The Attribution Model I Actually Trust at Small Scale
Run last-touch attribution with a manual spend-per-channel spreadsheet, reviewed weekly, until you're spending past $30K a month on paid media. That is the honest answer. Not a platform. Not a model with a name. A habit.
The short answer
Weekly, by hand, last-touch. I scaled a DTC brand from $100K to $3M in revenue using exactly this. Every Monday I pulled spend by channel, revenue by last-click source from the store's own analytics, and put both in one tab. No multi-touch model, no attribution software, no data layer debate. It worked because at that revenue size you don't have enough conversion volume for a fancier model to be statistically meaningful. A multi-touch model needs thousands of conversions a month to say anything true. Most small businesses have dozens. Fancy math on thin data just produces confident nonsense.
What changes the timing
Four things push you off last-touch and toward something more complex, earlier or later than that $30K mark:
- Number of paid channels running at once. If you're only on Meta and Google, last-touch is fine. Add TikTok, affiliates, and a retargeting stack, and last-touch starts lying to you about who deserves credit.
- Sales cycle length. A DTC impulse buy converts in one session. A B2B deal with a 60-day cycle touches six channels before close. Longer cycle, sooner you need something beyond last-touch, regardless of spend level.
- Whether a human sales team is involved. On the $2.2B infrastructure project I marketed, the buyer journey ran through PR, events, direct outreach, and a sales team over 18 months. Last-touch would have credited a single email opened four days before signing. Useless. We tracked touchpoints manually in a CRM instead of trusting any digital attribution tool, because the real influence was a relationship, not a click.
- How much of your traffic is branded search. If 40% of your conversions come through people typing your brand name into Google, last-touch will over-credit search and under-credit whatever originally built the awareness. That's a sign you need at least a first-touch column next to your last-touch one, even at small scale.
Signs you are overdue
You'll feel this before you can prove it with data. Watch for:
- You cut a channel because "it's not converting," then sales drop two weeks later with no other explanation.
- Two people on your team argue about which channel "really" drove a big month, and neither can point to a number.
- Your cost per acquisition on your best-performing channel keeps climbing even though nothing changed in the campaign itself.
- You've never once looked at what channel someone saw first, only what they clicked last.
- Your CFO or founder asks "what's actually working" and you answer with a feeling, not a spreadsheet.
What happens if you wait too long
The cost isn't attribution theory, it's budget. I've seen founders kill a top-of-funnel channel like organic social or PR because it shows zero last-touch conversions, then wonder six months later why their paid search costs doubled. The paid search wasn't getting better. The top of funnel that used to feed it just disappeared.
On that $2.2B project, if we'd used last-click reporting to justify budget, we'd have defunded the entire PR program. It never showed up as the last touch before a signed contract. But when we ran a simple pre/post test, pausing PR spend in two regions for one quarter, qualified pipeline in those regions dropped 22% compared to the regions where PR kept running. That's not a fancy multi-touch model. That's a controlled comparison, which is often more honest than any attribution software output.
Wait too long to build even a basic weekly view, and you'll make cuts based on what's easy to measure instead of what's actually working. That's the real cost: not bad data, bad decisions made confidently on incomplete data.
The most common mistake
The most common mistake I see is small businesses buying attribution software before they have attribution discipline. A $500-a-month tool doesn't fix a team that never looks at the numbers. I'd rather see a founder with a Google Sheet they check every Monday than a dashboard nobody opens after week two.
The second most common mistake is treating attribution as a one-time setup instead of a habit. Attribution isn't a project you finish. It's closer to bookkeeping. You do it on a schedule or it stops meaning anything.
Start simple. Last-touch, tracked weekly, by hand if you have to. Add a first-touch column once branded search or a long sales cycle makes last-touch start lying to you. Only reach for multi-touch modeling or paid attribution software once you've got the conversion volume, at least a few hundred a month, to make the math meaningful. Everything before that is just expensive guessing dressed up as precision.