Guides
Part of Campaign management tools: what they take over, step by step
Six things a campaign management case study needs to prove
A campaign management case study proves something only when it names the constraint, the cost paid and the part that failed. Here are six tests to apply.
Most published case studies about campaign software are unreadable as evidence. They report time saved, attach a percentage nobody can check, and omit the accounts where the same rollout was abandoned.
Six things have to be present before a case study tells you whether a tool will work in your agency. The six below are the tests. The four situations that follow are composites, labeled as such, showing how each test fails in practice.
What to take away
- A case study is evidence only when it names the constraint. Without one it is a testimonial with a chart on it.
- The useful accounts are the ones where the tool was the right purchase and the rollout still failed.
- Ask any vendor one questionwhat would have had to happen for this to be written up as a failure?
- Test the export before you need it. An export is only as complete as the objects the vendor chose to model.
- Review the seat list monthly and remove anyone who has not signed in. The meter and the workload grow on different curves.
The six things, and what each one proves
| # | What the case must show | What it proves | Named example |
|---|---|---|---|
| 1 | The constraint | That the result could transfer to you | Asana's published customer stories state the team size and the workflow replaced |
| 2 | The counterfactual | That the tool caused the gain, not the process rewrite | Salesforce Marketing Cloud case studies name the prior stack |
| 3 | The cost paid | Migration, training, the month of double entry | HubSpot's onboarding notes list data migration as a paid line |
| 4 | The part that did not work | That the account has not been edited into a success | Notion's template gallery shows teams keeping side spreadsheets |
| 5 | The meter | Which line grows when you succeed | ActiveCampaign prices by contact count, not seats |
| 6 | The exit | Whether the record leaves with you | Klaviyo documents what its export covers |
Element five is the one buyers skip. A tool priced per contact bills more as your campaigns succeed; a tool priced per seat bills more as your team grows. Those are different risks, and the case study usually names neither.
Every case below turns on the question that room turns on: where the authoritative copy of a fact lives, and who may change it.
The migration that succeeded and the reporting that broke
A team moved campaign records into a new system, kept the old spreadsheet running for a quarter as a safety net, and then found their quarterly client reports no longer matched. Nothing had been lost. The two systems were counting a campaign as live on different dates, and the report pulled from the older one.
Two Systems, Two Start Dates
Old spreadsheet
- Campaign start date
- Different
- Quarterly report source
- Pulled
- Records lost
- None
- Field ownership
- Unwritten
- Fix
- Paragraph
New system
- Campaign start date
- Different
- Quarterly report source
- Ignored
- Records lost
- None
- Field ownership
- Unwritten
- Fix
- Paragraph
What it turned on: nobody had written down which system owned the fact "this campaign started". The fix was a paragraph, not a feature. Deciding that ownership per field, before cutover rather than after, is the substance of a rollout plan.
The product that fit the work and not the week
An agency chose a tool that matched its process closely and still saw usage collapse within two months. The work had not changed. What had changed was that logging it now took a separate step at a moment when nobody had a spare minute, so people did it in batches on Fridays, from memory.
What it turned on: the tool asked for the record at the wrong point in the day. A system true only once a week is not a single source of truth.
A team will quietly go back to whatever is faster, and no demo produces this failure mode, because in a demo data entry is the task.
The seat count that grew faster than the work
A team bought seats for the people running campaigns, then added client reviewers, then freelancers, then a second reviewer per account because the first one went on leave. Nobody made a bad decision. Each addition was obviously reasonable, and at renewal the total was unrecognizable.
What it turned on: the meter and the workload grew on different curves. Which meters these quotes are built on, and which of them grow when you succeed, is covered in the guide to tool pricing structures.
The habit that prevents it is boring, which is why it is rare: someone reviews the seat list monthly and removes people who have not signed in.
The client who asked for the account back
A retainer ended and the client asked for everything: briefs, approvals, contracts, message history. What came out was a set of exports for the objects the vendor models, and nothing for the conversations, which is where most of the reasoning lived.
What it turned on: nobody had tested the export until they needed it. The obligation was in the client contract and the capability was never in the tool. Both parties were acting in good faith and the record still could not be handed over.
An export is only as complete as the objects the vendor chose to model, and a message thread is rarely one of them. Which objects one of these systems holds, and which parts of the job it never touches at all, is mapped in the guide to campaign management software.
Why the published ones read so well
Case studies are collected from customers who stayed. The accounts that churned are not asked for a quote, and the rollouts that were quietly abandoned never reach the marketing team.
What you read is filtered by survivorship bias before anyone edits a word of it, which is why the pattern across vendor case studies is so consistent and so uninformative.
Three Questions That Turn a Case Study Into Evidence
- Which customers of this size did not renew?
- How did the non-renewing customers differ?
- What did the first ninety days cost in hours?
- Which workflow part remains outside the product?
Three questions turn one back into evidence. Which customers of this size did not renew and how did they differ, what did the first ninety days cost the team in hours, and which workflow part remains outside the product?
A vendor answering all three tells you something; one answering none shows a brochure with a customer's name on it.
The habit of writing your own version of these, honestly, is worth more than reading anyone else's. Keep a short note after each rollout: what you expected, what happened, what you would test earlier next time. Writing a purchase decision as a claim that could turn out wrong is the discipline set out in the guide to comparing suppliers.
Common questions
Are composite cases not just made up?
They are illustrations, and they are labeled as such. The alternative is inventing a named agency and a percentage, which would be worse: a fabricated specific reads as evidence and is not.
How do we tell a real case study from a written one?
Look for a number that is against the vendor's interest. Real accounts contain a cost, a delay or a workaround. Written ones contain only gains.
Should we ask for a reference from a similar agency?
Yes, and ask the vendor to introduce you to one that left. The refusal is informative, and the occasional agreement is the most useful call you will have.
What should we record about our own rollout?
The date, the thing you expected to improve, and the measure you would have accepted as proof. Write it before the tool arrives, because afterwards you will remember having expected whatever happened.
