How to Tell Whether Social Media Is Doing Anything for Your Business

"Is this actually doing anything?" is the right question and the hardest one. Before you answer it, you need to know that both of the instruments available to you are broken in opposite directions.

Your website analytics undercounts social

Google Analytics defines its Direct channel as "sessions in which the user typed the name of your website URL into the browser or came to your site via a bookmark." In practice, that bucket also collects traffic that arrives with no referrer information, which includes links opened from messaging apps, some in-app browsers, and pasted URLs.

That is the phenomenon Alexis Madrigal named dark social in The Atlantic in 2012, reporting Chartbeat figures where "almost 69 percent of social referrals were dark." That number is old, comes from one vendor's set of media sites, and has no published sample size, so use it as an illustration of the mechanism rather than a current statistic. The mechanism is not in dispute: the most valuable kind of sharing, one person privately recommending you to another, is close to invisible in referrer-based analytics.

There is a second problem in the same tool. Google's documentation on the "paid and organic last click" model says it "ignores direct traffic and attributes 100% of the key event value to the last channel that the customer clicked through." A model that discards direct traffic and gives everything to the final click will always make search look strong and always make the slow work of becoming known look worthless.

Platform dashboards overcount social

Meanwhile the platforms grade their own homework, and the peer-reviewed evidence says the grading is unreliable.

Gordon, Zettelmeyer, Bhargava and Chapsky compared randomized experiments with the observational methods advertisers normally use, using "data from 15 US advertising experiments at Facebook comprising 500 million user-experiment observations and 1.6 billion ad impressions." Their conclusion: "commonly used observational approaches based on the data usually available in the industry often fail to accurately measure the true effect of advertising." Two of the authors worked at Facebook.

Lewis and Rao, studying 25 large field experiments representing "$2.8 million in digital advertising expenditure," found "the median confidence interval on return on investment is over 100 percentage points wide."

And if you want to run the platform's own causal test, Meta's Conversion Lift documentation requires a campaign "with a spend of $5,000 USD or more" and "a minimum of 500 optimized conversions." Below that, the rigorous answer is simply unavailable.

So: your analytics hides it, the platform's dashboard flatters it, and the honest experiment is priced beyond most small budgets. That is the real situation. Anyone who tells you otherwise is selling a dashboard.

What you can actually do

You cannot run a randomized controlled trial. You can do considerably better than vibes.

1. Pick one business outcome, in advance. New enquiries, booked consultations, orders, quote requests. One number, defined precisely, that you already track for other reasons.

2. Write down the baseline before you start. The last three months of that number, month by month. Without this, you will reinterpret history to match whatever you hope.

3. Add a "how did you hear about us" field. Self-reported attribution is biased and incomplete, and it is still the single most informative measurement available to a small business, because it catches the paths analytics cannot see. Keep the options short and let people type an answer.

4. Tag the links you control. UTM parameters on the links in your bio, your posts, and your newsletter cost nothing and rescue some of what would otherwise land in Direct.

5. Log what you published. A one line record per post: date, platform, topic, format. In three months this is the only dataset that will let you see a pattern.

6. Compare over a season, and account for seasonality. YouTube's own creator documentation names seasonality as one of three external factors that change how many people see your work, alongside topic interest and competition from other creators. Your business has the same. Compare this March against last March, not against December.

7. Watch second-order evidence. People mentioning a specific post when they call. Sales conversations that start further along because the person already understood what you do. Referrals from people who have never bought from you. None of these appear in a dashboard, and all of them are what "it is working" actually looks like early.

What good evidence looks like at small scale

You are looking for a change large enough to see without statistics. If your enquiries went from 4 a month to 5, that is noise. If they went from 4 to 12 and stayed there for three months, and half of the new ones mention seeing your posts, that is a business result even though no attribution model will confirm it.

Small businesses are lucky in one specific way here: you talk to your customers. That is a measurement instrument most large advertisers would pay a fortune for.

The exit criteria nobody writes down

Decide, in advance, what would make you stop. For example: six months of consistent posting on one platform, with no measurable change in the outcome number and no customer ever mentioning it. If you hit that, stop, and move the hours somewhere else.

Write those criteria before you start. Otherwise you will keep going out of guilt, which is the actual reason most small businesses keep posting into a void.

Two honest cautions

Attribution is not the same as contribution. A customer who found you through Google after seeing three of your posts will be recorded as search. The published research is unanimous that this kind of misallocation is normal, not exceptional.

A platform can work and still not be worth it. If Instagram brings you two customers a month and costs you eight hours a month, the question is what else those eight hours could do. Compare against the alternative, not against zero.

The version that tends to survive this test is not a heroic content strategy. It is a small, repeatable habit attached to something that makes it happen: one hour a week, a real reason to publish, and people who notice. That is the thing Ripple Room actually sells, and it is worth measuring the same way you would measure anything else here.

When this was last confirmed

Google Analytics documentation was retrieved in August 2026, and Meta's Conversion Lift requirements from an archived snapshot of its help page dated January 2025. The two academic papers are peer reviewed, dated 2015 and 2019, and address measurement rather than any specific platform. The dark social article is from 2012 and is cited here for the mechanism it describes, not for its number.