Marketing Attribution Models: First-Touch vs Last-Touch vs Data-Driven (What's Accurate?)

Why Attribution Models Matter for Budget Decisions

Attribution models are not a theoretical exercise — they directly determine where you spend money. An Indian ecommerce business using last-click attribution will see Google Shopping as its top-performing channel because most final purchases click through a Shopping ad. The same business using first-touch attribution will credit the blog post or social media post that introduced the customer, making content and social look more valuable. Linear attribution will distribute credit evenly, making every channel look moderate. The same marketing programme, three completely different performance pictures. This is why two marketers looking at the same data can reach opposite conclusions about which channel to scale. Google's own research found that advertisers using data-driven attribution (DDA) versus last-click saw a 6% improvement in conversions at the same ROAS — because DDA more accurately credits the upper-funnel touchpoints that drive consideration, preventing brands from over-investing in bottom-funnel channels and starving the top of funnel that feeds them.

• The same marketing programme can look completely different under different attribution models
• Last-click typically over-credits brand search, retargeting, and Shopping — all bottom-funnel channels
• First-touch typically over-credits organic search, social, and display — all top-funnel channels
• Advertisers using data-driven attribution saw 6% more conversions at same ROAS vs last-click (Google)
• Attribution model choice directly affects budget allocation — it is a strategic decision, not a technical one

Last-Click Attribution: The Default That Misleads

Last-click attribution assigns 100% of conversion credit to the last touchpoint before the conversion. It is the simplest model and was the default in Universal Analytics. Its appeal is intuitive: the last click 'closed' the sale. Its flaw is equally clear: it ignores every touchpoint that built awareness, consideration, and intent before that final click. In practice, last-click systematically over-credits brand keyword search (because buyers searching 'Brand Name + buy' are already decided), retargeting ads (because they reach the most conversion-ready users), and Shopping ads. It under-credits display, social, content, email, and any other channel that introduces the brand or nurtures consideration. For Indian D2C brands running Facebook and Instagram campaigns alongside Google Ads, last-click attribution almost always makes Facebook/Instagram look unprofitable — the social channels drive discovery and consideration, but the customer converts days later via a Google search, and last-click gives all credit to Google. This attribution flaw has led many Indian brands to mistakenly cut social media budgets that were actually driving their Google conversions.

• Over-credits: brand search, retargeting, Shopping, direct — all bottom-funnel close channels
• Under-credits: display, social, content marketing, email — all discovery and consideration channels
• Common mistake: cutting Facebook/Instagram because last-click shows zero or low ROAS when social is driving Google conversions
• Last-click is still GA4's default for some reporting views — check your attribution settings explicitly
• Best use: for high-intent, single-session purchase categories where most buyers do convert on first visit

First-Touch Attribution: Understanding Acquisition Sources

First-touch attribution assigns 100% of conversion credit to the first touchpoint that brought the customer into your funnel. Its strength is revealing which channels are best at acquiring new customers — especially useful for top-of-funnel investment decisions and brand awareness measurement. Its weakness is the mirror image of last-click: it ignores everything that happened between acquisition and conversion. For a customer who discovered your brand via an Instagram post, read a blog post, received two emails, clicked a retargeting ad, and then converted via Google Search, first-touch gives 100% credit to Instagram and zero to everything else. First-touch is most useful as a supplementary view rather than a primary attribution model. Indian businesses can use it specifically to answer the question: 'Which channels are best at generating new brand awareness and first-time visitor acquisition?' This is valuable for measuring the incremental value of content marketing and social media campaigns that would look worthless under last-click.

• Over-credits: organic search, social media, display prospecting — all acquisition channels
• Under-credits: retargeting, email, brand search — all channels that close already-warm leads
• Best use: measuring which channels bring in the most new customers (acquisition efficiency)
• Useful supplement when evaluating content marketing ROI and social media brand awareness campaigns
• Never use first-touch as your primary optimisation signal for paid media budget allocation

Data-Driven Attribution: The Most Accurate Modern Model

Data-driven attribution (DDA) uses machine learning to analyse all the touchpoints in converting and non-converting paths and assigns fractional credit based on each touchpoint's actual contribution to conversion probability. Unlike rules-based models (first-touch, last-touch, linear), DDA measures incremental contribution — if removing a touchpoint from the path would have significantly reduced conversion probability, that touchpoint receives more credit. Google's DDA model requires a minimum of 400 conversions in a 30-day period to function — below this threshold, GA4 defaults to last-click. When available, DDA is the most reliable model for optimising paid media spend. Google's internal research across thousands of accounts found that switching from last-click to DDA and using DDA signals in Smart Bidding produced an average 5-8% improvement in conversion volume at equivalent spend. For Indian advertisers, the 400-conversion minimum means DDA is available to mid-size and larger accounts but unavailable to smaller accounts that must rely on rules-based models.

