Start the report with the decision you need to make
A founder needs to decide whether to repeat a directory placement, maintain a partner resource, improve outreach, or stop a weak experiment. A report should provide enough evidence for that decision. Start each campaign with its audience, destination, deliverable, owner, cost, and review date. Then agree what a useful visitor action would look like.
For an API tool, activation might mean completing a successful test request. For a workflow product, it might mean finishing a first project. Choose a definition relevant to the product before the campaign runs. Otherwise, the team can unconsciously switch from customers to sign-ups to impressions whenever a different number looks more encouraging.
Separate four layers of measurement
The same campaign can succeed at producing useful coverage while still requiring work on acquisition or onboarding. These layers help identify where the next improvement belongs.
| Layer | Record | Decision it supports |
|---|---|---|
| Delivery | Live source URL, publication date, content scope | Was the agreed work completed? |
| References | Live backlinks and distinct referring domains | What references exist and remain accurate? |
| Visits | Observed referrals to the intended destination | Does the placement introduce relevant people? |
| Product use | Qualified sign-ups, activation, customers | Do those people benefit enough to continue? |
Keep observations and estimates visibly different
Your analytics can report recorded visits; an SEO tool may estimate a publisher’s search traffic. Those are different kinds of information. Label the provider, market, date, scope, and whether a number is measured or modeled. An estimate is useful context for prospect research, but it is not a count of people who saw your particular listing.
Search Console’s Links report is a sample rather than a complete inventory. Keep an independent record of important published sources so the absence of a row in that report does not erase delivered work. For broader search performance, use a consistent view of pages, queries, countries, and dates when making comparisons.
Calculate costs with the denominator stated
Suppose a fictional experiment costs $120 in publication fees and six hours of operator time valued internally at $20 per hour. Its total working cost is $240. If it produces ten qualified sign-ups and four activated accounts, the observed costs are $24 per qualified sign-up and $60 per activated account. These are illustrative figures, not a benchmark or a ShowMeYourSaaS result.
If one of those accounts becomes a paying customer, the observed cost per new customer in that small example is $240. Keep the tiny sample and observation period visible. It would be premature to turn one customer into a reliable long-term acquisition forecast. Different denominators describe different stages; avoid presenting the cheapest one as a general acquisition cost.
Use campaign tags and referral context carefully
Where a publisher supports campaign parameters, agree a consistent source, medium, and campaign name. Preserve a record of the final link and check that the landing page opens correctly. Also review ordinary referral sources, because some publishers use their own parameters and some visitors return directly on another day.
A campaign tag helps describe a visit; it does not prove that the last recorded click caused the entire buying decision. A person may discover a directory profile, watch a demo, ask a colleague, and return later. Use attribution as a practical reporting convention, and supplement it with customer conversations when a decision matters.
Turn the review into one next action
Review whether the agreed material is live, whether the right people arrive, and where they stop. Correct inaccurate profiles when delivery is incomplete. Improve the destination when visitors cannot find the promised example. Investigate onboarding when relevant sign-ups fail at the same setup step. Each problem calls for a different response.
Keep the report small enough for an operator to maintain every month. A source record, a few meaningful outcome columns, and a short decision note are often enough to support the next experiment. Add complexity only when a real unanswered question requires it.