Ask most comms teams how a campaign performed and you will get a number within seconds: impressions, reach, an AVE figure, a follower count that went up. Ask what that number tells you about whether the organisation achieved anything it set out to achieve, and the conversation usually stalls. That gap is the whole problem with communications measurement. It is not that comms people do not care about proving value. It is that the metrics most readily available are the ones least connected to what anyone commissioned the work to do. This is not an argument against measurement. It is an argument against measuring what is easy to produce and calling that accountability.
Why vanity metrics survive despite being useless
Vanity metrics persist for a boring reason: they are cheap to generate, they always go up, and they give a board something to nod at. None of that makes them evidence of anything.
Advertising Value Equivalency
AVE takes the column inches or airtime a piece of coverage occupied and multiplies it by what that space would have cost as paid advertising. It has been formally rejected by the industry’s own measurement standards, the Barcelona Principles, for over a decade, and the reasoning holds: editorial coverage and paid advertising are different products with different credibility and audiences. A sceptical piece that questions an organisation’s claims scores the same AVE as a glowing one, because the metric only measures space, not what the space says or whether anyone believed it. Multiplying that number by three or five to account for editorial credibility, as some agencies still do, is not a correction. It is inventing a number to make an invented number look more serious.
Impressions and reach
Impressions count how many times content could theoretically have been seen, not whether it was read, understood or acted on. A post with 50,000 impressions and a post with 500 can move an audience by identical amounts, or by none at all, depending entirely on who was in that audience and what they did next. Reach answers a media-buying question about potential exposure. It does not answer a communications question about effect.
Raw follower and fan counts
A growing follower count feels like progress because the number only moves in one direction and never contradicts you. It says nothing about who those followers are, whether they include the people the organisation actually needs to reach, or whether a single one of them has changed what they think or do as a result. Purchased or bot-inflated followings aside, even a wholly organic audience of the wrong people is a vanity metric with better provenance.
What unites all three is that they measure activity, not effect. They tell you comms happened. They do not tell you whether it worked.
Outcome measures: metrics tied to objectives
The alternative is not a different set of numbers pulled from a different dashboard. It is a different starting point: define the objective first, then work backwards to what would count as evidence that comms moved it. An objective stated as “raise awareness” cannot be measured honestly, because almost anything can be claimed to raise awareness. An objective stated as “increase the proportion of target policymakers who can correctly describe our position, from a baseline we have measured” can be tested, missed, or hit. The discipline is in writing objectives specific enough to fail.
The metrics that hold up
Share of voice
The proportion of relevant coverage or conversation an organisation commands relative to named competitors or peers, within a defined topic and timeframe. It is useful because it is relative rather than absolute: a quiet month for everyone in a sector is different from your organisation specifically losing ground. It requires agreeing the competitor set and topic scope in advance, otherwise it is trivially easy to define a comparison that flatters whoever is reporting it.
Message pull-through
Whether the specific arguments and framing an organisation is trying to land actually appear in resulting coverage, stakeholder statements, or how commentators characterise the issue. This requires deciding, before activity goes out, what the two or three core messages actually are, then checking coverage against them rather than against a general impression of tone. Pull-through is more labour-intensive than a coverage count, because someone has to read the coverage rather than just count it. That labour is the point.
Sentiment
Whether coverage and commentary are favourable, neutral or unfavourable, tracked over time and against specific events. It is more informative than volume alone, because a spike in negative coverage looks identical to a spike in positive coverage on a chart that only counts mentions. It is also prone to being measured badly: automated tools are unreliable on sarcasm, industry-specific language, and stories where the organisation is mentioned neutrally in a piece critical of something else entirely. Automated scoring is a starting point for a human to check, not a finished answer.
Referral and conversion tracking
Where comms activity is meant to drive a specific action, such as visiting a site, signing up or contacting the organisation, referral tracking shows whether traffic from a specific piece of coverage or channel actually converted. This is the closest comms measurement gets to hard commercial data, because it relies on standard web analytics rather than judgement calls about tone. It is also the metric most often skipped, because it requires UTM tagging or trackable links set up before activity goes live, not reconstructed afterwards.
The honest limits of attribution
None of the above should be oversold. Comms rarely operates as the sole cause of an outcome, and claiming it does is its own kind of dishonesty, dressed in more credible-looking metrics. A piece of national coverage might coincide with a sales uptick actually driven by a product launch or seasonal demand.
The honest position is that comms measurement can usually show correlation, contribution and plausibility, and can sometimes show direct causation where the mechanism is simple and trackable, such as a referral link converting to a sign-up. It can rarely prove that one piece of activity was the sole cause of a business result, and any plan that promises otherwise is promising something it cannot deliver. The right response is to be specific about what a given metric can and cannot claim, and to triangulate: message pull-through plus sentiment plus referral data, read together, is more credible than any one presented alone.
A worked example: mapping objectives to metrics
Take a hypothetical organisation launching a report to shift specialist policy opinion ahead of a consultation. A measurement plan built the right way round looks like this.
| Objective | What would count as evidence | Metric used | What it cannot tell you |
|---|---|---|---|
| Be seen as a credible voice relative to named peers | Coverage share in the specialist trade and policy press over the campaign window | Share of voice against three named peers | Whether the coverage changed anyone’s mind |
| Land the report’s central argument, not just generate coverage of it | The recommendation appears, in substance, in resulting articles and statements | Message pull-through against the two core recommendations | Whether it was accepted, only whether it was heard accurately |
| Avoid the report being read as self-interested | Tone of coverage, tracked against a pre-launch baseline | Sentiment analysis, human-checked | Why sentiment moved, without reading the coverage itself |
| Drive policymakers to read the full report | Traffic to the landing page from coverage links and targeted outreach | Referral tracking via UTM-tagged links | Whether they read it, only that they clicked through |
Each row exists because an objective was set first, not because a number happened to be available.
A checklist before you commission a measurement plan
- Is the objective written specifically enough that it could fail, not phrased as “raise awareness” or “build reputation”?
- Has a baseline been measured before activity starts, so change afterwards has something to compare against?
- Does every metric map to a specific objective, or is it included because it is easy to pull?
- Is AVE anywhere in the reporting? If so, why, and what is it being used to justify?
- Have the two or three core messages been defined precisely enough to check coverage against, rather than judged on general tone?
- Is sentiment being spot-checked by a person, or trusted automated and unread?
- Are referral links set up before activity goes live, not reconstructed afterwards?
- Does the final report state plainly what the data can and cannot prove, rather than implying causation it cannot support?
Getting started
None of this requires elaborate tooling. It requires setting the objective before choosing the metric, resisting numbers that only ever go up, and being willing to report a result that is genuinely mixed rather than dressed up to look unambiguous. That discipline is harder than running a coverage count, which is precisely why so few organisations do it consistently. If your current reporting is mostly volume and you want a measurement plan built the other way round, from objective to metric, email info@comosphere.com.
Recent Comments