• Uses machine learning to assign fractional credit based on incremental conversion contribution
• Requires 400+ conversions in 30 days to activate in GA4 — unavailable to low-volume accounts
• Produces 5-8% more conversions at equivalent spend when used with Smart Bidding (Google internal data)
• Updates daily as new conversion data flows in — adapts to seasonality and changing user paths
• Default attribution model in GA4 for Conversions report when data requirements are met

Linear, Time-Decay, and Position-Based Models

Three additional rules-based models serve specific use cases. Linear attribution distributes equal credit across all touchpoints in the conversion path — useful for getting a complete picture of all channels contributing without biasing toward first or last. Time-decay attribution gives more credit to touchpoints closer in time to the conversion, on the theory that recent interactions were more influential. It is more nuanced than last-click but still biases toward bottom-funnel. Position-based (U-shaped) attribution gives 40% credit to the first touchpoint, 40% to the last, and distributes the remaining 20% evenly across middle touchpoints. It recognises the importance of both acquisition and closing while acknowledging mid-funnel nurturing. For most Indian businesses, the practical recommendation is: use data-driven attribution if you have 400+ monthly conversions, use position-based as the default if below that threshold, and run last-click as a supplementary view to understand closing-channel performance. Running multiple models in parallel in GA4's Attribution Comparison report is the most sophisticated approach.

• Linear: equal credit to all touchpoints — good for complete channel visibility, no performance optimisation
• Time-decay: more credit to recent touchpoints — more nuanced than last-click, still biases bottom-funnel
• Position-based (U-shaped): 40% first, 40% last, 20% distributed — balanced for lead gen businesses
• GA4 Attribution Comparison tool: compare models side-by-side in Advertising > Attribution > Model Comparison
• Recommended default for <400 conversions/month: position-based for balance of acquisition and closing credit

Setting Up Attribution Correctly in GA4

GA4 uses data-driven attribution as its default reporting model when data requirements are met. Configuring it correctly requires several steps. First, verify your attribution model in Admin > Attribution Settings — ensure it is set to Data-Driven (or your preferred model). Second, check your Conversion window settings — GA4's default is a 30-day conversion window for purchases and 7 days for other events, but B2B businesses with longer sales cycles should extend this to 90 days to capture late-converting users who were influenced weeks before converting. Third, ensure all significant marketing channels are tagged with UTM parameters — untagged traffic appears as Direct, which inflates direct attribution and under-credits your actual marketing channels. In India, WhatsApp links shared without UTM parameters are a common cause of attribution gaps. Fourth, use GA4's Advertising workspace to compare attribution models and understand the delta between last-click and data-driven for your specific channel mix. The channels with the biggest positive delta in DDA versus last-click are the ones your last-click reporting was most systematically under-valuing.

1. 1Admin > Attribution Settings: verify attribution model is set to Data-Driven or your preferred model
2. 2Extend conversion window to 90 days for B2B businesses with long consideration cycles
3. 3Audit UTM parameter coverage: all paid, social, email, and WhatsApp links must be tagged
4. 4Use GA4 Advertising > Attribution > Model Comparison to identify which channels are under/over-credited
5. 5Set up GA4 Explorations > Path Exploration to visualise multi-touch conversion paths

Attribution in Practice: Making Budget Decisions

Attribution data should inform budget decisions but not mechanically drive them. The right process is: use your primary attribution model (ideally DDA) as the main performance signal, cross-reference with last-click to understand closing-channel performance, and use incrementality testing (Google's Conversion Lift experiments, Meta Lift Studies) to validate attribution estimates with causal data. Incrementality testing holds out a percentage of your target audience from seeing a specific ad and measures whether conversions change — this provides a true measurement of causal impact, something no attribution model can do with certainty. For Indian businesses that cannot run formal incrementality tests, a practical alternative is budget hold-outs: pause a channel for 2-4 weeks and measure whether total conversion volume changes. If pausing Facebook for 4 weeks while maintaining all other channels produces no change in total leads, Facebook's attributed contribution was inflated. If total leads drop by 20%, Facebook had a real incremental contribution that attribution models may have been under-reporting.

• Use DDA as primary signal, last-click as secondary — do not rely on one model alone
• Incrementality testing (Google Conversion Lift, Meta Lift Studies) provides causal data attribution cannot
• Budget hold-out test: pause one channel for 2-4 weeks and measure total conversion impact
• Channels with high DDA credit but low last-click credit are likely under-funded relative to their true contribution
• Never make channel cuts based on one attribution model's data without cross-validating with another model