How to set up a small business social media page and track inquiries
Set up a business page, plan its first month, and follow visits through to inquiries and bookings. A practical checklist and a hypothetical example are included.
Read storyUseful journalism to understand, manage and grow a business.
SearchLast interaction is simple, but it doesn't tell the whole story. Compare four attribution models and choose based on your sales cycle, channels, and data quality.
An attribution model is a rule for distributing the credit of a conversion—such as a sale or a business lead—among the channels and interactions that preceded that action. It serves to organize the reading of the customer journey and compare campaigns; it does not, by itself, prove that a channel caused the purchase.
The difference matters: a report can assign credit to a search, an ad, or an email, but that assignment depends on which interactions were recorded and the rules chosen. Therefore, attribution is a tool for making data-driven decisions, not a perfect measurement of the incremental effect of each channel.
| Model | How it distributes credit | When it can be useful | Main limitation |
|---|---|---|---|
| Last interaction | Assigns 100% to the last recorded interaction before the conversion. | Short purchase cycles or analysis focused on the final contact. | It may undervalue channels that introduced the brand or helped evaluate the offer. |
| First interaction | Assigns 100% to the first recorded contact. | To observe which channels initiate journeys or attract new prospects. | It does not reflect the interactions that helped close the sale. |
| Linear | Distributes credit equally among the recorded touchpoints. | Journeys with several relevant interactions and a need for a simple overall reading. | It assumes all contacts contribute equally, even if their function may vary. |
| Time decay | Gives more credit to recent interactions and less to earlier ones. | Processes where contacts close to the decision seem particularly relevant. | The weighting favors recent activity by design; it does not prove it was more decisive. |
These models answer different questions. Last interaction helps identify what immediately preceded a conversion; first interaction, what opened the journey; linear, how credit is distributed among recorded contacts; and time decay, how that distribution changes when giving more weight to recent activity. None is universally superior.
In a quick purchase, with few contacts and a relatively direct journey, last interaction attribution can be an easy operational reference to maintain. Even so, it is advisable to contrast it with first interaction: if the last click receives all the credit, discovery channels may remain hidden in the report.
When the sale requires consideration, inquiries, or sales follow-up, it is more likely that several contacts will intervene. A linear model offers a simple panoramic view; the time decay model is useful if you want to prioritize, as an analytical hypothesis, the activity closest to the conversion. The choice depends on how the customer buys and what question the business needs to resolve, not just on how many channels it uses.
For example, an SME that sells services through ads, website visits, emails, and sales conversations can compare which channel appears at the beginning and which at the end of recorded journeys. If both models offer different readings, that difference is useful information: it indicates that the result depends on the distribution rule and that it is not advisable to decide the budget looking at a single report.
The quality of the analysis depends on contacts and conversions being recorded consistently. A CRM can centralize lead information and its commercial evolution; web analytics can help record visits and traffic sources. For those data to be comparable:
Attribution helps organize questions such as which channels appear in journeys that end in a sale and which contribute to capturing contacts. To link that reading with ROI, the costs of each channel and a defined commercial outcome are also needed. Counting assigned conversions is not enough to know profitability: a campaign can generate contacts that do not convert into customers, and the models described do not necessarily distribute margin or the future value of the customer.
A prudent practice is to choose a main model according to the sales cycle, keep it stable during the comparison period, and contrast its results with another model. If the channel ranking changes significantly, avoid presenting a single credit distribution as a definitive truth. Use the discrepancy to review the journey, the quality of the record, and the business question you are trying to answer.
In an SME, the best choice is usually the model that the team can feed with reliable data and use consistently. Starting with a simple rule and improving tracking can provide more clarity than adopting a sophisticated scheme with incomplete information.
Set up a business page, plan its first month, and follow visits through to inquiries and bookings. A practical checklist and a hypothetical example are included.
Read storyA practical guide to understanding the technical terms of digital advertising. Learn what DSP, SSP, RTB, CPM, and CTR are to evaluate supplier proposals and optimize your campaigns.
Read storyReview strategy, messaging, design, and contact details to check that your online and offline ads express the same brand, adapting to each channel without losing their identity.
Read story