长尾关键词是SEO的蓝海。我开发了一套系统能从1个种子词自动扩展到1000个长尾词并且评估每个词的竞争度和价值。这篇文章分享完整方案。一、长尾词扩展的方法1.1 搜索建议扩展defexpand_keywords_from_suggestions(seed:str,api_key:str,depth:int2)-List[str]:从搜索建议扩展关键词all_keywordsset([seed])current[seed]for_inrange(depth):next_levelset()forkeywordincurrent:headers{X-API-Key:api_key,Content-Type:application/json}body{q:keyword,hl:en,gl:us,page:1}rrequests.post(https://api.serpbase.dev/google/search,headersheaders,jsonbody,timeout30)datar.json()# 从相关搜索扩展relateddata.get(related_searches,[])forqinrelated:ifqnotinall_keywords:next_level.add(q)all_keywords.add(q)# 从PAA扩展paadata.get(people_also_ask,[])foriteminpaa:qitem.get(question,)ifqandqnotinall_keywords:next_level.add(q)all_keywords.add(q)currentlist(next_level)returnlist(all_keywords)1.2 修饰词扩展MODIFIERS{question:[how to,what is,why does,how does,can you],comparison:[vs,versus,or,alternative,compared to],location:[near me,in [city],local,best in],price:[cheap,affordable,expensive,price,cost],time:[2026,this year,now,today,recent],quality:[best,top,good,bad,review],action:[buy,get,find,download,use]}defexpand_with_modifiers(seed:str)-List[str]:用修饰词扩展关键词expanded[]forcategory,modifiersinMODIFIERS.items():formodinmodifiers:ifcategorylocationand[city]inmod:# 跳过需要替换的continueexpanded.append(f{mod}{seed})expanded.append(f{seed}{mod})returnexpanded1.3 组合扩展defcombine_keywords(keywords:List[str])-List[str]:组合关键词生成长尾词combined[]fori,kw1inenumerate(keywords):forkw2inkeywords[i1:]:combined.append(f{kw1}{kw2})combined.append(f{kw1}for{kw2})combined.append(f{kw1}vs{kw2})returncombined二、长尾词评估2.1 竞争度评估defevaluate_long_tail_competition(keyword:str,api_key:str)-Dict:评估长尾词竞争度headers{X-API-Key:api_key,Content-Type:application/json}body{q:keyword,hl:en,gl:us,page:1}rrequests.post(https://api.serpbase.dev/google/search,headersheaders,jsonbody,timeout30)datar.json()organicdata.get(organic,[])# 计算竞争指标indicators{result_count:len(organic),has_ads:False,# SerpBase可能不返回广告has_snippet:len(data.get(people_also_ask,[]))0,has_knowledge:data.get(knowledge_graph)isnotNone,big_sites_in_top5:sum(1foriteminorganic[:5]ifany(dinitem.get(display_link,)fordin[amazon.com,wikipedia.org,youtube.com])),avg_title_length:sum(len(item.get(title,))foriteminorganic[:5])/5iforganicelse0}# 竞争分数score0scoreindicators[big_sites_in_top5]*10score20ifindicators[has_knowledge]else0score15ifindicators[has_snippet]else0scoremax(0,10-indicators[result_count]/10)return{keyword:keyword,competition_score:min(score,100),difficulty:(highifscore60elsemediumifscore30elselow),indicators:indicators}2.2 价值评估defevaluate_keyword_value(keyword:str)-Dict:评估关键词商业价值value_signals{buying_intent:0,informational_intent:0,local_intent:0}keyword_lowerkeyword.lower()# 购买意图buying_words[buy,purchase,price,discount,deal,coupon,best]value_signals[buying_intent]sum(1forwinbuying_wordsifwinkeyword_lower)# 信息意图info_words[how to,what is,guide,tutorial,learn]value_signals[informational_intent]sum(1forwininfo_wordsifwinkeyword_lower)# 本地意图local_words[near me,local,in ,nearby]value_signals[local_intent]sum(1forwinlocal_wordsifwinkeyword_lower)# 商业价值分数commercial_scorevalue_signals[buying_intent]*10value_signals[local_intent]*5return{keyword:keyword,commercial_score:commercial_score,intent_type:(transactionalifvalue_signals[buying_intent]0elselocalifvalue_signals[local_intent]0elseinformational),value_signals:value_signals}三、完整扩展流水线deffull_long_tail_pipeline(seed:str,api_key:str,max_keywords:int1000)-List[Dict]:完整的长尾词扩展流水线print(fStarting expansion from seed:{seed})# 1. 扩展expandedexpand_keywords_from_suggestions(seed,api_key,depth2)expanded.extend(expand_with_modifiers(seed))iflen(expanded)max_keywords:expandedexpanded[:max_keywords]print(fExpanded to{len(expanded)}keywords)# 2. 去重和过滤expandedlist(set(expanded))expanded[kforkinexpandedif3len(k.split())8]# 3-8个词print(fAfter filtering:{len(expanded)}keywords)# 3. 评估results[]forkeywordinexpanded:compevaluate_long_tail_competition(keyword,api_key)valueevaluate_keyword_value(keyword)# 综合评分overall_score(100-comp[competition_score])*0.6value[commercial_score]*0.4results.append({keyword:keyword,competition:comp[difficulty],competition_score:comp[competition_score],commercial_score:value[commercial_score],intent:value[intent_type],overall_score:overall_score,recommendation:(prioritizeifoverall_score60elseconsiderifoverall_score40elsedeprioritize)})# 4. 排序results.sort(keylambdax:x[overall_score],reverseTrue)returnresults四、实战数据从project management software扩展到1000个长尾词类别数量平均竞争度平均商业价值How-to234低中Best/Top189中高Comparison156中高Price/Cost123低高For [industry]198低中Alternative100中中五、总结长尾词扩展的核心多维度扩展搜索建议修饰词组合竞争评估SERP特征判断难度价值评估意图判断商业潜力优先级排序低竞争高价值优先持续扩展每月扩展新词长尾词扩展是SEO中最有技术含量的工作之一。它不需要创意但需要系统性的方法。这套流水线可以从1个种子词扩展到1000个长尾词并且自动评估每个词的价值。SerpBase的成本大约$3/千次扩展1000个词评估成本$3产出是一份优先级明确的关键词清单。