rather than arguing a position about it. The creative looks similar and the measurement is completely different, since one is judged on donations and the other on whether a decision-maker changed position. Our page on Facebook ads for nonprofits and charities covers the fundraising version.

How to measure an ad that sells nothing

The reason institutional advertising gets cut first in a bad quarter is that nobody set a measure before it ran. Four that work, chosen before the campaign rather than after it.

• A tracked destination. Every institutional ad should point at one page, and that page should be the proof: the figures, the pay bands, the named sites. Then the ad has a countable outcome even when the outcome is not a sale.
• A before and after survey of the specific audience, not the general public. Two hundred responses from the fifty mile radius around the plant is a better instrument than two thousand nationally.
• Applications, for the recruitment job. Count them by source, and count the ones that reach second interview, because volume without quality is the standard failure here.
• Coverage of your argument in the words you used, for the advocacy job. Not clipping volume. Whether the specific change you asked for appears in other people's writing about the issue.

Questions about institutional and advocacy advertising

What is the difference between institutional and advocacy advertising?

Institutional advertising promotes the organisation itself rather than a product. Advocacy is the version of it that takes a side, so the clean way to remember the distinction is that advocacy advertising has an opponent and institutional advertising does not.

Give an example of advocacy advertising

A trade association buying press and digital advertising that sets out how many of its members closed last year, names the specific reimbursement rule it blames, and asks for one named change, with its own name as funder at the top of the ad.

Give an example of institutional advertising

A construction firm running a spot in which a site manager explains that the company has rebuilt the same three schools over twenty years and now employs people it trained, with no service list and no call to action.

Does an issue ad need election verification?

Under Google's US election ads definition read on 31 October 2026, it depends entirely on whether the ad features a candidate or officeholder, a political party, or a ballot measure that has qualified. An issue ad naming none of those is outside that definition. Check each platform's current policy before you buy, and allow time for verification if there is any chance it applies.

Who actually sees this kind of advertising?

A much smaller group than the reach figure suggests, and that is the point. Each of the four jobs has one audience: future customers, future employees, the people who grant your licence to operate, and the people who decide the rule you are arguing about. Media chosen for total reach usually misses all four.

Is it worth it for a small business?

The recruitment job and the licence to operate job are, and they are cheap at local scale. A page with real pay bands and progression, advertised locally, outperforms national brand advertising for a business with eighty staff.

The description of your company that you did not write

Every one of the four jobs above is an attempt to control how an organisation is described when it is not in the room. There is now a version of that happening constantly and automatically. When somebody asks ChatGPT, Gemini, Perplexity, Claude or Google's AI Overviews what a company is like to work for, deal with or live next to, they get back a short characterisation assembled from whatever has been published about it, and that characterisation is doing exactly the job institutional advertising was bought to do, for free, in whatever direction the available text points. It is the cheapest reputation audit available and almost nobody runs it. Ask the question a journalist, a graduate or a planning officer would ask, with your own company name left out of the prompt where the question allows, and read what comes back and what it was built from. AI Knows Us puts those questions to the five surfaces on a schedule and records the answers and their sources, which is the before-and-after measure an institutional campaign has never had.